>>12810 Welcome back Kiwi, good to see you found us again.

>>12810 >Stepper motors, brushless motors with ferrous teeth to focus magnetic flux. Neat I had no idea that's how they worked. I figured it was some kind of gears tbh.

>>12810 >Artificial muscles can be made by exploding a hydrogen and oxygen mixture in a piston, then using hydrolysis to turn the water back into hydrogen and oxygen. could we miniaturize this and make artificial muscle cells? granted they'd be larger than actual cells but what if we could mass print them each the size of a grain of rice? Just a thought, as 3d printing increases precision and resolution there's no reason we could not fabricate some pretty interesting structures on the micro-scale

Great, so my comment >>12813 was the last in the old thread: >Cavatappi artificial muscles >Supercoiled polymer (SPC) muscles / TCP >Twisted string actuators (TSA) >Nylon fishing line muscles >Continuous ransmission (CVT) / torque converters >Mechanical multiplexer >3D printed pneumatics >>12810 >5. Coiled Nylon Actuators! These things have an efficiency rating so low it's best to just say they aren't efficient. What they are though is dirt cheap and easy as heck to make! Don't even think about them, I did and it was awful. I guess you mean by heating them from a battery source, but not with hot water which is already there.

>>12815 >but what if we could mass print them each the size of a grain of rice? Just a thought Actually, that's not too far-fetched Anon. It's certainly tracking the trend of miniaturization in technology and machining. It would certainly be a breakthrough, as you suggest. One of the most important men in the history of computing (and certainly to us here) is Carver Mead. During a talk he gave to EE students at Cal Tech where he (still !) teaches, he pointed out the proper place of vacuum tubes in the historical march of chip technology to the MOSFETs of the modern era. During the close of the talk he briefly evaluated future directions. One of those were micro-vacuum tubes, which are actually being produced on micron scales now there at the school. >also: >no cute anime catgrils in OP's pics <let us correct this oversight

(related x-post) >>12736 >adaptive motion smoothing

>>12819 Yes another anon recommended "There's Plenty of Room at the Bottom" by Richard Feynman . I should dig into that this weekend

>>12816 Ths video explores TSAs (Twisted String Actuators) for exoskeletons, which might give indications to how it could be used in a artificial body: https://youtu.be/lf4qPyHNzQk --write-sub TSA-powered glove: https://youtu.be/ZdsFULAslV4 --write-sub

>>12858 Neat! Thanks Anon I saw this pdf streaming by on BUMP's terminal window and stopped to have a look. Cool AF, and kind of addresses indirectly some of the topics I'm working through for pulley placements in windlass-based thigh motions under load.

>>12864 I suppose I could be more explicit exactly what I'm talking about. >

Adding another actuator which I think has great promise from the last thread. Hydraulically amplified self-healing electrostatic actuators with muscle-like performance https://www.science.org/doi/abs/10.1126/science.aao6139 Also called "Dielectric Elastomers" Links to explanations and other links >>10620 >>10639 Rough calculation on cost, forces needed and the number of muscles needed for full human mimicry.(the micro-controller cost might be less) >>12014 Rough equivalence to the force Tesla gets out of their cars to human actual forces >>12140 Rough cost budget of actual existing transistor and micro-controller parts to get full human mimicry >>12172

One I believe of high importance, switched reluctance motors https://www.controleng.com/articles/resurgence-for-sr-motors-drives/ One last one https://en.wikipedia.org/wiki/Piezoelectric_motor The piezo are very fast and powerful. They are used in pick and place machines for electronics parts at high rates of speed. The problem with piezo is high voltages. I think some of the actuator strategies used for piezo at the link could be used with solenoids or other magnetic type devices and it would make for low cost actuators.

>>12929 >Dielectric elastomer actuators are electrically powered muscle mimetics that offer high actuation strain and high efficiency but are limited by failure caused by high electric fields and aging. Acome et al. used a liquid dielectric, rather than an elastomeric polymer, to solve a problem of catastrophic failure in dielectric elastomer actuators. The dielectric's liquid nature allowed it to self-heal—something that would not be possible with a solid dielectric. The approach allowed the authors to exploit electrostatic and hydraulic forces to achieve muscle-like contractions in a powerful but delicate gripper. This is exactly what we need

>>12930 >The problem with piezo is high voltages Not sure, there are these voltage boosters. Might lead to loss of efficiency, though. I have piezo on the radar since I read about Festo using them. I was mainly thinking of valves for air and water, which might last longer than magnetic ones. Also for small actuators. Since this doesn't need much energy int the first place, the loss for boosting up the voltage might not be a problem. >>12932 >dielectric liquid >This is exactly what we need Yes. I agree.

>>12929 >Also called "Dielectric Elastomers" >related crosspost (>>6563)

>>12936 >Not sure, there are these voltage boosters. I had to think about that for a minute then I remembered that there is such a thing as a voltage transformer made of piezos. Is that what you are referring to? I think they would be different from ones made for transducers, motors, etc. Did you know you can make piezos out of PVC. Like the water pipe stuff? You heat it up, apply a high voltage field across it then cool it down while keeping the field on til it cools. For a short while they were making speakers out of it. I screwed around and didn't get any to play with and now I don't think they are around any more. Piezos are really wondrous. A solid state thing you can put voltage on and they move. The problem is they are pain in the ass. They don't move far and it takes high voltage. To get a lot of movement you have to complicate them with levers and...they just suck to do real stuff with but..they are very cool. I think the Japanese putting them in cameras was in reality just a "look at the cool techno-wizardry we can do" move. Succumbing to techno lust is a real thing and needs to be avoided. Some super complex wizard crap that is so difficult to get to work it becomes THE reason you work on something instead of what you were trying to build in the first place. Don't ask how I know about this.

I just now had an epiphany. What we need for a rough prototype is something that has fast movement when it's not under load but some strength when it's under load. Much like a regular human muscle. Well I was thinking about strings and came up with this. I really like this. Cheap and not too complicated. Now this is just a drawing. It's certainly not fleshed out. There would have to be some thinking done to make it work. You have motor with a cone shaped pulley on top of it. The string wraps around the cone. When you turn on the motor the string is pulled goes through a ring(purple) through a pulley(white) around and then up to the end of the limb. The purple ring acts just like the ring on a fishing pole that turns the string on a fishing pole. Now let's say the limb has little resistance to it moving then the motor winds the string up on the large part of the cone because the spring has pulled it down. High speed, low force. If the limb gets resistance the pulley and the ring that guides the string stretches the spring, pulling the sled the pulley and ring are mounted on up. This in turn means the ring that guides the string is now farther up the cone to a smaller section. The motor turning the same speed is now taking up much less string for every revolution. Higher torque but less speed. The pulley and ring are mounted on a sled of sorts that has a guide way so it can only go up and down the (red) slot. You could adjust the force by picking the right cone size, length and the spring force.

You could get some accurate movement with the spring cone auto trans if you had opposing muscles and ran it with pulse width modulation. So you run one muscle one way and the other the other way both with high force but one of them a little bit less. So you are holding the limb tight but moving slowly depending on the opposing difference in force.

>>12941 I used to own a Yamaha CVT (Continuously-Variable Transmission) Scooter. It was only 149cc, but I could regularly get it to 60mph on a flat highway. The neat thing was it would just pull through the powerband then entire way until it reached maximum speed: no shifting or drop-offs. I became interested to figure out how Yamaha did it, and found out it was a heavy-duty rubber conical cylinder that a transfer wheel would back and forth across the surface of transferring that engine power into the driveshaft. Thus the 'continuously-variable' bit. A working geometric expression of integral calculus! I say all that to say that the motor/cone part of your diagram reminds me very directly of the transmission on my scooter. You might look into it Anon.

Since the string would be tied to the cone you wouldn't have to worry about slippage like belts in CVT's.

>>12944 I suppose some forms of the idea use belts, but not my scooter. It was a direct-drive rubber takeup wheel that would ride firmly against the rubber drive cone. The takeup is run along different 'sections' of the cone depending on the drive conditions, which would vary the torque accordingly just like in your idea AFAICT. Anyway, belt-or-not isn't the question, it's "can we use this idea inexpensively to our advantage in creating robowaifus?" I'm betting we can.

>>12940 >>Did you know you can make piezos out of PVC. Like the water pipe stuff? You heat it up, ... Aarrgh!!1! DO NOT HEAT UP PVC. >>5106 >>9043

>>12940 >>voltage transformer made of piezos I mean the regular electronics ones. Boost converters is another name for those, wich can be buyed as a board, or build: https://youtu.be/QnUhjnbZ0T8 --write-sub https://youtu.be/hDZsBeYbDa4 --write-sub (Just in case, if anyone tries to get the inductor the same way, don't damage the lightbulb bc mercury) Btw, we don't need to build everything from the scratch. It's good to be able to, but doing so might also complicate things. If you want too make even the chrystals yourself: https://youtu.be/IGbVGKJldd8 >>12941 >cone gear or cone torque converter Interesting, but I would like to see an experiment based on it. Also, space might be an issue.

>>12947 >Aarrgh!!1! DO NOT HEAT UP PVC. I hate to be so blunt but the fact is you have no idea what you are talking about at all. Plumbers, electricians bend PVC by heating it up all the time.Here's a search for PVC heater bender. https://duckduckgo.com/?q=pvc+heater+bender&t=ffcm&ia=web They make a tool just for this.

>>12948 >Boost converters is another name for those, wich can be buyed as a board, or build I see thirty of them here for $20 https://www.banggood.com/30pcs-DC-2V-24V-To-5V-28V-2A-Step-Up-Boost-Converter-Power-Supply-Module-Adjustable-Regulator-Board-p-1614240.html?cur_warehouse=CN&rmmds=search but...what will you drive with them? If it's a motor then you have to make a bigger motor which cost big time. Also these little cheap things I don't think are going to drive piezos which to really drive them require much higher voltage. I'm not trying to be totally negative. I just don't think this will work out. If you look back at what I've posted a lot of it really won't work or will be so tedious that it will be too much of a pain but I throw ideas out there anyways.

>>12983 Doesn't mean it's healthy, especially if you want to do it on abigger scale. >>12984 The topic wasn't motors but dielectric muscles and small piezoelectric valves or such. My point was, that these boosters exist and some might work. One goes up to 500V from 3V.

>>12987 >The topic wasn't motors but dielectric muscles Ok I didn't see that. It wasn't clear to me.

>>12988 >dielectric muscles and small piezoelectric valves Okay, no problem. Maybe I should have quoted better. I think high voltages could only be used for small moving parts anyways. Bigger motors or muscles would need more ampere, and I don't want to get a shock touching my girlfriend. Or think her of holding a baby...

>>12810 I will model that madokami mecha

I feel like we're overcomplicating things. We should be thinking of this as a simple physics problem: What method gives the best power-to-weight ratio for a simple bicep curl? A 20 inch long blocks split half-way down the middle by a single hinge, pointing straight down with the upper "arm" fixed and a weight attached to the "forearm" and made to curl that weight 180° up using only electricity. The more torque out compared to weight of the arm of that length using an input of X watts DC for Y seconds is the winner. (I'm not sure how many watts or long would be considered reasonable for the experiment) Finding out the most energy efficient way to do this is the key to making knees, elbows, fingers, and probably any other joint we'd need. Because most robots would run on batteries instead of wall outlets, the weight of any inverters would have to be included as part of the arm, because even if it weren't really in the arm, you'd still have the weight in her somewhere. This will likely also make hydraulics and pneumatics impractical even if it were more practical with more limbs. I know this might seem like an oversimplification, but it seems like something that can actually be made into class experiments or a challenge just about anyone could compete in. And I didn't really know how to explain the rule, but no "C-3PO arms", so the blocks have to rest flat against each other, not having a linear motor or cables connecting the bicep to the forearm. If it were a leg that'd be like the back of the thigh connected to a calf, which would get in the way of sitting in a chair. It's not something you could easily hide inside of a human body or clothes and would really get in the way if they were used in prosthetic limbs.

>>13178 Mind explaining that a bit more in-depth Anon? >>13180 >because even if it weren't really in the arm, you'd still have the weight in her somewhere. True. That's one of the challenges a good engineer has to account for, and why the thrown-weight issue is pertinent in any wildly dynamical systems like a robowaifu with all it's limbs and lifelike motions. The Boston Dynamics robot 'dog' Spot is a great example of the design outcome necessary for success at this. All the actuators, batteries, inverters, etc., are kept inboard inside the robot's chassis, and the limbs themselves are kept as simple, rigid, and lightweight as possible. IMO we'll need to attempt something similar to Spot here if we're to succeed at this.

>>13203 The short version: I will literally model and break each and every part of that design down so it can be 3D printed into a model kit as a prototype design. Madokami and all. Now for the long version: I would use existing model kits like megami devices as a reference for part size but start from scratch with the techniques and workflow Ive been practicing since I first showed off a rough draft of Allie. Though that is only the shell of the design. modeling the insides of the thruster equipment would be where the real engineering occurs and would have to use rocket designs for that. schematics for reference. This isnt a sci fi board and i know its a gundam inspired drawing, but I wonder if magnets could be used in the wings to hold up the robo waifu in the air. since magnets that powerful do exist and those wings seem very flimsy with too many bendable parts. I have also 3D printed miniatures using small magnets for the joints before and that works perfectly. Though at the moment the only real problem is that you need another magnet underneath the model or it falls.

>>13205 Excellent. Well I'd say to have at it then Anon! Work hard and I'm sure you'll succeed. >This isnt a sci fi board... We actually had a few similar topics come up when trying to help Fairybot-Anon devise approaches for his literal-flying-mini-robowaifu from years ago. Personally, I consider the fact that drones are so widespread, that IRL flying-fairybot robowaifus will be inevitable eventually.

>>13203 >All the actuators, batteries, inverters, etc., are kept inboard inside the robot's chassis, and the limbs themselves are kept as simple, rigid, and lightweight as possible. I think the automotive term for this is "unsprung mass" >IMO we'll need to attempt something similar to Spot here if we're to succeed at this. I don't really think so. I don't think you can outperform them by aping them. We can't just assume that's the best way to do it just because it's the way the professionals are doing it. My life experience has been that the professionals don't always know best, but people rarely question them because they think there's that significant of a knowledge gap that might not really exist. This YouTube video refers to it as the "Smarter Monkey Fallacy" https://youtu.be/wkiL3Q7cq7A his videos are great even if his presentation is crap. The purpose of simplifying the idea to just bending a lever to lift a weight is so we can get a hundred monkeys at a hundred typewriters trying to solve the problem and anyone could provide a simple solution virtually nobody's thought of yet.

>>13210 >I think the automotive term for this is "unsprung mass" Indeed it is. Thrown-weight is a mechanic's term, which I am.

>>13205 You model her, I'll print her. Honestly, it would be nice if we could work together. I've always wanted to work on a waifu with another Anon. Please make a thread for her, I would be happy to make one for us if you'd like to work together. I'm not good at modelling exteriors but, I am a robotics engineer.

>>13180 >no "C-3PO arms", so the blocks have to rest flat against each other, not having a linear motor or cables connecting the bicep to the forearm I'm not buying that. It's not like normal arms work See here

>>13226 A normal, healthy-weight person can bend their leg to a squat where their heels to touch their ass, and can sit upright in a chair with their knees bent 90° and have the back of the knees touching the seat, instead of having a cable extending somewhere from the calf to the ass getting in between the knee and seat. It's better to have that range of motion available for all joints and not need it, than to need it and not have it.

>>13180 >>I feel like we're overcomplicating things. We should be thinking of this as a simple physics problem: What method gives the best power-to-weight ratio for a simple bicep curl? I can't imagine we can get any cheaper type motor that a switched reluctance motor. It's a couple coils and metal rotors. The only thing I can see as better is "Dielectric Elastomers" >>12929 Regular motors are really expensive. All of them cost more than switched reluctance BUT it's not like you can buy switched reluctance motors all over the place because they need micro=controllers and MOSFETs or transistors to drive them. The cost of these has plummeted so to get the exact precise control we need we will need micro-controllers anyways. What do we need for power to run these things? Notice that a average human can only produce about 200 Watts or so of power over a good bit of time with world class athletes at about 400 watts but you can get much higher brief spurts. Power = Force * Distance/Time Where: Power is measured in watt. Force is measured in Newton. Distance is measured in meter. Time is measured in seconds. 1 lbs force = 4.44822162 newtons We'll use 200lbs. force so someone can be picked up 200lb. = 889.64 newtons Call the arm 0.5 meter Time call it a second Power = Force * Distance/Time So 889.64 newtons*0.5meter/1= 444.82watts Now you can easily get a MOSFET that can be driven with 40volts and that gives us 11.12 Amps @ 40 volts. You can get MOSFET's that will do this for $0.50 USD all day long. You need two if you are running a switched reluctance motor and the micro-computer to drive 18 of these you can get for less than $10 USD. Link on force https://www.thecalculator.co/others/Horsepower-Calculator-490.html Link of force humans need to lift stuff. It's much less than the high 200lb. weight I used. https://www.quora.com/How-much-strength-would-you-need-to-lift-a-human-in-the-air?share=1 There's a problem here. 11 Amps is a lot. It takes a #12 wire to carry 20 amps. It takes a #18 AWG to carry 10 amps. https://i.stack.imgur.com/QqXFg.png You could probably cut that in half because it's going to be intermittent and the current carrying capacity depends on the wire heating up which it won't if this power is in brief spurts. So maybe you could use #22 gauge. Lot cheaper. So here's something cheap and a little bigger . "100-ft 20/2 Twisted Doorbell Wire (By-the-Roll)" https://mobileimages.lowes.com/product/converted/032886/032886856167.jpg Stranded would probably be better but not hard to find.

>>13205 >I will literally model and break each and every part of that design down so it can be 3D printed into a model kit as a prototype design. Madokami and all. NICE!

>>13229 >"Dielectric Elastomers" The problem with these is they are new and it's not like you can go out and buy one you have to make them. Of course the advantage of these is exactly the same. You can make them yourself and don't have to go out and buy them.

>>13231 >"Dielectric Elastomers" I forgot to add no one really knows how long these will last or how well they will do. "IF" they work then long term there really seems to be nothing so satisfactory as these for muscles. They seem perfectly ideal in all ways.

>>13229 Maybe I should repost this in the Meta Thread, because nobody seems to understand what I'm saying. The whole point of the idea was to get people to actually test to prove it and compete to create the most efficient arm, instead of just sitting around discussing and debating methods of actually testing it. Just saying to use a switched reluctance motor and showing a bunch of energy math doesn't really mean anything if you don't even explain where and how the motor is mounted to make the arm move.

>>13236 Where can we get these? I would assume that a smaller scale piston design hidden by some casing might be more efficient but requiring the wiring for sensors like construction vehicles. Diggers specifically. As for the power source, the goal I believe was to make a recharable robowaifu, so either using rechargable drone tier batteries, or some kind of solar power would probably work. Since it probably takes far less energy to grasp an apple than to create lift and maintain flight for prolonged periods of time.

>>12409 In this comment earlier I figured the force directly from pound force, used it to get newtons with an online unit converter. Newtons can be directly substituted for watts. I wasn't sure if this was correct but now using the other formula >>13229 I see the calculations come out the same so it's a good short cut to finding the amps needed at a certain voltage. pound force>convert to newtons>use amount in newtons to directly substitute for number of watts need for force

>>13236 >nobody seems to understand what I'm saying. Your right. You asked what's the best power to weight ratio, what length and how many watts needed. So... There's also a lot of links further up that go to the last thread before this one that go over all this. >The whole point of the idea was to get people to actually test to prove it and compete to create the most efficient arm, instead of just sitting around discussing and debating methods of actually testing it. If you build stuff and have no idea at all what forces are involved, no idea what power(watts) is needed then you are going to do a shit load of work that is not necessary when you could spend 10 minutes getting a ball park figure on what you need. Every time you change any of the weights, forces, etc. you need different wires, strength limbs, etc.. >Just saying to use a switched reluctance motor and showing a bunch of energy math doesn't really mean anything if you don't even explain where and how the motor is mounted to make the arm move. Picture of arm from above. >>13226 Maybe you missed it, The calculations are what you asked for. As for efficiency there's a mass, a ton, a whole lot of literature on what efficiency you can get from different type motors, hydraulics, air pneumatics, etc. It's pointless to just start randomly building stuff without at least using the stuff that people have spent thousands of hours working on the facts of how they work and what the advantages and disadvantages of them are. So no I don't know what you are asking for. The stuff you explicitly asked for a lot of it was given. Even a picture of a normal arms attachments as an idea.

>>13216 Well I also have a 3D printer, anon. If you want to start a thread then go ahead. Getting through the superficial diversity hire barriers in HR has been the only thing stopping me from getting into the field myself.

>>13257 >Your right. You asked what's the best power to weight ratio, what length and how many watts needed. So... I wasn't asking a question, as much as I was proposing standards for an experiment for people to do. I am terrible at writing and explaining things. I should have put the end of the post first: "it seems like something that can actually be made into class experiments or a challenge just about anyone could compete in." That's the point of what I said. Getting people to build and test out a different methods of flexing a single standardized limb, instead of just posting our thoughts here about what's the best way to do it. I guess what I'm trying to propose is a simple community-driven challenge/experiment where we actually try to outperform each other instead of just a bunch of ideas on paper. I wanted to make as many constants as possible to keep the math simple, both so more people will be able to compete and so you don't end up with limbs that are only more efficient because of the scale, and such. I said 20 inches because that was the first thing to come to mind and I couldn't think of any constants for time or wattage, so I just used X & Y. I'd love to hear some better suggestions.

>>13277 > I was proposing standards for an experiment for people to do. I am terrible at writing and explaining things. I should have put the end of the post first: "it seems like something that can actually be made into class experiments or a challenge just about anyone could compete in." That's the point of what I said. Getting people to build and test out a different methods of flexing a single standardized limb, instead of just posting our thoughts here about what's the best way to do it. Ok, you lay out some standards, show a design to test and then build it, "...instead of just posting our thoughts here about what's the best way to do it...". You first. Show us.

How to double the power of magnets with a Hallback Array: https://youtu.be/uQWHjj6ofwo

>>13236 You're right. Building something is better than dreaming up technology. However, there's not gonna be one answer and best method. Whatever, I'm going to work on some arm soon.

While working on OSRM, I've come to realize that DC cored motors inefficiency is a problem but brushless are too expensive. Coreless motors are where it's at. They're efficient like brushless but cheaper and don't need special controllers that increase costs.

check this. https://www.youtube.com/watch?v=hhDdfiRCQS4

>>13285 >You first. Show us. I will. Sometime between that post and now, I've found I'm going to get fired for not taking the vax. Also I've been working nights, which makes it hard to get anything done without noise complaints. There's also the fact that I know as soon as I post it I'm going to be met with replies about ways to change the experiment. I've already had one comment about how a 180° bend isn't necessary, but expect asking for a standard material and how thick or wide the arm should be, then eventually devolve into why I used an imperial measurement when we should use metric, etc. I expected a major backlash against the idea by people who prefer to just sit around thinking and talking, will find any way they can to ruin this. >>13313 >However, there's not gonna be one answer and best method It's just a general-purpose limb, mostly for bending at the elbow and knee. Scaling down to the fingers probably won't work and it won't be good for flexing face muscles or anything like that, but the application largely depends on what the leading design is. There's not gonna be one answer and best method regardless if the experiment is done or not, so why not do it anyway? It's application isn't just limited to waifubots either.

>>13318 >>I've found I'm going to get fired for not taking the vax That's really sucks. F the vax, F the people trying to force it on people and F the people who engineered corona. What ever you do don't quit. They will try to make you quit. Don't do it. That way they will have no liability. They will really try hard to do this. Watch you'll see. It may very well be that they will some day have to pay you for firing you. Force them to fire you. Of topic but we now have many serious test that show Ivermectin knocks out corona. We even have a whole country of billions, India, that has shown that corona can be stopped fast and eliminated. If they want to stop corona it's as easy as giving everyone Ivermectin, vit. D, and vit. C and corona is done in a few months for a couple dollars per person.

>>13322 >Force them to fire you. That was the plan, if only so it makes it easier to get unemployment if I can't get a new job soon. But please, let's not turn this thread about actuators into a Covid thread. When I leave the job, I'll switch back to sleeping normal hours, and it'll be easier for me to assemble the little CNC machine I bought but never got around to building. I don't trust my hand-eye coordination enough to do cut things by hand.

>>13324 Good luck Anon.

I was about to give up on creating and humanoid robowaifu because I can't afford it but I think this mechanism could be my savior: https://makezine.com/projects/the-chinese-windlass/ I think using this we could create high torque actuators using cheap small motors.

>>13489 I actually mentioned those in the OSRM thread, they are also known as the differential windlass and have incredible efficiency with potential for incredible strength. Using two connected to the motor can provide push and pull forces. https://makezine.com/2014/05/28/rise-robotics-cyclone-muscle/

>>12810 https://youtu.be/ENMZsPwCUcA Capstan Drive: nema stepper motor with no gears?

Anyone got a guide to which joints are safe to put your fingers and other fleshy bits near? Obviously it's not so important on low torque motors, but for life size waifubots you're getting into dangerous territory. I wouldn't want to use anything on a sexbot that might mangle my wangle. Imagine having to explain that one in the hospital.

Nylon actuators seem like a real bad idea also because you'll stress the thread way too much. I don't see why you would choose this method over a simple spool (which has been proven to be reliable and efficient since they use it on cranes) or pair of spools. >who wants their waifu to bench press a car? Patricians.

>>13671 Nylon is dumb as heck for an actuator in a batery and thermal constrained robot. >>13506 ...Why put your rooster near any mechanical joints? Just use an onahole integrated into her pelvis.

>>13671 >Nylon ... stress the thread way too much Which thread? I think I recall these actuators to be very reliable. The developers tested them a lot and made this claim, if I recall it correctly. >over a simple spool Because the spool implies a motor, and it might be slower and certainly louder. You could have both. The Nylon is only used if the weight is to high for the other ones. >>13674 >Nylon is dumb as heck for an actuator in a batery and thermal constrained robot We had this several times here on the site. You're thinking in terms of the option of heating the Nylon with electricity. But hot water can be used, which might already be there anyways. It depends on the design of the robot. Like it or not, some might use that approach. Even the method with electric heating might be interesting for some smaller movements, e.g. in the face. Though, dielectric elastomers might be better suited in that case. Nylon actuators are also quite slow, so it's for some special use cases, especially support for other actuators in case of heavy lifting, but not generally useless. Don't judge everything based on your approach.

>>13678 Okay I totally misunderstood the way these things work. From the OP picture it looks like they are motor-based like a simple spool, but the motor is perpendicular and contraction is provided by overtwisting. I am now aware that that is not the case and I'll concede that they actually are cool and potentially useful.

>>13687 What you thought of is a twisted string actuator. I just got my drone motors and want to try those out for that.

>>13678 >heating nylon as a wire and totally not causing a house fire. >Hot water? From where? The cooling system? How would it help movement?

>>13688 Good, that's a genuinely good actuator mechanism and I hope you share your findings. Just know that drone motors require cooling if they are used for more then a few seconds at a time.

>>13689 Please read through the existing information before making some bold posting. There are actuators with thin wires and Nylon. Also, yes, hot water comes from the cooling but could also be heated on purpose if necessary while being plugged in. Actuators help movements by actuating or so, I think.

Has anyone replicated some bldc-actuated limb projects on hackaday? https://hackaday.io/project/181799-redacted-the-first-fully-open-bipedal-robo There are several great open-source projects which are a good place to start to actually build a robot instead of theorizing about what would be the perfect actuator and never getting anywhere. I was on a tear with nitinol braided actuators being the perfect actuator, but the part manufacturing held me back to the point where all I could work with wasn't nearly good enough. We need to start with something at all in order to have a hands-on perspective of what works and why to build these robots at all. Even a perfect first design nearly always has something go wrong with it, which is why iterating versions is a superior method to getting stuck in theoryland and never making anything.

found this just now https://www.youtube.com/watch?v=guDIwspRGJ8 >We have achieved strong, fast, power-dense, high-efficiency, biomimetic, soft, safe, clean, organic and affordable robotic technology. Dumbbell weights 7 kg (15,6 lbs) , forearm with hand only 1 kg (2,2 lbs). >This artificial muscles robotic arm is operated by water and consumes 200W at peak. We invent and produce our electro-hydraulic mini valves to have complete controllability of speed contraction and compress the whole powering system (for a full body) inside humanlike robot torso. >At this moment our robotic arm is operated only by a half of artificial muscles when compared to a human body. Strongest finger-bending muscle still missing. Fingers are going to move from left to right but they don't have muscles yet. Metacarpal and left-to-right wrist movement are also blocked. This version has a position sensor in each joint but they are yet to be software-implemented. We are going to add everything mentioned above in the next prototype. >The movement sequence was written and sent by simple commands to a hand.

>>13723 To my knowledge, no one has replicated a Hackaday project here. That project is ok as a concept but, is severely flawed. All of the mass is centralized high up in the hips which means it would lack any stability IRL. It also suffers from having brushless motors close enough for the magnetic fields to adversely affect each other. The flaws in the design continue but, this is not the thread to discuss why that design would not make sense IRL. As for this thread, it was made for the express purpose of discussing actually useful actuators. Anon's enjoy dreaming about super coiled actuators and other ultra low cost actuators that do not actually make sense to use for various reasons. My guess is they want their waifus to be cheap to produce. I will continue to tell everyone to use DC and brushless motors as they are the only sensible option currently available unless she's constantly plugged in. Steppers could also work and are very cheap for their strength, they're just heavy.

>>13724 That thread mentions him in the "Humanoid Robot Project Videos" thread: Automaton dev from Poland >>5136 >>10179 >>13304, but I don't post every video of his. Crosslink in the actuator thread also exists.

>>13724 If only we had power systems capable of delivering hundreds of watts over prolonged periods. If only water was not heavy. These actuators are such heart breakers. They seem almost fantastical until you realize how inefficient electrolysis is for a pneumatic system. There's also the major problem of making sure the hydrogen and oxygen do not escape the system and reconstitute into water at the right time. This Eastern European man explains how the system actually works. https://www.youtube.com/watch?v=gy-Cl8X6Itk

>>13725 >thread, it was made for the express purpose of discussing actually useful actua It's the official thread for actuators, so we'll discuss all of them here. Being strongly opinionated doesn't mean you're right. >continue to tell everyone to use DC and brushless motors as they are the only sensible option For her primary joint movements it might be the best approach, but can still be combined with other accuators in case of lifting something for example. >unless she's constantly plugged in She can be plugged in most of the time, while not moving from one place to another. While sitting she can be plugged in. Then doing some movements while laid down or standing somewhere won't need much energy.

>>13723 >the-first-fully-open-bipedal-robot I answered here, in the meta thread: >>13737

>>13735 Hm, it would my personal biases have poisoned my judgment. I forgot there were those among us that would want a plug in waifu. I only want a waifu that can follow me so I only think about robowaifus through the lens of one that must be self powered.

>>13724 That's very impressive. Having hydrogen and oxygen in the same tube is...risky if I understand it correctly. Maybe he has a way to separate them. Not sure. He is making a gas to drive pneumatic actuators which is a good idea. It happened to occur to me possibly a spark gap which can expand water, or other liquid, would work as well??? Not sure it may be even more inefficient than splitting water. Maybe you could use some other sort of element that spit into a gas although I can't thing of one off hand. I'm complaining about the shortcomings but don't think I'm not impressed mightily by what he has done. It's very excellent.

>>13755 >>13729 >the major problem of making sure the hydrogen and oxygen do not escape the system ... >Having hydrogen and oxygen in the same tube .. H and o2 aren't going to be at standard pressure, that's something to factor in. No need to worry so much about escape, since that's still an issue in pneumatic and hydraulic systems regardless (and more dangerous in the latter). What's really interesting here is the way the actuators are arranged like muscles - with apparently opposing muscle pairs for greater control rather than jerky one-way movements or rigid machinelike movements. However don't rule servos out entirely they, and solenoids as well can do a lot still. Consider the human body, it has smooth muscle, skeletal muscle (slow and fast twitch) and cardiac muscle. Similarly, we should expect to use a diversity of tactics to the actuator problem. 1 .Hydraulics or large pneumatics where the most power is needed (shoulders, bicep, legs), 2. Where some power but more control is needed: wrists, neck, feet/ankles: smaller pneumatics (they don't have to be micro, but an array of micro pneumatics shouldn't be ruled out), the hydraulic transformer that other anon had mentioned, and/or pulleys which sound low-tech but don't have to be: try to envision something like a metallic herringbone chain inside a lubricated graphene or nylon weave bladder. 3. Finally for fine dexterity: nitrile wire or something similar from the "soft robotics" school: fingertips, facial muscles (when we cross that bridge), etc. Anyway I wanted to respond to you both before sharing this too: https://www.youtube.com/watch?v=6JNq1COqB_s[Embed] https://spectrum.ieee.org/kenshiro-robot-gets-new-muscles-and-bones

>>13506 >...to put your fingers and other fleshy bits near? Redundant sensors are your friend. Also, having the system only use so much force as necessary for each movement. Also, there are the terms 'compliant' and 'backdriveable' waifusearch> backdriveable OR backdrivable THREAD SUBJECT POST LINK The Basement Lounge >>4807 backdriveable Robotics Hardware General >>12511 " Actuators for waifu movement! >>7086 backdrivable " >>8207 " " >>9198 backdriveable " >>11700 " " >>11855 backdrivable " >>11881 backdriveable Work on my Elfdroid Sophie >>7675 " " >>7693 backdrivable " >>7695 " " >>7701 " Actuators For Waifu Movement Par >>13761 backdriveable, backdrivable ' backdriveable | backdrivable ' = 13 results waifusearch> compliant OR compliance THREAD SUBJECT POST LINK C++ General >>1073 compliant Robotics Hardware General >>12511 " R&D General >>2048 " Wifu that gives you Hope. >>6823 compliance " >>6827 " General Robotics/A.I. news and c >>6229 " " >>6231 " " >>6381 compliant " >>6919 " " >>12586 " Actuators for waifu movement! >>8207 " " >>8221 compliance " >>11023 " " >>11028 " " >>11030 compliant, compliance Prototypes and failures >>12882 compliance Why consider alternative CPU arc >>4506 compliant, compliance Robowaifu Power and Control Syst >>11175 compliant Actuators For Waifu Movement Par >>13761 compliant, compliance Madoka.mi prototype thread & rob >>13511 compliant ' compliant | compliance ' = 20 results Cycloidal drives with a gear ratio of maximum 10:1 are backdriveable, along James Bruton. A soft outer shell might also help with your concerns. Doing such a search with waifusearch requires the option "-y false" when starting the program from the shell. >=== -I hope you don't mind Anon, but I cleaned up your waifusearches a bit. I suggest you post them on the board inside codeblocks, as I've done here for you. This will help ensure the browser rendering of contiguous spaces is monospaced. -Also, I'd suggest you upgrade your Waifusearch copy to the latest v0.2a (>>8678). As you can see here, among other improvements, basic Boolean OR is available for different (though related) search terms. -Also, there's a newer version of the all_jsons just posted; 211117 - https://files.catbox.moe/jgmdxs.7z Cheers. >t.Chobitsu -edit waifusearch results

>>13754 Yeah, I knew the links weren't working. But for some reason the option to remove the hyperlinks didn't work when I tried waifusearch the last time. It seems not to be a bug, but I don't know why it didn't work the last time. I reposted the article, and I will delete this posting here later. So you might delete yours, if you wish. The new posting is still messy, probably because he didn't use tabs, my copypasta didn't copy them or I don't know.

>>13724 >>13728 Reliability test of hydraulic actutors (literally working with water) by Automaton robotcs: https://youtu.be/iQhYXE6cAEY This does it for me, I might actually use something like that. Because it's easiely imaginable to reuse the water by catching it with onter tube around the muscle.

Hasel actuators are something to keep an eye on. They might not be interesting for the face, because of high voltage, but maybe somewhere. >Reminder: Hasel actuators https://youtu.be/PGGQc5q2NGo >Use cases, advantages https://youtu.be/TjglKIkLFSI >Speedtest: https://youtu.be/Lsn-Z1wKaEo >General progress, more stroke https://youtu.be/ep0CV9PYtSo >Accumulation conveyor, efficiency https://youtu.be/bjDdUn39H7s >Fast repetitive movements https://youtu.be/2ggwTlWbkZI >Dev kits https://youtu.be/i1QmwOxGGFA https://youtu.be/LhQgKnkxXXA https://youtu.be/usvoiGBAflY

>>13966 This is similar to a concept I was thinking of, except my design only contracts like a muscle. Any idea what that "self-healing liquid dielectric" is?

>>13978 All liquids are self healing. Dielectric just means they're insulators. Many HASEL actuators have been made with castor oil and transformer oils because they can handle large voltages. The system works through hydraulics and large voltage differentials causing conductive plates to attract each other. Really, the oil is only a working fluid that needs to easily flow when the plates squeeze it and, withstand large voltages so that shorts don't happen.

>>13989 >All liquids are self healing. Well, that's obvious, but I'm not the one who called it that. It just seemed like he was trying too hard to make it sound more sophisticated than it actually is. >Many HASEL actuators have been made with castor oil and transformer oils because they can handle large voltages. That's good to hear. The thing I was going to make was very similar, but uses electrostatic repulsion instead of attraction, so I wasn't sure what to use aside from the main electrode. I was thinking of a dielectric powder instead of oil, but I guess either or a mix of both could do. Here's a crude doodle, it's a loop of wire inside a plastic bag with the dielectric. A single HV AC power source charges the loop, so both sides are the same charge and repel. Alternatively I was thinking of just a single wire, half of the loop, with dielectric doing all the repulsion like putting sand or something on the top load of a Tesla coil and turning it on to blast them off, but 'catching' them with the bag to push perpendicular to the wire to create the force. But maybe it would be better to use a conductor instead of a dielectric, for an added induction-levitation-like force? Or go in the opposite direction entirely and rely on wires, for something that'd look more like the other pic.

>>13989 >HASEL Actuators This was educational: https://www.youtube.com/watch?v=Lsn-Z1wKaEo They are also for sale apparently. What does "self healing" even mean? It sounds like a marketing gimmick.

>>14021 >What does "self healing" even mean? It sounds like a marketing gimmick. Yeah, I made that joke already. It just means that it's a capacitor with a liquid dielectric, so when the capacitor plates attract and the liquid is squeezed out, it'll "self-heal" by the liquid moving back in once the plates move apart.

>>14018 >It just seemed like he was trying too hard to make it sound more sophisticated than it actually is. That's a polite way of putting it. >Dielectric repulsion That's fascinating. I take it it'll maintain constant repulsion until a drain is triggered. I would always use a dielectric oil for any electric charge based actuator. As you are no doubt aware, they have many advantages over most other dielectric mediums. Incorporating a dielectric powder could be interesting for experiments. Air is an excellent dielectric but, it makes for very weak and inefficient dielectric actuator from my albeit, very limited experience.

>>14024 I hadn't heard of the Hasel method until this thread, my original idea was inspired years ago by the beginning and end of this video: https://www.youtube.com/watch?v=P5za9sa4-qk but I genuinely don't know enough about electricity to know what would be the best material to use. I've seen people put flakes of aluminum foil on the top load of a tesla coil and they fly off. I've also seen it done with bits of paper, and that video uses sand. I was thinking of eventually doing the same thing with a little tesla coil of my own and just figured I'd try a bunch of things to see what gets the most distance. A lot of my ideas rely on brute force trial and error, because I have no idea what I'm doing. It wasn't even until later that I realized I could probably just use a bunch of insulated wires in phase with each other, (instead of 180° out of phase like Hasel) hence that wire ball pic. >I take it it'll maintain constant repulsion until a drain is triggered. I was just going to use HV AC in an ISM band, and figured the strength of the contraction could just be controlled with amplitude modulation, or maybe pulse-density modulation like a real nervous system. I didn't think about maintaining the contraction by maintaining a DC charge, because you'd probably need to keep putting energy into it just to do any meaningful work.

>>14022 I'm quite sure they mean if there's a hole somewhere the oil coming out dries and closes it, so it might be flexible when dried, maybe has something like silicone rubber mixed into it.

>>14039 That would make sense, but I'm not getting that from the diagram.

>>14039 That was never brought up in several HASEL research papers I've read. It's a good idea if you could figure out the appropriate additive to seal holes when reacting with oxygen or nitrogen.

>>14061 Thanks, then it was just my imagination. Would have been nice, though.

>>12810 >hasel actuators Would a continuous coil of the circlar disk actuators work? Also, I think the preferred direction of actuators are those that work in tension (contraction) or both ways (bi-directional) instead of expanion. This way, the actuator or string does not collapse on itself in buckling, easier to connect the pulling string, ect. A real muscle works in tension and tendons / joints / bones all work like wires in bowden tubes. I am going to start printing CF Nylon cycloidal gears with PC Max enclosures to make servos in the style of the anti-backlash cycloidal gearbox on hackaday. The carbon fibers, nylon, polycarbonate mix, and maybe some PTFE powder should give me plenty of dry lubrication that doesn't need frequent servicing. I would like to try planetary gearboxes, but I feel that any amount of warping from 3d printing very finely detailed gear teeth would lead to noise and backlash. Cycloidal gearboxes seem to be better optimized for 3d printing BLDC gearboxes at home. PolyMax PC/PC Max, PolyMax CoPA nylon, and NylonX are all higher temperature plastics with anti-warping compositions, and strong as fuck for being printable on entry-level 3d printer machines (all metal hot-end extruders, heated bed, and non-heated enclosures). Machining delrin gears would probably be better strength (~60MPa printed part tensile strength vs 83 MPa delrin tensile strength) and friction-wise. I will have to see how bad cycloidal gear vibration is for low-speed applications like a robutt muscle. If cycloidal gearboxes turn out to be heavy pieces of garbage, I will report back here with results and forge on ahead with planetary print-in-place gearboxes like I planned to originally, before I found an existing model for cycloidal gearbox actuators on hackaday.

>>14513 >Coil of disk hasel actuators Should indeed work, would be complicated to build though. >Contraction is better then extension You are correct. It must be said that unless gravity or a spring is used to provide a return force, you will need to pull both ways. >Cycloidal gearbox Mind sharing the link to the Hackaday project? Cycloidal gears are notorious for vibration, make sure you carefully balance the gears. Most modern designs use out of phase gears to balance out the forces, that can help a lot. Did you design a print-in-place planetary gearbox or, is this design online? Either way, please share links and information. Good luck, cycloidal gears have low backlash and high ratio to area potential, would be good to have as an option for Anons.

>>14517 >Hackaday cycloidal gearbox project https://hackaday.io/project/167855-simple-cycloidal-robot-leg-for-quadruped Also, >stepped gear planetary gearbox >file related Making a stepped gear in a single stage planetary actuator would give a working gear ratio with high backdrivability and low vibration compared to a cycloidal gearbox.

I got ahold of a BLDC dev kit (moteus) and tried it out. It is indeed the motor most suitable for human x robot interaction since it is light, backdriveable, powerful, available, long-lasting, and has a large platform of hobbyists doing development work right now. These beat shape memory actuators in precision and ease of use by miles. There's no working fluid, no compressors, no pumps or tubes, and they are ready to have many types of controllers bolted on right from the factory. This is the fastest way to build a good strong motor that isn't a non-backdriveable geared DC motor. The main downsides I'm noticing are that they aren't very cheap at 30-70$ for just a small low torque motor, 60-80$ in controllers per actuator, up to 500$+ for a geared high torque actuator. Also, there's effects at play that make the motor/actuator system surely undesirable for some people: >cost >bulky circular shape >cogging (motor has intermittent torque when the magnets align with the poles) which feels like it has "cogs" as you turn the motor with your hand (not perfectly smooth rotation like you might want) >noise (quiet static and grinding noise from electronics) >rigidity by virtue of being solid >backlash if using a gearbox with small motor >high inertia if using a big direct-drive motor If you don't care about the trade-offs, I'd recommend a BLDC + Controller system. I have a gearbox in the works to bolt onto the motor to see if 3d printed planetary gearboxes will work. I'd also like to see if there's a coreless/ironless/slotless motor that would be able to direct-drive a large surface (thighs) smoothly.

>>14689 One last thing I forgot to mention is the potential that low KV gimbal brushless motors have. They are used for camera panning, and have significantly higher coil turns/resistance, so they cannot have position accurately measured in the same way as some BLDC controllers are set up. The potential benefit is that they are cheaper and run on lower current that can be pumped out by cheaper controllers and power supplies.

>>14689 Thanks for the nice breakdown Anon, it's appreciated. I'd say cost is the most obvious (and limiting) factor for the significant majority of us. I hope we can at the least find alternatives for pricey single-use controllers, and spread the costs out among fewer, general-purpose drivers. >coreless/ironless/slotless motor What does 'slotless' motor mean?

>>14691 Please use "ctrl + f" to find information in the relevant thread. Slotless means coreless and that's been explained before in this thread. Picrel for more info

>>14694 Ahh, I see. My apologies for appearing lazy. I would guess the little micro motors in tiny drones are of this type. I know they get pretty hot, but the RPMs achievable are frankly, amazing. Again, thanks.

Anyone have any updates to building a robot actuator?

>>15126 I'm not really aware of anything new on that front Anon. I expect we'll have something before long though.

>>15174 I'm making a nylon gearbox for the best feasable actuator mentioned in >>14689. I have about one or two more weekends until it is prototyped. This same actuator could even be applied in any other system for cheap power and precision.

In >>1002, there's some links to some decent looking electromagnetic actuators that are fairly simple to build. They're limited in that they only switch between two states, and do so quickly, without a smooth transition. I'm hoping to adapt them for arm and neck mechanisms in a mini robowaifu, and the current plan is to use a small spring and variable current to try and produce a finer position control/state transition. That said, this seems like an over-complicated method, so would anyone here have some recommendations for alternative avenues of consideration?

>>15735 I think we'll probably find plenty of uses for fast-acting, linear actuators Anon. I hope to see work here along that line, thanks! >>15737 >DC motors are the easiest. Just get a gear motor that exceeds your torque requirements and add a potentiometer to control position or, just use a readily available servo. Good thinking, Kiwi. Can you sort of diagram that for us? I think I understand most of the general points there, but I probably lack understanding in some of the details.

>>15313 I've made a gearset prototype from 3d printed nylon, and it's garbage. However, I will be tweaking it and trying to make a wolfrom stage planetary gearbox the same way. Here is a paper which has some details (key points: human-safe, high backdrivabilty, high gear ratio, small size) https://ieeexplore.ieee.org/document/8867893

>>15878 Any chance you can post pics of your WIP Anon?

>>15879 >>15890

>>15891 Thanks!

My idea for an artificial muscle is to have a good stretchy material, which has a certain amount of liquid in it and no liquid will get added or removed, rather the stretchy container of the liquid will be squeezed in proper places to deform the container, thus changing the shape and causing it to contract so as to work like a muscle. Right now its just theoretical, I haven't tested anything. >>14018 I also kinda like this idea Maybe there could be special structures kinda like meshes that would contract when turned separate directions on each end and could do it really quick, I think I may have seen something like that before.. its kinda like how the lower arm bones go when the lower arm is turned

>>16629 I'd recommend HASEL actuators. This post has good information: >>13966

>>16631 The only issue I see with the HASEL actuators is that they require direct current (DC) high voltages (3,000 - 10,000 V). I'm thinking maybe one could do this a different way using low power maybe with a small motor which controls clamps or something. but I guess it might get kinda unnecessarily complex.. 2 sacks with fluid could be connected and while one be squeezed the other be full or similar, might work well for an arm, just throwing out ideas.. eventually I'll do some tests to see what works well and what doesn't

>>16632 At 15cm scale, you'd be best off implementing a rotary prioritizing selective mechanical multiplexer. With a power motor and an index motor that decides the output. This allows for a very small mechanism to provide high power to many outputs. (Relative to individual actuators for each joint.) Picrel is a decent example made with lego which should help make it easy to understand. Here's a link with valuable information: https://hackaday.com/2021/05/04/prioritising-mechanical-multiplexer/

>>16636 https://www.youtube.com/watch?v=tCQ2I_rga0k Apparently the squeezing actuators can work.. for certain applications. But you may be right that I should explore alternatives..

>>15126 I downloaded this code for making a cycloidal drive in OpenSCAD, and looking into it. https://www.youtube.com/watch?v=FdTGeYKB5NM - Code is in the video description (--write-description). I hope it works this time. I already had some small drone motors for a while, but also five Nema 17 coming in soon. The reason why I didn't do this before was that I had some tutorial for Fusion 360, but I wasn't capable of transferring that into Solvespace. I also had a builder for Blender, but this was before I got my new PC.

>>16692 There's a new video on eccentrically cycloidal drives, and generally a good video on cycloidal drives, by RoTechnic: https://youtu.be/qMDU5tlGUwU - I thought I posted about the eccentrically ones before, but can't find it right now. Whatever, I'm trying to finally figure out at least one way to make a more regular cycloidal drive that I can adjust to my needs and actually print and build. The online examples are all a bit different from each other and many require some very specific gearing, and the plastic parts are mostly only available as stl files, so I can't just adjust the files for another size or another gearing or another reduction. The eccentrically cycloidal version is highly interesting, but it's still under patent protection and making my own version in OpenSCAD would probably require math skills which I simply don't have (or didn't have, I think one video explains it very good). I'm gonna try it anyways. Maybe I can do it with his Python code: https://github.com/roTechnic/internalECGears Another video from the same guy: https://youtu.be/xv5cd7Bg7Uk and the one where he explains the design fundamentals of how to design a regular cycloidal gear: https://youtu.be/y9vLVXjz2c0 I'm also thinking about a bevel drive for the shoulder now, inspired by Dan Royer and his Sixi Robot: https://youtu.be/W6xzBVVuml8

>>16709 I think I found a finished model, which I can use to wrap my head around. Actually it's several versions in one file. Understanding the formula or not, I think I can print and use it, and probably also modify to my use case. https://3dmdb.com/en/3d-model/cycloidal-speed-reducer-in-openscad-by-mattmoses/9065832/ Same on Thingiverse: https://www.thingiverse.com/thing:8348/files > By request of Syvwlch ( http://www.thingiverse.com/syvwlch ) and WilliamAAdams ( http://www.thingiverse.com/WilliamAAdams ), here is a stand-alone public-domain OpenSCAD cycloidal speed reducer. As with the Wankel Engine and Roots Blower I recently posted, this is intended more as an example of an interesting mechanism than as a practical device. If you want a practical printable speed reducer, you might consider one of the other alternatives like > the worm drives on this Tank http://www.thingiverse.com/thing:8080 or > differential planetary gears http://www.thingiverse.com/thing:7390 or > cascaded spur gears http://www.thingiverse.com/thing:7379 > or this planetary gear reducer http://www.thingiverse.com/thing:8460 > There are several cycloidal-type mechanisms already on Thingiverse, such as http://www.thingiverse.com/thing:3617 and http://www.thingiverse.com/thing:3736 > There are also several interesting external sites like: http://www.zincland.com/hypocycloid/ > http://fabricationsofthemind.com/2010/07/09/extruder-design-1-printable-1001-hypocycloidal-gearbox/ > https://github.com/triffid/Differential_Hypocycloid > http://reprap.org/wiki/Differential_Hypocycloid > http://en.wikipedia.org/wiki/Gerotor > http://en.wikipedia.org/wiki/Gear_pump > and many many interesting youtube videos such as http://www.youtube.com/watch?v=bRn1K2XeWVE[Embed] > http://www.youtube.com/watch?v=3WvPF6uGCq4[Embed] > http://www.youtube.com/watch?v=CG2sPuqEXBg[Embed] > http://www.youtube.com/watch?v=AMtyFwMDL7w[Embed] > http://www.youtube.com/watch?v=h236SP86nnQ[Embed] > This present script is based on a design by M.F. Hill described in his 1928 patent "Internal Rotor", number 1,682,563: > http://www.google.com/patents/about?id=mdF5AAAAEBAJ&dq=1682563 > Note that this design is based on an offset hypocycloid, similar to Figure I in Hill's patent. Most of the contemporary designs appear to be based on an offset epicycloid, more closely resembling Figure V in the patent. > The motivated student can modify the code so it generates epicycloidal-based profiles. Hint: start by making a module ``epitrochoidBandFast(n, r, thickness, r_off)". The motivated student could also probably clean up my train-wreck of code and/or figure out how to do arrays in OpenSCAD. > Note also that these rotors can be used for pumps - see the gifs in the comments for an example.

Great material Anons, thanks.

>>16629 I like the idea. Maybe we can use it somewhere. It's simpler than pushing in more liquid and then using a valve to keep it, till it can come out again. But it might also make more noise with the motors for the squezzers, and I'm not sure how strong it would be. Whatever, I think it's good to throw around some ideas. >>16637 Okay, cool. (Didn't watch it yet)

I'm surprised you're not building a smaller 140cm version of the actuated ball joint dolls yet. We have standalone models for ai conversation now you can download from yandex. Anyway, here's how to have a smooth servo. https://youtu.be/jsXolwJskKM I should mention that vrchat supports osc out now. If you could convince someone over the web to be your waifu with some speech synthesis, you could use the movement of them in vrchat full body to remotely control your robo bjd in your living room. Have fun.

>>16890 Thanks for your contribution. But which "actuated ball joint dolls" do you mean?

As with the other Anon, thanks for the contribution OP. But this support having it's own thread so I'll be merging it to one of the others. Not sure if its Actuators or TITS yet, since you touch on both. I'll think about it a bit. Cheers.

>>16890 >If you could convince someone over the web to be your waifu with some speech synthesis, you could use the movement of them in vrchat full body to remotely control your robo bjd in your living room. No, it's a way too risky and awkward. We would still need additional layers controlled by AI: >>9709

I'm currently looking into some bldc motor calculations, or guestimations. I found this calculator: https://www.societyofrobots.com/robot_arm_calculator.shtml and used it to get some idea about what might be possible. It's still hard and complex, since I don't know how to calculate the maximum fringe case of lifting a water container of 8kg near to the body with two hands. Then again, I don't really need it, I could go with a small water container. Size and batteries will be more of a limiting factor, I guess. Interesting in that regard (the tutorial): https://www.societyofrobots.com/robot_arm_tutorial.shtml I won't buy anything big soon, but wanted to start learning more about it. Gimbal motors seem to be the most interesting, though high torque bldc motors have more RPM and could be reduced to have more torque, but are heavy: GM8112 motor by iPower weighs 400 grams,, has 6 kilogram-force centimeter (kgf·cm) = 0.5886 Newton meter (N·m) A high torque one from Ato has 0.8 Nm, but weights 3kg: https://www.ato.com/250w-bldc-motor - but it has 10x more RPM, so would still be faster after reduction if I'm not misguided. Hmm.

>"Dielectric Elastomers" I finally watched this video here on inflatable muscles: https://youtu.be/iiStyGqAG9Y - Here some search terms for further research: Fergal Coulter, hexachiral structures, Minimum Energy Structure - these might be useful for creating other kind of inflatable or deflateable muscles or structures like soft shells: https://youtu.be/IJ1dgxSZuAY The process for creating such variant of dielectric elastomer based muscles is quite complex: https://youtu.be/Qmuf_6h7Kl8 >Multiple layers of hard silicone (Shore A 73) are printed with a seamless hexachiral structure on an inflated silicone balloon. >When the layers are cured, the balloon is deflated, leaving a Minimum Energy Structure (MES) tube. http://fergalcoulter.eu/?p=73 Searching for hexachiral structures lead me to this: >The objective of this study is to improve these two mechanical properties, without significantly compromising the effective yield stress, in the regime with significant material and geometrical nonlinearity effects. https://www.researchgate.net/publication/323503727_Auxetic_hexachiral_structures_with_wavy_ligaments_for_large_elasto-plastic_deformation (PDF related)

>>16900 This is a simple static calculator, which boils down to computing the torque needed to statically lift the described geometry (a center of mass placed at a distance from the root motor). On one hand, the dynamical motor load will be larger than a static estimate, on the other hand motors often have higher peak torque available. I think there is no realistic robot actuator technology on the market or in the engineering lab rn except for electric (BLDC), pneumatic and hydraulic actuators. Pneumatic actuators could be usable in theory for actuating fingers and face (and they are indeed used in some high-end japanese humanoid robots) if you can find a way to print/mold/produce the required valves and pneumatic lines and the pump cheap enough for that to make sense. Precision hydraulics is mostly impossible for a hobbyist. What remains is designing (because you likely won't be able to buy one) a BLDC+reducer servo for the most loaded joint of your robot arm, you know which one... And the hardest part here is a reducer (which could be solved by buying a batch of higher-quality load-bearing gears from a sympathetic machinist, or you could borrow ideas from SEA actuators and use a long ball-screw rod as a component: [1][2], or you could borrow ideas from [3] ). Peak motor power could be improved by an order of magnitude with water cooling and good enough power electronics, and it has been done: https://spectrum.ieee.org/schaft-robot-company-bought-by-google-darpa-robotics-challenge-winner Research-grade actuators of artificial muscle are for now useless, and look like a waste of time. I hope to see an excellent BLDC-actuated lightweight robot arm being designed and validated in our group, and I very much respect the prototyping of such devices being built by fellow anons. If/when you are serious about the alpha-version of the arm design, we can start modeling it in mujoco. 1. https://iopscience.iop.org/article/10.1088/1742-6596/1633/1/012052/pdf 2. http://www.romela.org/wp-content/uploads/2015/05/2014_design_of_a_compact_lightweight_electromechanical_linear_series_elastic_actuator.pdf 3. https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.435.2907&rep=rep1&type=pdf

Excellent servo design: https://www.sciencedirect.com/science/article/pii/S2095809921005282 If we could manufacture a similar one, a question of hip actuator would be solved. There is a lot of metallic machined parts and a lot of gears (some could be bought though as these are relatively common planetary reducer gears), so a sympathetic machinist or a hackspace with a good machine shop is required. I also have some background thoughts about the backlash introduced by the planetary reducer (as opposed to cycloidal). But this is a straight solution of the powerful leg servo problem.

>>17025 Do I understand this correctly, the gear has a second motor attached to it, so the second one can help in certain situations?

>>16996 Thanks for the PDFs >>17025 This is very similar to how planetary gear transmissions are used in automotive applications. https://youtu.be/9sPi7DEbyI0 https://youtu.be/xOiAfOH-fU8

>>17026 Yes, there are two motors, the second is much smaller and its only purpose is rotating the ring gear with set speed, to regulate the transmission ratio of the main reducer. This way the system can balance torque vs speed as is required by the current situation. >>17030 You are welcome! I think PCM in this system is a bit of a meme, and it would be better served by the usual watercooling. Would be cool to get some standard planetary reducers (they are used everywhere, even in electric screwdrivers) and try to use their gear assemblies for a prototype of such servo. You would need an encoder, there are now good magnetic encoders (austria microsystems has a lineup of these https://ams.com/en/as5040 ) and a controller. There are some open-source/open-hardware servo controllers but I haven't seen a very good one, though maybe it still exists somewhere. We would have to either take some pre-engineered solution and adapt it or design our own. I could help here, but in the end someone will have to manufacture a prototype and iterate on it until it works.

>>17036 >controller For BLDCs we have simpleFOC. If you want to move abway from BLDCs then that's maybe not an option. I understood it so, that for BLDCs you won't need a additional sensor to measure the motor position.

>>16996 I agree with everything, and I've been stuck on fixing my printer to be perfect for a few months before I want to print all the components to a nylon gear BLDC actuator that I posted to this board before. Remember that ball-screw actuators have a very high backdrive force and aren't suitable for safe human interaction without a connection in series to an elastic element or equivalent. Pneumatic devices are loud, need plumbing, and need pressure sources that make them not ideal for human interaction.

>>17063 >Pneumatic devices are loud, need plumbing, and need pressure sources that make them not ideal for human interaction. Yes, though that japanese humanoid https://en.wikipedia.org/wiki/Actroid managed to keep noise low somehow. Also I expect our planetary drive reducer to by very noisy, and I don't see non-arcane methods https://en.nanotec.com/products/9893-quiet-gearbox-first-stage-plastic-gears-helical-toothing of removing this noise. Hope the anons will solve this problem eventually. >I want to print all the components to a nylon gear BLDC actuator that I posted to this board before. Cool, hope you will keep us posted. Expect difficulties, but we'll likely route around them with out DIY engineering mindset.

>>17040 Sure, the BLDC driver can measure the rotor position by native EMF sensing, but some method of absolute reckoning will be necessary anyway and a reducer is a nontrivial mechanism that may break and even if it works it will introduce backlash. Absolute encoder attached to the real joint solves this problem.

Related: >>17086 >Paradox Drive prototype >>17072 >Which BLDC for MaidCom?

I'm never sure if gears belong in this thread for actuators, but I think so. This video here https://www.youtube.com/watch?v=TzJkD87eQNI made me understand why planetary gears are interesting for robowaifus. I thought most of us would go with cycloidal drives, and ignored the planetary gears. Maybe because I one read something negative about backdriveablity, idk. However, different gear ratios in one gear are of course very interesting. Still need to catch up on the "Paradox Drive", though.

Drilled a hole in my 50$ 8318 and superglued in a magnet for the hall effect sensor. More progress to come on soldering to a moteus and printing the controller bracket before actuator is operational. If you guys put one of these in the hips, whoo boy, I can't remember the calculation result exactly but I think 80 amps or so and a 8318 bldc gives 4Nm torque without a gearbox. Going to assemble it and then test the torque with a cheap aluminum arduino force sensor.

>>17201 Thanks. Great to see some progress, I hope you succeed. I'm surprised you do this without a gearbox, and I'm curious about the results. (However, please try to size down your images, they don't loose detail while being 20% of that size, for Android 'Imagepipe' works quite well but it might be from F-Droid)

>>17206 >pic related Not my best work, but the power is soldered together with 100 amp banana jacks. Once I re-print the bracket for the controller I will do a test to see if it works.

>>17217 Calibration test: >https://u.smutty.horse/mkibuzgvdao.mp4 It works. Next step is trying positions after a bench mounting setup is made.

The 8318 motor plus the bracket and controller weigh around 800 grams together, and I have found that most knock-off mini cheetah robot dogs use 8108 motors. These thinner motors can sustain 22+ amps constant current vs 60+ amps of 8318, and weigh around 2-300 grams vs 5-600 grams. I don't think 60 amps are required for normal movement, and I will find out. It's clear that the meta for quadruped actuators is to go as wide and thin as possible to maximize torque per amp/weight. The 8308 motor I'm doing a test with costs about 80$ instead of 50$ (8318), so we will see if the trade-off is worth it. There is an 8110 motor for 45$ which may be the sweet spot. Biped robots could require twice the torque per hip actuator of a quadruped for the same weight for all I know. It's physics, right? I have an idea for a hydraulic silicone mckibben muscle actuators, and I will replace these rigid metal actuators in my next robot iteration if my idea works. The pure silicone actuator should solve the issue of having a rigid metal motor in your waifu's hips when you try to cuddle her. Bldc motors will work better for a turn-key iteration 1 proof of concept.

>>17742 Links to these cheap 8xxx motors? 45 seems stupid cheap for that size.

>>17742 Thanks for sharing your thoughts!

>>17743 Aliexpress has scuffed and probably rewound motors for cheap but they work: >https://aliexpi.com/zVkc You still would need an expensive high-amp bldc controller. I see potential in the frameless motors linked here: >https://aliexpi.com/w0y0 I might outright skip the integration of 8318/8308 motors in order to begin building a robot leg with the motor and gearbox fully built into the motor hole itself. It would look like a pancake with a thru hole instead of a big stupid can of beans for the hips. I will try an "MBS" magnetic encoder in it as well since simple FOC controllers use a center axis magnet that won't work on a servo with a thru hole. Idk what I will use to drive it, maybe a moteus hooked up to the magnetic ring encoder for now. The frameless motors are twice as expensive, but have more torque for less weight at 100A (8318 vs fl106-12), almost unlimited design freedom, come with what looks like an absolute (optical?) encoder for a few extra bucks, and you might only need 2 big ones for a bipedal bot. They are compared to low current capacity 8108 motors here: >8108 1.25 Nm @ 20A, fl106-12 2Nm @ 9.6A with headroom for 27Nm @ 104A >8108 ~245g, fl106-12 328g, 8318 ~650g I've read some papers that suggest a robot leg moving relatively quickly needs high peak torque values in the hips, and I have been attempting to model a leg with the software "Yobotics! simulator instruction set" as is common practice in a few papers. Will report back with findings. >tldr 8318 are heavy and bulky, 8110/8108 are not powerful enough for hip motors, and fl106-12 might be perfect for twice the price. >silicone actuators My attempt to make "straight fiber mckibben actuators" by using stiffer silicone down the length of a long silicone bladder did not work. The stiffer silicone still stretched lengthwise and was not stiff enough to produce an axial contraction. My next experiment will be something inspired by these "zero volume air chamber" muscles: >http://jobirobi.com/actuator/vujpwyio11e3qkjhivnwd32ap0cz4w What I might try is stacking the air chambers vertically and staggering them along the length axis to make them more homogenous when inflated.

>>17882 >scuffed and probably rewound motors What does this mean? >potential in the frameless motors Could one make these parts with a CNC machine? >have more torque for less weight at 100A How much Amps can we get from some batteries? >robot leg moving relatively quickly needs high peak torque values in the hips I had the idea of adding a motor on the back of the hip, pulling strings which go to the legs. The motor would only rotate a bit to one or the other side, then reverse and pulling the other string.

>>17938 >scuffed motors The ones they sell on aliexpress are scratched and the windings appear new, they also sell burnt out motors of the same size used in agricultural drones. I think the Chinese agricultural drones are ran until they get all scratched up and burnt out, and then they rewind and sell for cheap. >CNC these parts If you mean the robot parts that attach to the frameless motors, then yes. I would be doing that right now if I had a CNC machine because 3d printing a leg and gearbox seems like a bad idea. I'll probably end up trying it anyway, failing, and then moving on to something else. I read that there are tight machining tolerances needed for integrating frameless motors in a robot, but tolerances are for the weak. Just go for it. >batteries >https://www.rotordronepro.com/c-rating-drone-lipo-battery-packs/#:~:text=The%20C%20rating%20for%20a,and%20that%20means%20higher%20performance. Search for a lipo battery with the voltage and C rating you need. The controller you choose will likely be the voltage and current bottleneck. >motion coupling I think they call that idea a bowden cable mechanism. There are pros and cons, and it would probably be hard to execute. Most of the people who make 2 joint legs use 2 motors on either side of the hip joint, with one motor driving a belt to the knee joint.

>>17971 Thanks for your answers, but it leads to more questions. >burnt out motors of the same size How do they burn out? Abuse? Overheating? >then they rewind and sell for cheap. Does this mean some of the claimed torque and such is likely to be fake? >Search for a lipo battery with the voltage and C rating you need I just asked because some of the ampere ratings you mentioned implied very huge batteries. >motion coupling, bowden, ... would probably be hard to execute. I don't see why and hope not. Walking will need more torque than bending her legs while lying.

>>17986 >Walking will need more torque than bending her legs while lying. Actually, not AFAICT. If we can closely mimic the musculo-skeletal kinematics & mass-dynamics of a normal, walking adult female (pretty much our target goal anyway), then the forces required should be relatively minimal. About like keeping several rudders steered close to point, as opposed to doing a hardover to leeward, etc. Think about it (and try it yourself) Anon. Once, you're up and standing balanced, how much effort is it taking to keeping you going forward in a normal, casual walk? Not much really. Now getting upright and many other forms of kinematic motion may perhaps be strenuous (and by extension--hard on a robowaifu's batteries & cooling systems), but the train of motion req'd once you're there is astoundingly small. Basically, your bones support most all of the strain against gravity. 'Controlled-falling' if you will. The whole systems-engineering of bipedal locomotion for us is really a miracle of design & aesthetics! :^)

>>17989 I agree with walking being energy efficient and getting up being more demanding. If she's is lying down and moves her legs, then there's no weight on them, though. No upper body on top, especially not being lifted up.

>>17995 Heh, fair point.

>>17986 This is only a guess, but judging from this page here: >https://aliexpi.com/r59X >This is Xaircraft A12 motors, The motor maybe can not work, if know how to repair , you can change the bearings and rerolling the wires The dead motors have grey spots in the wires and some of the bearings are probably stuck from the description. The bearings and wires are replaced, which is why the bearings and wires look new on the motor I bought that has a scratched and worn casing. I bought a functional 8318 motor for a little more money from a different page. I am building a test bench to measure torque per amp. >batteries You could get a high C rated battery with less power capacity to save weight. >bowden If it's connected via a tube to a link further away, the wire may have to move with the link. Think of a wire pulling on an arm that's being moved with another motor. This is the basic principle of a belt drive, which is predictable system: the belt is in one axis, and any rotation on the first joint would turn the next joint where the belt is connected the opposite direction, provided the motor driving the second joint is located on the axis of the first joint. Let's say there's a hip joint motor and a knee joint motor on the same fixed skeleton and axis of rotation, and the knee joint motor is connected to the knee joint by belt or bowden cable. Where would the shin point when the hip joint rotates, and the knee joint motor is held still? The shin would always point towards the same direction. If the hip joint rotates both the leg link and knee motor while the knee motor stays stationary, then the shin rotates with the hip. Imagine the bowden cable being slack and pulling on a joint 3 far away from the motor's axis. It would be hard to factor in the angle of each joint plus the bowden cable slack to position the joint it's powering.

>>18005 >>The motor maybe can not work, if know how to repair , you can change the bearings and rerolling the wires Thanks, maybe it's worth trying. I once though about building a machine doing this wiring if we need to build our own motors. >Bowden... >It would be hard to factor in the angle of each joint plus the bowden cable slack to position the joint it's powering Not sure if I was able to completely visualize that correctly, but I think the gist is, that such bowden controlled movements are hard to calculate. So, my guess is then, that it will need to be learned by trial and error, maybe under some guidance, and the system would need to use that data somehow.

>>18019 >bowden cables for robotics >https://youtu.be/ahSS5HUylT8

Limited access to magnets, for whatever reason, could be a showstopper for us. Even if we had copper wire and machines to shape the steel plates for electric motors. No magnets, no servos. Good news, some young guy build a better synchronous reluctance motor - it doesn't need magnets: https://interestingengineering.com/innovation/17-year-old-motor-outperforms-traditional-designs Context: https://youtu.be/ojJHwi3YIus

>>18278 Nice! It'd sure be nice to have this anon participating with us here ITT.

Allegedly muscles which are 17-times stronger than human ones. Equal strength, but useable, would be good enough. >Artificial muscles are indispensable components for next-generation robotics capable of mimicking sophisticated movements of living systems. However, an optimal combination of actuation parameters, including strain, stress, energy density and high mechanical strength, is required for their practical applications. Here we report mammalian-skeletal-muscle-inspired single fibres and bundles with large and strong contractive actuation. The use of exfoliated graphene fillers within a uniaxial liquid crystalline matrix enables photothermal actuation with large work capacity and rapid response. Moreover, the reversible percolation of graphene fillers induced by the thermodynamic conformational transition of mesoscale structures can be in situ monitored by electrical switching. Such a dynamic percolation behaviour effectively strengthens the mechanical properties of the actuator fibres, particularly in the contracted actuation state, enabling mammalian-muscle-like reliable reversible actuation. Taking advantage of a mechanically compliant fibre structure, smart actuators are readily integrated into strong bundles as well as high-power soft robotics with light-driven remote control.

>>18331 Thanks Anon, great paper! (BTW I really appreciate when anons post the actual documents here. :^) If researchers can manage effective implementations of artificial muscles using various carbon forms like graphene and CNT, then we can begin to seriously imagine the sci-fi notions of superhuman-strength robowaifus in the not-too-distant-future IMO.

>>18339 >superhuman-strength I was more thinking about it in a way that strong muscles will allow to make those muscles smaller and therefore allow more tissue or space being used for energy storage. Maybe it's also more energy efficient to use them within a more human like frame. Also, it could be used for redundancies: Muscles would break down over time, but she would still have more to which she can switch over.

>>18341 Yes, you are absolutely correct anywhere within the near-horizon timeframe, Anon. We are pressing in so literally on the edge for at least a dozen (of the even-more) domains we are tackling together at once in devising our robowaifus -- that even slight improvements should be 'redistributed' as it were back out into the systems as a whole. For the physical robotics parts of the equations, I'd say that 1. mass 2. energy 3. power (ie force) are the top three areas needing extreme efficiencies here (and in that order (>>4313)). >Muscles would break down over time, but she would still have more to which she can switch over. Redundancy is important, maybe even critically-so for certain areas. Any type of 'self-healing' materials or systems we can use will also be of great benefit in general. Beyond that, it certainly rests upon us to also engineer in ease-of-maintenance. From a systems engineering perspective (>>98), this last priority has always proven to be a yuge boon for big projects if its done well (and the converse is also true). >=== -fmt, prose edit

>>18331 I am pleased to finally come across a "soft" robotic actuator that takes a second or less to actuate and even less time to reverse. I see scaling problems with using radiation to heat a thick bundle because the energy might heat the outer fibers more. I would like to give this a try, but I didn't see the type of elastomer they used, and the graphene filler has its own set of problems to use in a DIY environment. Maybe if a silicone could be altered with liquid crystal particles, used as the elastomer, and (available carbon fillers: carbon black, graphite, chopped carbon fibers), it could achieve softness suitable for cuddles. The last thing I want to say about using cutting edge research is that it's usually unoptimized and hard to implement into a final product, I tried this and failed terribly due to the paper's assumptions (big math formula) that did not hold up to my scenario (I probably did the formula wrong).

>>18392 Thanks for your insight, and for trying.

After watching the Cutieroids video again >From Anime To Reality: Embodying An Anime Character As A Humanoid Robot https://www.youtube.com/watch?v=HEOOpia1EP0 During the Q&A segment, he was asked about Hatsuki walking. He was talking about the limited actuators available that had sufficient torque, etc. So this got me thinking about the incredibly high amount of linear force that relatively-primitive, screw-based, linear actuators can provide today. I mean like tons of PSI with a machined steel system. What if we could arrange two of these vertically in the lower half of a robowaifu's torso, oriented into her hip joints using strong cabling to move her legs around. A) Can we still get enough force to move her if it's a 3D-print with cheap systems (kind of a must for our overall goals here on /robowaifu/ tbh)? B) Can they be made to move quickly enough that she won't seem to be walking through molasses? There are other issues about balance, kinetics, etc., but those are better-suited to the Bipedal Locomotion thread (>>237) on another day. My current concern is more about can we manufacture a pair (quad) of cheaply-manufactured linear actuators with sufficient power & speed to move Hatsuki (and all our own robowaifu designs) around effectively & pleasingly.

>>18421 I think these lead screws are available to buy. But I'm skeptical about your idea, I don't see why using a linear actuator that way would make sense.

>>18424 >I don't see why using a linear actuator that way would make sense. Because of the issue that business guy mentioned (roughly): >"No appropriately-small & appropriately-powerful servos to drive walking in this full-sized robowaifu'' Tucking the actuators up into the lower torso deals with the thrown-weight issue, and their potential for high-power solves the torque requirements for leg-actuation. The primary issues to solve are the two I mentioned: 1. inexpensive production 2. high-speed actuation

>>18421 Linear screw actuators are ideal for applications requiring high accuracy and force. A humanoid robots legs would benefit. Teslabot uses an advanced ball screw actuator as a good example. 3D printed linear actuators frankly don't make sense for waifus. Filament prints have highly uneven surfaces that generate noise and heat, reducing efficiency. Luckily, because 3D printers have made steppers and lead screws cheap and abundant, they can be adapted for our use easily. A NEMA 17 stepper with a leadscrew can efficiently provide newtons of force. They require constant current to hold position though. Natsuki Cutieroid will never walk simply because her design prohibits it. She's a heavy lumbering show piece. Which is fine, she's built to be beautiful rather than functional.

>>18441 >Luckily, because 3D printers have made steppers and lead screws cheap and abundant, they can be adapted for our use easily. Hey you're right! They're everywhere. Hmmm. What about this then: putting the NEMA 17's (four of them?) up in the lower torso, lead screws directed near to the hip joint areas, and some kind of teflon-lined 'ball-socket'(term?) on the screw ends, which connects to and drives a lever action system which in turn moves the legs? By using a lever system rather than cabling, this should simplify increasing the actuation response-rate of the legs. This is needed since the screw actuators can deliver abundant force, but only slowly. The tradeoff in design here is higher strains locally at the joints & connectors, needing tougher/heavier materials; and also increased strain on the steppers (and batteries), leading to more heat & shorter stepper-lifespans. A cable-pulley system could ameliorate some of these issues, at the cost of greater mechanical & design complexities + increased maintenance costs. Slower response-rates, too. Actually, I think if we can devise a system where the levers are the leg's skellingtons, then we've probably found the sweet spot tbh. As to 'locking-off' the leg positions, why not devise some sort of clamping system that will act as a brake (maybe a tiny, actuated, alloy disc-brake system fastened orthogonally to the screw-axis?) right on the screws themselves? That way the steppers (and the batteries!!) can take a much-needed 'rest'. :^) BTW, it seems to me this would also be an efficient approach to the 'back-driveable' safety problem as well (ie, just 'let off' the brakes in a moderated fashion). Great input Kiwi. BTW, can you link to technical sauce on Teslabots pls? >=== -prose edit

>>18441 >Filament prints have highly uneven surfaces that generate noise and heat, reducing efficiency. What about wrapping every printed surface that must slide against each other (such as a linear screw actutor's) with a metric-boatload of plumber's tape? Surely this would greatly reduce the friction coefficient between the two surfaces?

>>18441 >Linear screw actuators are ideal for applications requiring high accuracy and force I'm not convinced that walking is about high precision motors, though, but I might be wrong of course. I think sensors and fast reaction is more important. That aside, I always saw bipedal walking without help and long distances as a luxury for the more advanced models. Ignoring that requirement has always been one of the points why we can pull it off in the first place. She either rolls, has an additional device or gets help. >Steppers Steppers are heavy compare to their power, though. Therefore they might be good for waifus which aren't mobile, or not much, or maybe for rolling waifus with slowly moveable joints, but not for walking. I feel we're chewing up the same points over and over again. >They require constant current to hold position though Sorry, but this is clearly not the case with a stepper plus lead screw, otherwise 3D printer arms would move down automatically after you turn your printer off. >>18444 Okay, good luck with that. I just don't get the part where you have lead screws but make them pull something, instead of getting the servo closer and use gears. Ironically, you were the one always preaching how mass matters and that it should be close to the hips. >>18449 Sounds good for parts where it fits, but ball bearings aren't that expensive either. We also have the options of metal plating and ceramical paint. If it's a pole rotating in a hole, then a little metal tube cut added to at least one side might be sufficient.

Someone asked in anther thread about "inexpensive linear-screw actuators" and I just had I think a really good idea. Ok the thread could be made of "all-thread". This is a standard material in all mechanical supply houses and you should be able to get it in the electrical or plumbing department of big box DIY stores. It's just what it says, a long rod of thread. Use nuts to crawl up the thread by rotation. I'm constantly going on about switched reluctance motors because I can see nothing cheap. High silicon iron, conductors, a couple MOSFETS, a microcontroller and coils of wire for the fields. Tesla is moving to a hybrid switched reluctance motor(SRM) because nothing has more power or less cost than this, so far. Here's a great brief article on SRM. https://www.controleng.com/articles/resurgence-for-sr-motors-drives/

Cont. If the (toothed)rotor is attached to a nut and a big washer AND the inner rotor is attached to all of these when the rotor rotates it will crawl up or down the all thread. If you have washers above it then the washers will squeeze the silicon or foam or whatever used to simulate muscle and it will fatten as it crawls up the rod and look much like regular muscle. The stator will be attached to the muscle to keep it from rotating.

>>18771 >switched reluctance motors because I can see nothing cheap should be nothing "as" cheap

Instead of steel rod, all thread, I think the best way is to put the SRM inside the bone. It would have a string that rolls up on the rotor of the motor inside the bone. Just like a fishing line comes from the top of the rod, makes a 90 degree turn and rolls up on a spool. The string goes out the bone and makes a 180 degree turn. What it runs through is foam that has a covering. When the foam, rubber, silicon, whatever the simulated muscle material is, is pulled by the string from inside the bone it contracts AND it expands like muscle. It would look just like this air actuator. https://en.wikipedia.org/wiki/Pneumatic_artificial_muscles Imagine, instead of air pushing the walls out, the process of shrinking the foam muscle (pulling the ends in) would push the foam out looking the same as a muscle contracting. It wouldn't have to be foam. It could be air bags with a stiff covering just like that on the air muscle. The process of shortening it would cause it to expand like muscle. I give a lot of calculations on how many microprocessors, MOSFETS, forces needed and stuff like that here. Rough calculation on cost, forces needed and the number of muscles needed for full human mimicry.(the micro-controller cost might be less) >>12014 Rough equivalence to the force Tesla gets out of their cars to human actual forces >>12140 Rough cost budget of actual existing transistor and micro-controller parts to get full human mimicry >>12172

Read the article on the SRM I linked above. These things are like magic. Huge RPM with super fast control and lots of power. One of the advantages of putting them in the bone is that you could mass produce a limited amount of sizes. Say five. They could stacked for extra power. So little muscles, little motors turning a solid rod, (on which the string is wound up to pull the muscle). If you need more power you just slide another motor down on the rod and run it in series. Power leads could go down through the bone saving you a lot of trouble routing. The only thing I can see better is "Dielectric Elastomers" but I don't really know how to make these easily. They require further research. They would be great but how to make work???? Links Here's a link to a good book on these called "Dielectric Elastomers as Electromechanical Transducers: Fundamentals, Materials, Devices, Models and Applications of an Emerging Electroactive Polymer Technology" >>8502 Dielectric Elastomers can also be used as sensors A good page with links to 100 papers on Dielectric elastomers >>8505

>>18771 >switched reluctance motors I mentioned those before (not claiming to be the first one here on board). Recently they came up in the news, but I can't find the link. Anyways, costs and availability are the issue. Where can I get them? Are they even working with low power and voltage or does it only matter for something like a car? >>18772 I thought about faking muscle movements before and like how you are thinking. I thought about using air for it, but I'm open to ideas. Anyways it's for later optimization. For now we even didn't build a hard shell robot. Also, please keep noise in mind.

>>18781 >I mentioned those before (not claiming to be the first one here on board). Recently they came up in the news, but I can't find the link. We have a tool for just this type of need, Anon. (waifusearch v0.2a - >>8678) >reluctance Also, there's a new entry in our library for this word now (>>18786).

>>18775 This sounds like an innovative concept Anon. Any chance you could make a sketch of your ideas here? >I give a lot of calculations on how many microprocessors, MOSFETS, forces needed and stuff like that here. If you can share a brief >tl;dr consolidation here for this new idea here too, it would probably be helpful to us all. >>18776 >Say five. They could stacked for extra power. If we can stack motors vertically up the inside of bones, that would be a big win for us all regarding volume. Two for one and all. Volume consolidation is always a critical consideration for us, expecially during these early prototyping years.

>>18776 >Dielectric Elastomers New library entry for this one too, Anon (>>18791).

>>18787 >We have a tool for just this type of need, Anon. >(waifusearch v0.2a - >>8678) Ha, I knew if I would write this, that this would be the answer. :) You're right, I just didn't install it yet on my other computer.

>>18781 >Anyways, costs and availability are the issue. Where can I get them? Are they even working with low power and voltage or does it only matter for something like a car? I thought about faking muscle movements before and like how you are thinking. I thought about using air for it, They do make them smaller. It's works for ANY size motor. It appears to me that most all motors are moving to SRM. They are in air conditioners, and all sorts of stuff. The reason is microcontrollers. Without them they are noisy and rough. You have to use the MC to create a pulse width modulation PWM output that's linked to the speed of the rotor and what speed you want the rotor to get up to. What this is is a fake sinewave. Normal motors, some anyways, you can hook right up to normal house power. Can't do that with SRM. You have to control the rotor with a computer. But some of these are like less than a dollar. The one I keep pushing, and I have no financial interest in them, is anywhere from $4 individual to to $15 each for the board version. The motor itself I don't know where to buy them. The are mostly embedded. The good news is you can make them from sheet metal, or better transformer type metal sheets. For DIY I think the best way would be to shear the sheets out with a die and a stamp. I'm still thinking about this. maybe you could make serviceable ones with washers cut in two. Here's I think a better paper than the first I linked. A significant quote,"...Switched reluctance motors exploit the fact that the forces from a magnetic field on the iron in the rotor can be up to ten times greater than the magnetic forces on the current carrying conductors....". That's a lot. https://www.machinedesign.com/motors-drives/article/21826520/are-switchedreluctance-motors-for-you and the quick link https://en.wikipedia.org/wiki/Switched_reluctance_motor

I don't care for Pneumatic muscles for what I think are good reasons. They are not very efficient. They are noisy. They can be slow unless you have higher pressures and big pressure reservoirs. The valves must be controlled wioth a transister and a microcontrollor. Well then you buying a valve AND the same parts needed t do something electric which will be likely more efficient, faster and less costly. These pneumatic valves are not cheap. The problem with SRM s you will likely have to make them but sheet metal will do for starters and experimentation. Later you can get better materials, coils, etc. These rotors are just flat sheets of metal glued together. That's all. Look at this video to give you some idea, https://www.youtube.com/watch?v=g3qWBmz-j2k This is a ford not Tesla, see at time 11:05 he is going to show you the sheet metal laminations. This has magnets but you do not have to have them. I think they are for better starting torque but I don;t know for sure. We will likely leave these out for cost reduction. And make sure you understand this is just one configuration. The rotor can be shaped many ways. Although if the car companies are doing it this way likely a good reason. You can also see if you have a metal stamp you could stamp out the slots and then glue these together.

BTW this is Sandy Munro. He's been in the manufacturing business for decades and his company takes things apart and provides cost accounting and also he helps companies cut cost and raise quality by redesigning things to be low cost and fast to put together, He's great and very knowledgeable but not so charismatic. Old school engineer. He has a bunch of Tesla tear down videos where he shows you all the guts. Chobitsu asked for some drawings and stuff. I will but it may take a little time.

Thanks for the waifusearch v0.2a I was looking for just that a couple days ago and couldn't find it.

A video that shows how Tesla motors work. It's a modified SRM https://www.youtube.com/watch?v=esUb7Zy5Oio

>>18824 MAGNETS; How do they work? That was interesting and helpful, thanks.

>>18822 > Sandy Munro Very impressive engineer tbh. >Chobitsu asked for some drawings and stuff. I will but it may take a little time. No rush. It will be useful for understanding but we still have time before it's needed I think.

>>18823 >I was looking for just that a couple days ago and couldn't find it. Be sure to use the latest JSON archives with the tool as well. Both are always linked in our Library thread, Grommet (>>7143).

Another good video. This one is not super clear. The subject is complicated but it's all there. This is one of those you need to watch several times to grasp. Tesla Model 3's motor - The Brilliant Engineering behind it https://www.youtube.com/watch?v=esUb7Zy5Oio

You'll have to suffer my retard drawing skills. I can't draw and I'm using a trackball to do so. The drawing shows a bone with internal motor. The motor is mounted on a rotating shaft. The shaft has a cable wound around it. The motor turns, winds up the cable on the shaft and pulls the cable. Its very much like a fishing pole reel that winds up on a spool but in this case it's a small rod it winds up on. The small rod means lots of torque for a small motor. In this case our motor is fast, so winding up a small rod is quick. The cable goes through the bone end where on a human the tendon is attached. Cable travels up to bone it is pulling and is attached to it. So it pulls much like a muscle but the pulling part is the motor inside the bone. Around the cable is a silicon, air inflated bag, foam, or whatever to simulate muscle. This is covered in a fairly non flexible weave just like in the air muscle I linked above. This gives the appearance of the muscle expanding just like real muscle. In fact it doesn't but it looks like normal muscle expansion. I will add another post next with another drawing to flesh out the idea a little.

This drawing shows how with three motors and cables you can control any joint. The top is a side view. The bottom is a bit of a retarded view looking along or somewhat down the bone. Maybe you will get this additional idea. There may be in a human more than three muscles controlling a joint. But understand you really only need three motors for any joint positional movement. But to simulate more muscles you could put more than one cable on each motor spreading out to "simulate" more than one muscle. Each additional cable would have an attachment point like a normal muscle and have fake muscle foam, or whatever, to show it as expanding like normal muscle. It may be that this is not needed. I haven't studied intensively all muscles. But even if more are needed we can fake it with only three motors per joint but adding extra cables and foam expanding muscles. Now I show this second drawing with all internal cable pulls attached to the bone internally. In fact the more accurate robowaifu simulation would be like the first drawing with cables going through the bone and exiting out of the same area on a human where the tendons attach, then attaching at the top of the muscle on the bone being controlled. Just like the first drawing. It would look just like real muscles and tendons. Some advantages of this idea. The motors could be made in a few sizes and be mounted on the rod cable spools. Make a small size motor and then add motors for more torque by making the rod/spool longer with more motors. The mounting, the motors and rods all being the same. Saves you from making a huge amount of different motors. Make say...three sizes. All the rest of the changes in torque could be made up by varying the number of motors mounted on the rod. Having motors internal makes for a sleek look, less noise, and you can detail the covering without worrying about what to do with the motors. Power can be routed through the bones once again making a neat package out of the way. Mounting the motors as I did is for a reason. Our motors must be small in diameter to fit into the bone, so the torque of the motor is dependent on how long it is. By making a long vertical motor we can get good torque in a small diameter package. Also as noted they can be stacked.

>>18895 >>18896 I see. Excellent, that gets the ideas across really well Grommet. So do you think multiple moters can be stacked up, each one tasked with pulling different 'ligaments'?

I made a mistake saying only three motors needed for each joint. While trying to count how many I needed I realized, Duuoh! that they need to twist. With just three I don't think you can do this. So I'll have to rethink this. Three motors for most movement will work but some extras will be needed for twist. Off hand I think you would need 6 motors to twist in a human like movement which is what I'm aiming for.

>>18897 >you think multiple moters can be stacked up, each one tasked with pulling different 'ligaments'? EXACTLY! This also means the motors can be aligned along the inside of the bone with routing of the ligaments in different areas. This gives us more diameter for the motors because they are stacked. Makes no difference the length of the ligaments. They all exit the bone from the normal human tendon location so they can be routed differently. I envision the bone being made in a split fashion. The inside of the bone would have mounting points for the different motors and controller. After mounting motors, MOSFET power transistor, micro-controller, routing wiring, connecting ligaments, you put the bone together. It could be wedged like wood working, with a few small screws, just like pegs in wood furniture construction to lock closed. It could be rough as the whole thing will be covered. Heat may become a problem depending on work load. May have to put controllers on one end and motors on other. Possibly might have to have some sort of water circulation for cooling. This is going out there a little and is further down the line but maybe the heat from these could be routed to bladders that are our fake muscles. Making waifu warm. So the excess heat from motors, controllers, etc. warms the waifu. Further down the line, but thinking out loud about it. For "lite" work you could probably get by with some metal heat spreaders connected to the bone material. If you are going to have it do heavy work, a lot of heat to dissipate. Maybe 400 watts or so. 400 watts is human athlete working fairly hard, so a lot. One horsepower equals 746 watts. A horse is fairly powerful but we will want that kind of power to use waifus for elderly care. BIG market and will also keep people from banning them because the elderly vote like crazy and if you try to ban their waifus they will revolt. Hair dryer is around 1,000 watts so that gives you an idea of the heat created. I think in the beginning it would pay great rewards to add temperature measurement interior to the bone to see just what kind of temps are created and then work up some sort of programming that stopped it if it was getting over heated. If heat a big problem "then" think about liquid cooling or something like that. Any liquid stuff added is going to be a mess and should be avoided if possible. On the other hand I think it's inevitable that if you want warm human feel and touch you will eventually add some sort of circulating liquid. Maybe in the muscles as I said. Maybe there's other ways to heat and if they could be found and not use liquid it would be better. I see using liquid as the "no thought" first case, go to, idea but a far superior way would be to find some no liquid way so the thing would not leak all over. Air circulation maybe? I have no doubt if you fill it with liquid over time the thing will spring a leak.

On the elderly angle. No matter what you do you will be involved in politics and Women will try their best to ban these and they have more political power. So the sneaky cheat way to do this is sell them for the elderly but make the outer skin and unlayment so it can be changed to a sex bot. This is why I think the skeleton, the muscles should be separate. Over this a skin and simulated fat layer could be placed like a full body sock. Kind of like those stocking/pants that little girls wear, but full body. You could sell the same bot with no genitals and then add whatever you wanted. I've got all kinds of ideas on this. One I mentioned before is to hell with silicon. I think that microfiber stuff they make towels out of mixed with a little stretch fiber would be ideal. This stuff is really soft and washable easily. Train the waifu to wash itself by getting in the shower. And by changing the underlying fat layer you could change the look. New skin would be cost effective to change every so often.

The idea for the fake muscle is shown in Hik's post >>16629 and here >>14018 The arm attachment is shown in another comment I made >>13226

>>18899 >I envision the bone being made in a split fashion. The inside of the bone would have mounting points for the different motors and controller. After mounting motors, MOSFET power transistor, micro-controller, routing wiring, connecting ligaments, you put the bone together. It could be wedged like wood working, with a few small screws, just like pegs in wood furniture construction to lock closed. It could be rough as the whole thing will be covered. I think that's a reasonable approach to the design and construction for these robowaifu power-skeletons. Thrown-weight is always an issue, so you want to keep the mass of the motors as near to the bone's proximal pivot point (eg, the hip joint, in the case of a thigh bone) as feasible. Yes, it's a good idea to separate the heat of the motors and controllers -- spread it out. Since the controllers are far less massive, they can be situated down at the distal end of a bone w/o too many penalties I think. Also, glad to see you already thinking ahead about manufacturing rigging, Anon. Such tooling is always essential if you're building anything more than a one-off. >heat dissipation Yes, heat will always be an issue for our robowaifus. While I think it would be very nice if we could limit ourselves to a heat-collection system made up primarily out of flexible air tubing (for several reasons), in the end I believe the laws of physics will dictate that mostly we'll use water tubing instead. The specific-heat ratio between the two is crazy. We'll find an effective means to use liquid cooling I'm sure, Anon. >>18900 >On the elderly angle. No matter what you do you will be involved in politics and Women will try their best to ban these and they have more political power. So the sneaky cheat way to do this is sell them for the elderly but make the outer skin and unlayment so it can be changed to a sex bot. I've stated here more than once I believe that monetizing companion robotics at a big-time scale will happen first within patient-care scenarios. Those who simply want sexbots can already get those from existing suppliers. But we here want waifus; I think there is a huge amount of crossover for that with the needs of elderly & patient-care. So yeah. Good ideas about the modularity systems extending even to the shells & skins. That approach will be essential anyway once robowaifus become wildly popular, ofc. Good ideas, Grommet. Keep 'em coming! :^) >=== -minor prose edit

>>18903 That type of crosslinking is very helpful for robowaifu researchers, Anon. Thanks, and please keep it up! :^)

>>13229 I'll add one more cross link. It's figuring force needed to lift something and then converting this to watts. Very handy because that tells you what size of wire, what type of transistors you need, how many and what batteries needed to drive the waifu. Bunch of basic stuff. I believe as a general rule a good short cut to finding the amps needed at a certain voltage use, find pound force>convert to newtons>use amount in newtons to directly substitute for number of watts need for force Example in above link.

>>18972 This kind of stuff is always important and helpful. If we can keep the weight of things down, that's the best, first step I think.

Orca Series from Iris Dynamics: https://youtu.be/xZ7T1H1uPSc Any opinions on this? The smallest one might be small enough for a thigh in a human-sized gynoid. Not relevant this year, I guess, but worth to keep an eye one. >Orca 6-24V >Optimised for 24V >Peak Power 899 W >Max Force 394 N (89 lbf) >Force Accuracy ±0.24 N (0.053 lbf) >Positional Accuracy ±150 μm (±0.0059 in) >Acoustics < 20dB >Force Controlled >Traditional motors are position or velocity controlled. This means their logic controllers deliver commands to move to a specific position or at a specific velocity. Our motors are inherently force controlled which means they can feel the force they are imparting. This is a critical safety feature in applications where that force is imparted upon a person as our motor will know when it encounters an object and can back off or apply lower force. https://youtu.be/pmwMfgaGuyU

>>18980 That's remarkable Anon, thanks. I like how the design seems pretty rugged under challenging environmental conditions. Any idea what one of these actuators that might be tuned for, say, lifting a robowaifu's thigh from the hip might weigh?

>>18984 >lifting a robowaifu's thigh from the hip might weigh? I didn't look into their data sheet. I would just look for the lightest one. I most likely won't need that kind of actuator within the next one or two years, and if I consider it later then it might differ anyways. There would also very likely be another servo in the hip involve to get her to stand up. These fast and regular movements made me more thinking of walking and dancing. Well, okay now I look: >2.23 kg for the stator and 1.38 kg for the shaft More than I anticipated, though I guess the shaft might be able to be made lighter by using a custom design. So it might be useful for the later and heavier robowaifu designs, which isn't a surprise.

>>19005 >So it might be useful for the later and heavier robowaifu designs, which isn't a surprise. Yes, eventually we'll be manufacturing more automotive-quality-like robowaifus, once the basic & inexpensive robowaifus kits, etc., are are off the ground and flying along.

Potential ODrive (dual drive BLDC controller) alternative, for a similar price (compared to other copies on AliExpress) but maybe better shipping and such for Americans: https://sequremall.com/collections/odesc/products/odesc-dual-drive

This book seems to cover a lot of ground on possible technologies for robot muscles http ://www.sze.hu/~molnarka/EAP_ANYAGOK/Springer%20-%20Electroactive%20Polymers%20for%20Robotic%20Applications%20-%20Artificial%20Muscles.pdf >=== -disable hotlink

Thanks for the information Anon, these sorts of concepts are right up our alley here on /robowaifu/. However by necessity we try to keep related information together inside topic-focused threads, since that eases research here after the fact. Many of us find the catalog view quite helpful in that regard. We already have an active thread here on actuators in general (>>12810), which also includes the specific topic of electro-reactive polymers. I'll soon be merging your thread into that one. Thanks again for your contribution Anon, please do continue on here with us. >=== -prose edit -lol i linked ITT OP

I think it'd ought to be some kind of muscle. it can't be just plain actuators. Waifus should be soft.

>>19530 >Waifus should be soft. Definitely. We've tossed around several ideas over the years about how to do this. The general consensus at this stage of tech is that electric motors of one fashion or another are the only really viable choice for their power to weight, and especially in their ancillary control systems (basic batteries & smol controller electronics). While this too will likely change with time, at this stage it's probably our best shot. >tl;dr We'll put padding around our robowaifu's skellingtons, Anon. :^) >=== -minor fmt edit

>>19530 >Waifus should be soft. There can be different variants. Even the hard shell variant can have soft layers on top. Either way, we will need to use some motors.

>>19531 >over the years

>>19533 Haha, fair enough. Please show us the way, Sempai! :^)

>>12810 based thread. For the coming revolution will be glorious. As man continued till he made machines to fly, to move faster and to live easier. So too shall man's desire lead to him achieving independence and freedom

>>19535 Well I'm just saying strength is not really an argument against pneumatic muscles I think.

>>19558 >Well I'm just saying strength is not really an argument against pneumatic muscles I think. Totally. Point granted, Anon. The primary issue with so-called 'air muscles' is the other equipment involved, namely air compressors and valving. These are power-hungry, noisy, heavy components. Not to mention bulky. I personally hope we can in fact utilizing similar air bladders in such a way to make our robowaifus comfy for snuggles. Possibly even figuring out clever ways to make their volumes dynamically transform (as with muscle-flexing). But we'll certainly need to figure out how to do this type thing without relying on compressors, which are decidedly un-comfy IMO. :^)

>>19592 >thing without relying on compressors, I guessed for a moment I maybe might never shared my idea of using compressor tanks for paintball. But either I did or this >>1422 is from someone else. I also had the idea of using a small (foot) pump for air and drive it slowly with a motorized mechanism while the robowaifu is plugged in. Problem is, that this is of course extra weight and required space. Yes, these tanks can be recharged to a certain extend by using foot pumps, of course not as fast and as full than using a compressor.

>>19527 >Electroactive Polymers for Robotic Applications - Artificial Muscles - Springer These are really cool. I've commented a lot on electric motor stuff but only because I haven't really figured out a way to make these work. Not that I'm saying they won't. The electric reluctance motor stuff I've talked about is a kludge but something people could build more easily. Ultimately these active elastomers would be the perfect thing and far more ideal.

So to get an idea of what it would cost I'm looking at castable rubbers. The two I see right off are Polytek PlatSil 73-25 Platinum Silicone Rubber (80lb Kit) $1,524.00 and Polytek 75-60 Liquid Polyurethane Rubber (80lb Kit) $571.00 So right off you can see the Polyurethane has a huge cost advantage. There's a company called smooth on that makes a lot of this stuff that I've heard make good products of the same sort. I don't have any financial interest in them. They have this product called VytaFlex urethane rubber they have formulated for casting concrete. This might not be so bad because it apparently has good tear resistance. "...Tear resistance of different materials Substances with high tear resistance include epichlorohydrin[quick search of this stuff, it sounds toxic] natural rubber and polyurethane. In contrast, materials such as silicone and fluorosilicone have low tear resistance...." https://en.wikipedia.org/wiki/Tear_resistance Going to need high tear resistance. TPE. These might be even better. They make sex toys, for sure, out of these and I think sex dolls also. I haven't found a cost though and I don't know which kind they use. It's melted to force into molds so you can 3D print this stuff. https://en.wikipedia.org//wiki/Thermoplastic_elastomer I would think a 80 pound, 5 gallon pail might be just about the amount of material used for a full size doll so that volume might be a good comparison to make.

>>19601 Sure, IMO this is a reasonably-fesible alternative for the smaller-scale usage I mentioned, Anon. And in our general design approach for a RW Foundations (>>14409) robowaifu, she will have a so-called 'Dollhouse' (basically a docking, storage, communications, and charging station pod) where she could just as easily recharge her air tanks as well as her electrical batteries. Good thinking, Anon! :^) >=== -minor prose edit

>>19638 >Ultimately these active elastomers would be the perfect thing and far more ideal. You're right Grommet, it would be amazing to be able to drive our robowaifus this way. IMO, if/when we actually manage to do so, it will turn out that we are now beginning to come much closer to the energy efficiencies actually present in molecular-motor designs that were crafted for biological systems, including us humans.

>>19645 >I would think a 80 pound, 5 gallon pail might be just about the amount of material used for a full size doll so that volume might be a good comparison to make. That would be excellent if we can actually pare the usage down to just that amount IMO. Maybe even less eventually? (It would help to reduce that mass involved, basically.)

>>19653 One of the reasons I talked so much about electric motors is you could have fake muscles made of thin bladders of rubber type stuff and not whole silicon muscles. That stuff is expensive and might be the difference between $500 or even $1,000 just for rubber elastomer materials as opposed to maybe $40 for bladders Hey, just to get an idea I searched for amount of muscle in average women and got "...Skeletal muscle is the most abundant tissue in our body, accounting for approximately 42 and 35 percent of body weight in men and women respectively..." So 35 pounds 100 pound women. The thing about those elastomers is the contacts to feed them power. I don't think that's going to be easy. People study contacts even in regular circuits with copper wire. It's not as easy as people think. You might get something to work for a short time but very soon it would quit. This connection stuff can be really tricky. I'll tell you a story. I was dealing with some electrical stuff and ran across a guy who had a electrical contactor that would go out frequently. Like every 8 months. The solenoid it fit in was expensive but alone the metal contact that connected the electricity was $1,000 dollars. WOW! We were all trying figure a way to cut cost of this thing and I mentioned at that price he could get someone to machine a pure silver contact and use that. He said,"I tried that, it failed right away". So even pure silver would not work as well as this mystery metal that the company that designed the contractor used. So he had to keep buying these $1,000 contacts for his contactor.

>>19661 >as opposed to maybe $40 for bladders Yes, let's please go with this alternative Grommet! :^) >So 35 pounds 100 pound women. Interesting. Didn't anon do a study here a couple of years or so ago on such considerations? How much does a human hand weigh wasn't it? >So even pure silver would not work as well as this mystery metal that the company that designed the contractor used. So he had to keep buying these $1,000 contacts for his contactor. That sounds like some kind of massive industrial switch.

I will try to build a hasel actuator. These are the parts I got so far. Still waiting for the electronics kit, the baby oil and the foil paper and the electronics kit. The electronics kit is coming from China though and that usually takes a long time. Although it's already at customs

>>19803 Thoroughly looking forward to your prototyping test progress, Anon. Please make plenty of photos!

>>19803 Great, looking forward to see your process in the protoyping thread: >>18800 >>19804 Yes, but also please run them through a image resizing program, so they're ideally under one MB.

>>19818 >Yes, but also please run them through a image resizing program, so they're ideally under one MB. Sorry I don't have any such control after the fact. I can either spoil, or simply delete files. It's up to the poster to optimize images beforehand. Here's a good suggestion for a simple-to-use tool for such by anons. https://trimage.org/ >>19803 Anon, please note the above request? >=== -add hotlink

>>19803 Very cool. You using an electric fly swatter for the driver?

>>19645 Ecoflex is about 1000$ per 80lb, available in 18 Gal buckets on amazon. The reason I shill it so much is it's large support platform of research use and customer support. The material properties are far superior to regular pdms that you'd get anywhere else because ecoflex stretches way more and is way softer. Good luck getting polyurethane to function as well for bladders. In my experience of using it, you can do quite a lot of experiments using less than 1/100th of the small trial units (30-50$). The cost of silicone per hydraulic actuator is quite low, and the majority of the weight in any of these actuators is the working fluid. Ecoflex is a little lighter than water. I've got the final stage of hydraulic actuator in my to-do list, where it's a water pump to a slender carbon fiber braid around/in a silicone bladder. Cost per actuator and pump is very low, and I must complete pic related servo motor frame first. It will be the fated frameless BLDC actuator that is highest in torque per weight and current (65$ per motor if buying 1 on aliexpress). Trying to get the tolerance right is challenging and I need a lathe. This goes into the legs, the hydraulics go into the toy.

>>20885 >slender carbon fiber braid around/in a silicone bladder Wouldn't carbon fiber be too brittle? Could you use fish net made from plastic instead? Dyneema is really super strong. People use these for rigging on boats, replacing stainless steel, and nets. Very long lasting strong stuff. Here's a link on it. https://www.dyneema.com/applications/nets/aquaculture-nets For prototypes nylon would be good enough. Nylon would likely be super cheap and work fine.

>>20885 What is in your picrel? Looks like it's made of TPU

>>20885 Thanks for the very valuable info, but if this goes into the direction of a specific prototype, please proceed in the prototype thread.

>>20909 PLA for a motor frame prototype.

>>20911 Will proceed in the right thread if I post any more.

>>20928 >>21194

Development kit includes the actuator + BLDC control driver board. Looks like the same as the ones they used in their quad prototype. https://mjbots.com/products/moteus-r4-11-developer-kit

Clockwork has been a thing forever. It may still be viable.

>>21945 Yep I think you're right Kiwi. In fact in my autistic pursuit of low-mass in the extremities (the thrown-weight problem), I've been envisioning numerous designs using piano wire (or even deep-sea fishing line) arranged through guides and actuated within the central torso core. Keeping the majority of mass (eg, batteries and actuator drives) colocated as nearly as feasible to the robowaifu's Center of Mass (ideally between her navel and her upper hips) is important for many kinematic and mechanical-advantage reasons. BTW, do you have an higher-res images of this fantastic design. Based old Italian guy was really amazing tbh. Cheers. :^)

>>22086 > >300 actuators >Impractical, the mass alone would increase energy consumption. We only need one actuator per degree of freedom of the human body at most. >>No wires >We need wires, even if all actuators are clustered together somehow, wires are needed to connect the mcu's and power. I recommend twisted pair everywhere. This is in response to another thread but it dealt with actuators so I'll split my response with actuators response here. First mass has nothing to do with the amount of actuators. Those little tiny helicopters have tiny mass motors. Musk huge aluminium injection presses in his car factories have massive motors. The number is not relevant. 300 actuators as a number needed will likely be too small. Let's add them up for a full skeleton. waifa Number of joints to control hands 8 per finger x 5 fingers x 2 hand(wrist) forearm 4 x 2 Arm 4 x 2 shoulder 4 x 2 head 4 neck 4 back 4 x 6 thigh 4 x 2 leg 4 x 2 foot 4 x 2 toes 8 per toe x 5 toes x 2 total 240 This is not even counting the face which to have pleasing to the eye waifu will need a lot on muscles. And don't forget we will want some sort of serious trick turbo muscles in all three mucus orifices. It depends on what you want. You want something that looks humans you will have to include all these. You want R2D2 you can have much less. As for wires I did include power and control and we will need sensor wires but the idea is to have motors if at all possible fixed into some solid structure like bone with the control board fixed directly next to it. Look at this bladder picture here >>16629 If you pull on a bladder such that it contracts it will be like the picture and look like real muscle even though you are only pulling a string on the opposing side of the muscle. Here's an arm picture here >>13226 here's where I show a picture, poorly drawn, of a motor inside a skeleton pulling a string(tendon) to show sort of what I'm thinking of >>18895 >=== -fmt edit

I just had what I think is a brilliant idea. After adding up all the very large number of muscles to gracefully operate hands like a human I got to thinking that it would be a LOT of work for the network to do this. Then an idea popped into my head. Bézier curves . Like are used for smooth curve drawing on computer screens. So after thinking about this for a while I looked it up and low and behold other people say they have been used for robots. SO it's not a complete idiotic idea. Now I had already thought about a compressed way to move feet and the waifu body. The basic idea is to tell the feet where to go and how fast. I can see now that Bézier curves would also work here to have the body move in smooth motions. Since you are only sending a end point, starting vectors, (say you need to step over something you want a high Z axis vector), and velocity vectors the actual bytes sent are not so big. You let the local micro-controllers figure how to make this work. I talked about this here,(It's not complete. Needs a little work but I think the basic idea is sound) >>21602 Now let's talk about the fingers. So just like the feet you have a end point but this time you use the wrist and the place where you send it is the main movement , where the end point of movement should be, the direction to send it and the velocity you want it to move at. Now here's where I think I have a brilliant idea. Notice I said elsewhere you have 80 different actuators in the hands. Lots of stuff to keep up with BUT what if instead of individually controlling all of them you instead gave each finger tip a point and then the controllers would work out Bézier curves to get them there. So you are only sending finger tip data. So say waifu is reaching for a glass. You send the wrist data and then finger tip. As it gets closer it's finger tips have already moved to the proper location to grasp the glass and as the wrist slows down you can add finer control to the finger tips. The finger tips of course control where the fingers move without needing to send data to every single actuator in the hand. Finger tips control the fingers. I wonder if Boston Dynamics is doing this. Their robots look good but still look a little fake to me. (I'm not dissing them. I'm sure they way smarter than me but I do sense a bit of fakeness in their robots compared to humans) Maybe they are using this same sort of control with Bézier curves and doing this you get a constant velocity as they are moving. Maybe real humans have a more varied velocity as they move their limbs. I of course could be totally wrong it's just a thought. I think this will work. You should see me flailing my arms around the room like a fool trying to see exactly what my body does. Most of the time we pay no attention to this at all. I was thinking about stuff like catching a ball or catching a dish you dropped. You would move the hand and arm in a direct path quickly. But if you were picking up delicate things you could slow the arms down and feed more bandwidth to the fingers. The goal is to never overload the network(nervous system). It may be possible that using Bézier curves will give us a much smoother lifelike movement without delving into AI corrected movement. Though I bet if you used AI for movement it would turn out better.

>>22111 Let me see if I understand your concept Grommet. Are you saying that the core decides the next stepwise action for a (limb, say), calculates a Bspline to direct it to the FK target and sends only that signal data across the network?

>>22113 Sort of but not quite. The peripheral controllers can only get directions from the brain and then calculate how to move there, excepting if they hit something as you and others have discussed. That being said I'm working with rough idea here so...could be changes. On the macro level this sort of framework could start out with a rough movement not so good and then as the waifu moves it could send correcting ending position, direction and velocity vectors as it moved. This gives you quick movement but allows finer control as it preforms whatever movement required. Gives it fast compute of gross movement but also good control because it gives it time to "think" over the movement period. It also allows us to change our programming or functions to get better as we learn what works while providing a fixed framework to start with. All the micro-controllers and end actuator controls have a fixed set of instructions which can be fine tuned by the main processor, while only using a small set of instructions. Rough progress then further refinement as you learn and upgrade. I made up the framework but got the idea of the refinement progress as the waifu moved from a paper where they used it with success and also a sort of general thinking about wavelets for processing. Because wavelets, I think generally, are a sort of a signal processing way of adding tiny more specific waveforms to larger signal gross waveforms to represent complicated signals. And they tend to be fast. I'm not saying I know how to program all of this or even yet how to set up the set of equations but I do know about some of these things generally and know that this sort of computing has worked on other task. So I would be a long ways off from making this happen but it can't hurt to start with what works. On the big picture level the main brain sees the world for the waifu to move in and tells the peripherals how to move and where to go. They don't understand the world only how to move their one piece of it. On the other hand by giving a complex instruction to the peripherals instead of minute by minute control from the brain you compress the network bandwidth which will likely be low with these cheap parts we will use. (1MB/s with CAN bus 2.0) I really like CAN bus because it seems most all the problems we need to solve have been solved for this and it's everywhere for all sorts of stuff. Further thought, you could have a separate CAN bus for each limb, and the head/backbone(5) or maybe one for the head alone for facial movements(6). https://www.autopi.io/blog/can-bus-explained/ Now I'm saying this several different ways on purpose. It's long but maybe it will be clearer. So I read what you said again,"Are you saying that the core decides the next stepwise action for a (limb, say), calculates a Bspline to direct it to the FK target and sends only that signal data across the network?" and realize that I haven't answered it correctly. The answer is no. The bspline is only for the fingertips and toe tips. All the other commands are as I stated earlier. >>21602 >>22111 The wrist and the foot are the point of reference for moving the limbs. The brain gives a go to point(end point), a direction vector and velocity vector, (in that order I would think. This knowing where it's going first might later could be used in running or using past recent movement to guess the other components and get moving right away), to the feet and wrist. The toes and fingertips(important only the tips data is important) get a five pointed bspline. Since the bspline represents a line then fingers can then compute with reference to the wrist where the finger tips go, which then computes how the fingers move to get the the fingertips where they need to be. Same with the toes which will determine balance in the feet by moving the toes. Most of the time only these signals are sent. For refinement all of the limbs could be sent the same data as they are moving or they could be sent the "stop" data. This would give them the velocity vector for stopping at the point they are going to. So data to decelerate the limbs before they get to the end point. Notice by using this broad framework you can first get a gross, get her going, program and then slowly add refinements. The same can be used for bandwidth considerations. If the bandwidth is pressed you still have an endpoint so maybe it's not so graceful as adding all the refinements as it moves the limbs but it would get it where it needs to be. Slower movement could use more bandwidth to give refinement. I would have a override for the joints that move the end point feet and wrist. I bet, after flailing my arms around and trying to make sense of my movements, that 99% of the time there's a basic straightforward movement based on the least distance to move but if there are obstacles the waifu brain could send vectors to the individual joints to avoid hitting things. In this case it's likely the waifu will be in a cluttered area so it will need to slow down anyways giving us more time, and bandwidth, per movement. One last thing. You need a reference. I would use the shoulders, the hips and the lower neck as references. When the brain sends commands it would be in reference to these points. You don't want a constant stream of (I'm here) of every joint. This allows the waifu a rough guide of the limbs for most movement but it could also see and calculate where the limbs are if it needs to be precise. Since the waifu knows where the neck, shoulders and hips are and all the limbs do also then it saves all this position data beng sent.. yet the waifu could use sight to monitor and refine this if needed. It's all about trying to do rough good enough movement when it's busy while allowing super fine control when things slow down and need to be precise.

>>22119 Thinking a little more, maybe a bspline is not the most efficient. There are other line creations functions. https://en.wikipedia.org/wiki/Non-uniform_rational_B-spline Maybe the lowest bandwidth set of commands is to do the wrist and foot as I said and then send a set of fingertip and toe tip position, velocity etc for each fingertip and toetip. I do believe that the wrist(actually the point where the forearm meets the wrist) and the foot(actually the lower leg meets the foot) points combined with toetip and fingertip, referenced to the before mentioned point wrist and foot points, will give you low bandwidth but precise control. Move your hand around and hold your forearm. You will see if you tell where the fingertips are then the rest of the wrist and fingers will move to make the tips go where you want. I actually should start saying the forearm end and lower leg end. Wrist and foot are not actually correct for gross limb movement. I hope I didn't confuse anyone because the nomenclature I used was incorrect. I think I'll start using the phrase leg end and arm end points. So the foot connected to the leg end and the wrist connected to the arm end.

>>22120 I better say this again to be super specific. Most will understand right away so bear with me because some won't. I made a mistake and was not precise. When you say I'm moving my foot or hand it can have two meanings. One you are moving the whole foot which actually means the lower leg socket point, but you also could mean "articulating the foot", which is fact means moving the toe tips. So if I said move the foot I really meant the lower leg. If I'm talking about "articulating" the foot I mean the toe tips. Same for the hand which means the end point of the arm meeting the wrist.

CAN bus ESP32 125KBPS - (Default) should be enough for network if you divide each limb and head into different busses. "...The ESP32 has an integrated CAN controller and therefore doesn’t need an external controller necessarily. You only need to specify the RX and TX pins. Any GPIO will work..." https://esphome.io/components/canbus.html

This guy here skipped the cycloidal drive where I'm still kinda stuck, and used a planetary gear with helical gears. The 8308 motor he's using costs around 85€ with shipping. https://www.youtube.com/watch?v=gxXatMr0LX4 https://www.aaedmusa.com/projects/openqdd Here some comparisons (shipping included), higher type number is of course (I hope) a bigger motor: 5010, 90kv, 30€: https://www.aliexpress.com/item/1005002552665265.html 5010, 360kv, 17€: https://www.aliexpress.com/item/1005002337180816.html 5010, 340kv, 40€: https://www.aliexpress.com/item/1005005134738612.html ... 8308s, MIAT-M8308s, 90kv, 171€: https://www.aliexpress.com/item/1005003780936078.html 8308, 90kv, 85€: https://www.aliexpress.com/item/4000563826414.html ... 8318, 120kv, 57€: https://www.aliexpress.com/item/1005005728960661.html 8318. 100kv, 42€-100€, broken to working: https://www.aliexpress.com/item/1005005081311825.html 8318, 120kv; 86€: https://www.aliexpress.com/item/1005003217445525.html ... 14970, stator core alone, 437€: https://www.aliexpress.com/item/1005004529938888.html Related: >>17201 - and thanks again.

>>23490 Nice set of links NoidoDev!

okay torque is the amount of weight of a motor is supposed to endure right? What will be the dimensions of the stainless steel bar for say the forearm, how much will the exoskeleton for the forearm weight? Will that include the motor itself in its weight? According to my guesstimates the entire arm should weight say 7lbs, which would mean the motor should have a torque of 7lbs for the base of the arm. Am I off the mark?

>>23719 Well no the arm probably wont have an inner skeleton come to think of it probably... 5 lbs?

>>23719 >okay torque is the amount of weight of a motor is supposed to endure right? Actually, it's more about the turning force needed on, say, an actuator shaft to rotate a robowaifu's limb for example. But don't take my word on it, we have actual MEs here. I'm just a lowly programmer lol. :^) https://www.khanacademy.org/science/physics/torque-angular-momentum/torque-tutorial/a/torque https://www.wolframalpha.com/examples/science-and-technology/physics/mechanics

>>23725 >But don't take my word on it, we have actual MEs here They're awfully quiet.

>>23729 Please don't go off-topic again friend. Do you understand what torque is and how to calculate/use it?

>>23731 How is that off topic?

>>23740 I don't want to have to do everything but its looking that way if people are not going to their fair share. Torque= rFsin(θ) r is the distance of the arm which if the robot is 5 feet is 2.5 feet or so. Force is mass times acceleration. the mass is 7 lbs assuming. We want the arm to move 45 degrees. 2.5*7*acceleration*sin(45)=the torque of the stepper motor. I already said I'll do this thing with or without you guys.

>>23750 thats just the force its just rf really, you still need to calculate the objects resistance to rotation (moment of inertia) to figure out how much force it actually takes to rotate it 45 degrees

>>23766 Thanks for the information. But I did say I'd leave so I have to leave now I mean. But last post unless there's some sort of desicion making mechanism to this project...

>>23750 I posted something about how to best calculate things before. You can use Waifusearch to find topics on the board. I also recommend asking ChatGPT nowadays for general advice. >>23750 Thanks. Good to know. >>23767 There is not one project. The group that exists is lead by Kiwi, I think, and they have their Discord and Matrix. You want some kind of organisation but you also made the decision that "your" project will be patented, so you put yourself in charge already. This site here is like a tavern where we exchange ideas and information. The part where we keep track of things and cooperate will always lean towards open source and sharing knowledge.

Going to try this motor. I don't like having to use gear motors. The price to performance and ease of integration into a design given its integrated driver and encoder makes it too good compared to DIY solutions. >=== -rm link

>>23964 Lol, apparently you triggered a spam filter Kiwi. Mind re-upping the link a bit differently this time? :^)

>>23964 Metal gears really seam to be quiet, compared to 3D printed ones. They might interest me as well for some use cases at least. I assume, these ones are not really backdriveable, though?

I'm all the time finding stuff on this board that's really good that I missed. These links comment on a type of actuator to be used for working exoskeletons. >>719 >>720 Lots of good ideas there. This led to a link on exoskeletons(that no longer exist or is broken) and from there I found a link on Bionic parts for damaged humans. So I link the internet archive link on exoskeletons and the current link for human bionic parts. Both of these will be good for coming up with actuator ideas. https://web.archive.org/web/20220131052129/https://exoskeletonreport.com/ https://bionicsforeveryone.com/

>>13701 >Nylon Actuator Just wanted to mention that I've read something about this on Reddit recently and it was mentioned that these kind of muscles need to be annealed in an inert atmosphere. This was in the paper, but this paper didn't get shared much. The videos how to make these muscles never showed the complete process. (I'll try to get the paper at some point)

>Inspired by Waterbomb-Kresling origami, we developed three types of vacuum-driven soft cubic actuators with different origamic skins, featuring contraction, bending, and twisting-contraction motions. These various modules can be assembled in different ways to achieve a variety of applications including weight-lifting / rotational end-effectors, simple 2-finger gripper, multi-module gripper, grape picker, crawler robot and pipe-climbing robot. https://www.youtube.com/watch?v=TLbEILTPcGM https://www.liebertpub.com/doi/10.1089/soro.2022.0064 https://sci-hub.st/https://doi.org/10.1089/soro.2022.0064 (not yet) I thought about something like this, but with actual hinges in some places, to make it last longer. Irrelevant for their use case since they can replace the part after a time, but inside our waifus this has to work for a very long time.

>>24232 That's a really neat idea NoidoDev, thanks! :^)

>>24232 These are really impressive. The power is really high with these.

ABENICS - spherical drive ball joints with low gear lash. https://www.youtube.com/watch?v=Z92dAnhv3JY

>>24329 That thing is incredibly cool but don't see all those gears lasting long. I could be wrong. It is very compact and solves a LOT of problems all in one neat package. Not directly about the above gear but gears in general. Seeing the above made me remember some papers and books I had on my drive so I'll link them while I remember it. I save all sorts of stuff on gears and machines, etc. Here's a paper on Cycloidal gears. They are used in watches because they have little friction The beauty of these is that while hard to machine they are much easier to make with 3D printers as they follow a math formula to make them. In the paper in also references to other books on these. https://www.csparks.com/watchmaking/CycloidalGears/index.jxl I have the ultimate book on universal joints and different ways to join rotating machinery. I'll see if it will upload. Couplings and Joints - Design, Selection and Application 2nd ed - J. Mancuso (Marcel Dekker, 1999) Too big. Here's an address to download it. http://libgen.rs/book/index.php?md5=FCD6B678C3CE84E7E117D80DFDB47043 This is really an extraordinary book on rotating powered joints. Covers all sorts of stuff. Also a history of movable, drive-able joints of sorts. It also specializes in flexible joints which is just what we need.

>>24328 Thanks, but this was already in one of the older threads. The questions are: How to make these? Are these for sale? How expensive? How loud? Will they break after a while? >>24353 >Cycloidal drive I gave up on translating the formula for it a while ago. I'm just not good enough at math. Also, there are so many ways to make them and some need very many parts of metal rods. I have it high on my list to try again, because I need them. But I'm kinda unmotivated to try.

>>24353 >Couplings and Joints - Design, Selection and Application 2nd ed - J. Mancuso (Marcel Dekker, 1999) >in memory of my daughter, my father, my mother, my wife, the rest of my children, and my grandchildren Jesus, what the fuck happened to this man's family???

>>24328 This mechanism has been posted before and I will reiterate that this mechanism is far too complex to be viable for anything. I understand how it can seem exciting, but realize that it's over 2 years old and hasn't been used beyond demos. Pan tilt mechanisms are the default answer for multi axis joints for a reason. They are simple, easy to build, have simple kinematic equations, and can easily be incorporated into designs. https://www.learnrobotics.org/blog/the-complete-guide-to-program-pan-tilt-servos/

>>24355 Eccentric cycloid drives are much simpler and better than the older stile. They're essentially cycloids that are inside out. https://learn.colorfabb.com/eccentrically-cycloidal-drive-model/ https://m.youtube.com/watch?v=qMDU5tlGUwU&pp=ygUZZWNjZW50cmljIGN5Y2xvaWRhbCBkcml2ZQ%3D%3D

>>24357 Let's hope he didn't mean it how you're interpreting it. He's just honoring his family. >>24359 Yes, still need to get the formula into OpenScad though. I had some ideas a while ago, about trying how to do this, but didn't so far. I want to do decide on my own gear ratio. Also, we need the other ones as well, because size. These ones here are bigger, I think, and if that smaller part gets to small I'll need to make it out of metal to prevent it from breaking.

>>24359 how is this simpler than a bevel or worm where all you need is the same same tooth profile like any regular gear

>>24358 >>24359 >>24363 Why bother with gears when we have electro hydraulic soft muscle actuators and dielectric elastomers? Only problem i can see from using them is proper power distribution for different muscle groupings. >>24362 I certainly hope he didn't mean what i thought he meant. If so, then i feel sorry for him.

>>24359 >Eccentric cycloid drives That looks really cool and great idea. >>24363 >how is this simpler than a bevel or worm I think they have less friction. I was looking around at peristaltic pumps and found this. https://tim.cexx.org/?p=1282 It uses a planetary gear design as a bearing and he added a peristaltic pump into the design. He has links to the original planetary gear design. This looks really good. It's a good way to design bearings that can be 3D printed.

>>24368 I forgot to add. He prints it all n one piece. I think if you you printed the outer race split down the middle you could join it without sticking and twist the two outer races, then screw them in place, to take up slack in the bearing. Maybe split the inner gear also. You could likely get good clearances like this and not have them stick together.

>>24363 It has advantages and probably disadvantages over a regular cycloidal drive. If you compare it to other gears, then it's a completely different comparison. I don't remember much about every gear, but we need a lot of reduction in a small space and they need to be fast, quiet, and back-driveable.

>>24355 >Cycloidal drive I actually looked into my files instead of just thinking negatively and discovered that I most likely already have the necessary code e.g. from this videos which were surely posted before. One of them: https://youtu.be/FdTGeYKB5NM (Dan Fekete) I don't understand the code, but I can work with it, I think. The important part is that I can change the gear ratio and the file changes to adapt to it.

>>24380 I also just found this: https://www.youtube.com/shorts/xq6ngavxCS0

Nice recent contributions ITT anons. Thanks! :^)

This is an idea going into a direction some of us certainly had. Trying to reduce the necessary number of actuators: https://hackaday.io/project/171924-braker-one-robot >Braker One is a proof of concept 3D printed robot based on a novel inexpensive DIY actuator system. Similar to hydraulics, it distributes power from a single power source such as an engine (gas, steam, sterling, turbine) or a motor (AC / DC / brushless / stepper) to multiple actuators. It does so through a series of interconnected miter gears located in every joint. ALL gears spin continuously delivering power to all joints. Two miter gears in each joint that spin in the opposite directions have disk brakes attached to them. Use of MR fluid brakes is also possible. >When a joint needs to rotate, a caliper or an Eddy Current brake located on a limb acts on the continuously rotating disk and by doing so rotates the joint. Instead of converting energy into heat however, brakes convert spinning disk's energy into limb movements. When a brake is not applied, the joint is fully COMPLIANT and rotates freely! Brakes are smaller than motors, providing higher power to weight ratio. Similar topic, using clutches like bicycle breaks: https://youtu.be/2-eAZlVFlkM[Embed] >This video introduces the Distributed Active-Semi Active (DASA) actuation approach. The DASA actuator was invented by Alex S. Shafer at the University of Western Ontario under the supervision and direction of Associate Professor M.R. Kermani. The actuation systems is developed for low-cost, high-performance, human-safe robotics. To find patents and publications on DASA, you can visit: https://scholar.google.ca/citations?user=nJ9rSQwAAAAJ&hl=en - I downloaded one paper for now an attached it. Found via "Ferrofluid for simulating lifelike muscle movements in animatronics?" https://www.reddit.com/r/robotics/comments/15q89n8/would_it_be_practical_to_use_ferrofluid_for/

>>24620 This is pretty interesting stuff Anon. >It does so through a series of interconnected miter gears located in every joint. I'm curious if he's going to attempt any type of 'pass through' power transmission system -- say through the shoulder down to the elbow for actuation; or through the shoulder, through the elbow, down to the wrist for actuation, etc., etc. This is our actual problemspace to solve ofc. Regardless, it's encouraging he seems to be working towards a humanoid robot of some kind. Therefore he's a kindred traveller on some of the same roads we journey together! Cheers. :^)

>>24620 >Braker One That's the same guy that invented the fiber optic sensor idea where you use fibers linked to one camera for sensing,(fibergrid). Pure genius. One of the best ideas I have seen in a long time. This idea...eeeeh...maybe. A bit difficult. He says it's way out of the box, and it is. I can't say it won't work and is not a good idea. I don't think it would be so good for waifu's. Too much noise but it is next level thinking. I hate to say anything bad about it because it's so original and approachable that in it's place it's really brilliant. You could do all of this with 3D printers.

>>17882 Update: the seller marked up the same product 50$ more than before, almost 2x the price. These are no longer worth buying vs smaller outrunner motors in small qty. I may have to find a way to wind my own motors or buy a new alternative on alibaba straight from the supplier. Every robot component from china just got marked way up this year, and I think I know why, but I'll keep that conversation in a different thread. I have one of each of almost every kind of motor, and I'll make a spreadsheet of weight, price, power, ect. Before buying any more motors I'll need to do a robot motion simulation to find what will work and where.

Throwing out a "maybe" wild ass idea. What if you used "roots" type cycloid blowers for actuators? Normally the blower sucks in air and compresses it. It seems that you could force in fluid or air and run it backwards. Where the fluid rotates the lobes and then the gear, normally driven by the engine in operation, is now driven by the blowers. This occurred to me looking at these things after seeing cycloid gears printed on 3D printers. Since the blower lobes are large it could be that a low pressure fluid over a large volume would allow you to use low cost plastics for the lobes. Might have to have metal for the output. The output could be a pulley with a string on it that winds up like a muscle being pulled. https://www.tec-science.com/mechanical-power-transmission/cycloidal-gear/roots-type-blower-supercharger/ https://en.wikipedia.org/wiki/Roots-type_supercharger There's also screw type blowers that are even more efficient but could be used the same way. This idea may not be totally whacked. I the actual hordepower used by a blower, meaning HP used to blow could be also gotten out of the blower run backwards. I saw these quotes on how more efficient screw type blowers are, "...“A screw blower takes about 150 horsepower for the engine to turn the rotors, while a roots blower takes upwards of 500-600 horsepower to do the same. ..." “The screw blower requires less energy to do the same amount of work compared to the roots blower. Given the same 40 psi of manifold pressure, the screw blowers require on average 400 less horsepower to generate this amount of boost..." WOW! this screw blower uses 400 less HP! So how much power does this thing put out. Must be really high. Have you seen how small these are on a dragster? There's not that big and they are using, and could put out, more than 400HP. https://www.dragzine.com/tech-stories/blower-talk-roots-and-screw-blowers-in-drag-racing/ So let's just add some quick and dirty numbers to this. Let's say it can make 150HP like the screw blower. We need about 1hp or 746 watts of power. So the size would be 150 times smaller and even if it's half as efficient as the real dragster type it's still 75 times smaller. A normal roots blower is 15" long. The 1HP version would be, at 75 times smaller, 0.2" long. Of course you realize this is very rough numbers but it does give you some idea of what can be done at the extremes.

BTW these dragsters due to these blowers put out somewhere around 11,000HP!!!! https://www.popularmechanics.com/cars/a39954318/top-fuel-drag-racing-horsepower/

>>24843 >>24844 Neat. I love racing Anon. Dunno about the basic concept though, tbh. A) Seems like there will be a significant mechanical-disadvantage trying to run these compressors in reverse, using just low-pressure gas (presumably air). B) It's going to make a bodacious loud racket. These blowers are LOUD, b/c the kind of work they're doing. Cool idea though.

>>24853 >using just low-pressure gas (presumably air). I'm thinking water.

Why do motors always have these skewed screw holes? This threw me of my design, because I already had this solved with four holes at a same distance. Then I had to redesign it, which is just so annoying. I wonder if there's a better way to deal with that. Maybe I will make two parallel gaps instead of holes.

>>24879 So this is normally used for compressing fuel in a combustion engine? Or for the hot gas? And you think it might help with compressing water for hydraulic actuators?

>>24883 >So this is normally used for compressing fuel in a combustion engine? This is used in dragsters/funny cars to force air into the engine. According to links I found they inject some fuel before the blower to keep the blower cool and more after the blower. They don;t want too much fuel before the blower because it takes room from the air. They use nitro-methane for fuel because it's high vaporization temperature sucks some of the heat out of the compressed air, as it vaporizes, and cools it. They need this because they are getting 11,000 HP out of them as I said. A stupefying amount of HP that astounds me. The basics are engines need air and fuel. The supercharger-blower compresses and packs air in the cylinders. So the real reason for the blower is packing in the air to react with the fuel. The blower type you showed is a roots. Roots blowers as you can see are a bit pulsed. They make pulses. See the picture I loaded. The lobes on the roots compress the air in batches. The next one, screw type, it's my understanding that it does not necessarily compress the air but like a normal air fan speeds it up which is then turned into pressure so less pulsing. Vastly raising efficiency. The pulses make heat and waste energy by turning it into heat instead of flow. Flows what we want. Not compression or heat. So running this backwards. Feeding it, I say water instead of air, would turn the flow into rotation. Where the belt from the motor would normally turn the blower, instead the flow would turn the belt, or pulley. Note I did say this was a wild idea. The reason I mentioned it was the pressure these things work at, 40psi in dragsters, is not really that high but they are boosting a tremendous amount of power from that small amount of pressure. A general idea about pumps and motors is you can run them backwards in many cases. A motors a pump run backwards. So if they can get so much boost out of them we should be able to get a great deal of power at relatively low pressure. The low pressure is important because standard 3D printing material is not that strong compared to metals. It appears to me that the very long length of the rotors means a small pressure acting over this big long area gives it a lot of force. I fully admit I may be wrong about this. What impressed me is these, not so big, screw type superchargers are using 120 HP or so from the engine and I read the screw type compressors have efficiencies approaching 98%. Now of course this is for air compressors for large businesses like heavy equipment shops where they use tons of compressed air. Likely we will not get 98%. I'm not stating 98% as a fact we will get but I am saying that they appear to be very "inherently" efficient and they appear to be something that gets a great deal of power from a small space AND they appear to be something that is complicated to make normally, but NOT with 3D printing. Look at the VERY rough calculation I did. "...A normal roots blower is 15" long. The 1HP version would be, at 75 times smaller, 0.2" long..." Now this is super rough but even at 5 times bigger that's 1" for 1 HP or 746 watts which all the power we need for a really robust robowaifu. I admit it';s a wild idea but I looked at the numbers and it doesn't look exactly wild when you do them. An of course I could be complexity missing something...or not. It's something to think about. An outer possibility.

I see a glitch in the screw supercharger idea. "...For smaller rotor sizes the inherent leakage in the rotors becomes much more significant, leading to this type of mechanism being less suitable for smaller compressors than piston compressors..." but...using water would lesson this a great deal so...who knows. Maybe, maybe not.

I don't know why you guys are declaring victory with the physical robot. Poppy is like $10,000 and imoov doesn't have legs although it's far more reasonably priced. You guys won't even spend $100 we have to bring the cost down. I was thinking of using bevel gears for the legs. Anybody has better ideas that don't involve expensive actuators?

>>24937 I don't think this is a problem but we will have to make our own actuators. Look up at other posts I did in this thread and the first actuator thread. The way forward is magnetic reluctance motors and combinations of these possibly with hydraulics using water. The power is there. It can be done but you can't buy this stuff off the shelf. The good thing is it can be built. I've provided a great of data on this. Musk is getting stupendous power out of these type motors and you DO NOT need rare earth mineral magnets. Not at all. I've been steadily researching this stuff and making notes and doing cost accounting by looking at the parts cost. They are very reasonable. I found a guy who explained the best way to drive MOSFETS. So as far as I'm concerned it's just a matter of putting it all together. The math works. This will take a good while. I expect it will take me 5-6 months or more to figure this out. I need a 3D printer and it will take a good while to learn it plus all the software. Of course I plan 5 months, it will probably take twice as long, or more, as these things always do. The best way to drive MOSFETs. You use a set of AND or NAND gates from the 74___ series like, 74HC08 logic gates 74HCT logic gate The reason these are used is MOSFETs can not be directly driven by micro-controllers. These take a small signal from a micro-controller and peg the output to a higher voltage. MOSFETs have a value called GSr (gate to source resistance) which is the total resistance when it is turned on. Some are extremely low, and need to be, have to be to keep them from burning up, BUT you must drive the voltage high to get that. So you use the 74___ series to do so. So Mosfets are $1 for really good ones and $0.50 and below for probably good enough. The gates are 74HC08 logic gates, 4 gates in each, so,$10 for 10, drives 40 Mosfets. This is all you need to drive fairly high current devices. Bevel gears won't help I don't think. You have to drive them with something so now you are right back to some sort of electrical motor.

>>24944 That'd be great thank you. On the mean time I'm looking into controlling servo motors with a raspberry pi zero and might go back to the speech recognition ai cause as it is it is too slow regardless of it being on the cloud. I should have just bought a raspberry pi 4 or something with the money I've spent but oh well. I think this will do. So yeah I'll do the hand, a barebone talking to the waifu, might look into sophiedevs head and try to make molds for the skin with 3d printer and maybe clay to save money. As you can see that in itself is a bit of a handful Id appreciate if people can work on other stuff as well or improve upon it. You know work as a team. Out the stuff I didn't mention are head movements, body movements, the leg, the actuators, poppy's exoskeleton is kind of skinny too I don't know if thevexoskeleton needs to be modified...

>>24672 https://files.catbox.moe/umd93d.mp4 Rebuilt my motor with the thinner fl106-06 bldc in it, and now it's only making normal cogging noises when it turns. Looks promising so far, and I'd like to make a forged CF motor casing instead of plastic. I'll follow the Easy Composites tutorial on YT with using a PETG clamshell casting + draft angles + release spray. I might put in a nylon gasket between the steel and CF so it won't corrode or crack.

>>24885 I've been considering hydraulics later on in my own design, and ended up reading a bit as a refresher-- might be worth giving one of those books a look-through (the one I read was "Hydraulics and Pneumatics: A Technician's and Engineer's Guide"). Don't even have to read through it proper, just scrolling through and reading the diagrams you find interesting gets the job done. Anyway, it lists the different types of hydraulics pumps (gear-lobe and rotary-screw are the ones you mentioned) and references efficiency and impulse tables. afaik, lobe pumps are considered notably more efficient than screw pumps-- maybe this comes down to the differences between air and "incompressible" fluids? At the very least, a lot of the concepts that work well for superchargers center around compression and its effects (Venturi, etc) to change gas velocity and such. There are also efficiency, torque, and scale (sizing) tables for hydraulic actuators, like the rotary actuators you mentioned. Sadly, I haven't seen much info on actuators that could be waifu-sized, but it seems like the field is advancing pretty quickly-- "biomechatronics" and all, now that we have semi-functional neural interfaces for prosthetics, is getting a ton of commercial attention. Not that we're stuck with commercial products ofc, just... I've found it helpful to let the big companies and their R&D budgets work through the annoying stuff for me :P

>>24968 >lobe pumps are considered notably more efficient than screw pumps-- maybe this comes down to the differences between air and "incompressible" fluids? Hmmm...interesting. I didn't know that. Have think about it. I know the lobe pump inwith air is causing heat build up from the lobes alternating pressure on the air. I guess fluids being mostly incompressible do not have that problem. My tractor/backhoe/loader has a constant pressure pump system. It has bunch of little piston pumps in a ring and their stroke is based on pressure needs so when no pressure needed, no stroke. I can't remember all the details. I actually rebuilt it. Real pain in the ass. I suspect the simplistic system is a peristaltic pump with an elastic reservoir that is electricity driven. Not saying this is the best but it would be cheap and easy to build.

>>24968 My understanding how we would use pneumatics or hydraulics in waifus is the following: Soft muscles that contract or expand, and a pressurized tank in the waifu, which is pressurized during the time she is plugged in. NOT using a loud pump, but a slow but strong motor using a foot pump. Related: >>19601

>>24984 Here's a quick paper on the subject, with picrel as an excerpt [doi: 10.1016/j.epsr.2017.08.003] Guess the whole push for green energy can be useful after all lmao You probably know more about this than me, but I've heard that certain types of piston pumps use a series of chambers with different compression cycles, trying to flatten the impulse as much as possible-- sounds like that's what your tractor has? btw, topic unrelated, but I might need to pick your brain on keyshafts+key installation down the road-- my coworkers are great for engineering stuff, but we're kinda spoiled when it comes to equipment and doing stuff the "right" (expensive) way, since it's on company dime. I'm guessing you've installed em before? >>24986 Not sure how practical that is tbh-- at least for hydraulics, it's a non-starter, since we're talking less than 0.5% volume changes per 1000psi. Pneumatics is a bit more doable, but without a massive tank, I just don't think we'll be able to store sufficient volume, passively, unless we do something crazy like putting it through a state change and storing it as a liquid. ... Actually, that could be worth looking into. Not gasses that need cryogenic temperatures, mind you, but if we could find a liquid that phase changes near enough room temperature, that ~1:1000 expansion ratio is looking sexy as hell. Anyway, a rule of thumb for hydraulics is that the system's pressure is effectively the pressure exerted by the system's load, You can get away with stuff like a buffer (think reverse retard chamber to "store" force), but that usually kills efficiency unless you have some very fancy valves that disengage the buffer unless the system is drawing from it, or idle (and thus "charging" it)

>>24990 >You probably know more about this than me Apparently not because that chart surprised me. Why is a screw type pump so low in efficiency? I don't know. It's not that way for air. You would think a steady flow of a screw type pump would be far more efficient than a pulsating piston. Even regular gear pumps they say are better. I wonder. I can't find that paper from that number. There's mass of links. Whats the actual name of the paper? The reason I said use peristaltic pumps is while they don't have the best of anything they are fairly quite and when you stop them they hold pressure. They do pulse but with more rollers this becomes less and you will need some sort of air spring to had a reservoir for pressure. I think they would be cheap and good enough. But maybe not. >keyshafts+key installation...I'm guessing you've installed em before? Yes but I don't design them just use what came with it. If I remember correctly, and I may not, electric motors generally have straight shafts and pump type stuff has beveled shafts to press it on tight. >we're talking less than 0.5% volume changes per 1000psi Ehhhh...I don't necessary follow that but I'm surprised about the pump efficiency also. The storage is only to get the pump started. Instantaneous. It may well be that if you are right we will have to go all electric. My ideas about hydraulics has to do with low cost manufacturing. Not any love for hydraulics. In fact I hate them and have said so earlier. I rebuilt the pump on my backhoe/tractor, that wasn't the problem. Turns out there's a little reed valve spring type thing in the power steering column that when you twist the steering wheel inside the steering column this little spring loaded valve would let more fluid into the power steering. A seal in that drained my pressure. You have no idea how long it took to figure that out. I hate hydraulics but...if that makes it cheaper then, hydraulics it is. Look up this article. You don't need as high a pressure as you think. "Fluid-driven origami-inspired artificial muscles" A good way to visualize this is an accordion or a scissors jack. Low pressure, large area.

Another interesting pump that can also be used as a motor. https://en.wikipedia.org/wiki/Progressing_cavity_pump

>>24990 It's important to note that I wouldn't claim that it makes sense to use that for everything. I just want to keep it on the table for at least smaller movements or as additional actuators. Keep in mind: I want to build a gynoid with a soft body. Should I fill the soft parts with some filler, or rather use it for some soft muscle?

Mirroring>>25099

>>25114 Expanded I want to use wipers or similar dc motors with gearboxes over servos due to them being powerfull, having 90 degree gearbox (tho large gearboxes should be avoided doe to mass added) which allows for better space management & power transmition (belts/steel cables/chains) where large and blocky steppers/normal servos that lead to compromises of power over fembot form. I was looking up how other people make custom servos outta salvaged motors and they either rip control boards from cheap/broken servos or equip fairly inexpensive assemblies with crazy expensive encoders and drivers. Have any one of you build your own custom piece that can be hooked up to SBC?

>>25179 cont. Basically, if i get it right, it should be as easy as hooking encoder (soemthing dirt cheap as 10k linear potenciometr) to analog input on SBC, from there send PWM signal to driver that can control speed and direction )again something dirt cheap as L298N). Connect it to your DC, afix encoder, test & set movement range and it's done. Anything major imma missing?

>>25040 https://doi.org/10.1016/j.epsr.2017.08.003 "Direct driven hydraulic drive for new powertrain topologies for non-road mobile machinery" To be clear, I'm not an expert on hydraulics (far from it, in fact-- haven't seen em since college), so a lot of the questions you're asking I'd have to research in order to answer. Plus, I'm pretty sure the "why is it inefficient?" question would be worth a published paper or fifty.... If I were to "shoot from the hip" to answer (which means I'm 99% wrong), I'd say the different efficiency cases comes down to the properties of the fluid (or gas) that you're moving, especially viscosity and compressibility-- imagine trying to pump honey with a "propeller", vs. a screw pump (which I imagine would work quite well) Intuitively (so now I'm 99.9% wrong), if we think of a piston pump, the only real "wasted motion" is retracting the piston and refilling the void it's about to pump. Valves would cut the piston of from the system any time it's not contributing energy, and thermal losses from momentary compression aren't a big issue for stuff with such low compressibility. That's just my 2c tho, and it's worth noting that the paper I referenced is specifically for small-scale, direct-driven pumps with commercial hydraulic fluid. > (re: driveshafts) Thanks, man. I’ll need to come up with a mechanism that allows a keyed gear to slide along the keyshaft’s axis, while not allowing rotation. Thankfully, a bit of slop (backlash) is tolerable, but if you have any ideas to minimize it, let me know. (rel >>24744) > (re: hydraulic volume change under pressure) I think we had a bit of a miscommunication there, I was referring to NoidoDev's idea of a passive, compressed storage system for active operation. The compressed-gas hydraulic bladders you mentioned will work, of course, but in that case, it's not the water "storing" that pressure, just transmitting it. Bottom line with "incompressible" stuff is you'll never get more volume out of it than the tank displaced in the first case, no matter how much pressure the tank has (both by-definition and purely theoretical, I know) As I understand it, the easiest way to understand how much energy there is "in" a compressed system is to think of the ratio between compressed and uncompressed densities. (multiplied by the force exerted at each particular pressure, of course) For example, let's take two pistons exerting (the equivalent of) 1000lbs on a 1in plunger (say 1000psi inside), one full of compressed water, the other with compressed air, and say they're each at "100% volume" (not yet extended). Now, as they extend, the volume the compressed medium fills will increase, so the pressure exerted by the piston will decrease. Once the piston full of compressed air extends to 101,103, 105...% volume, the force it exerts (and internal pressure) will have decreased a little, but there's still plenty of Work (physics-def.) it can do. The piston full of water, on the other hand, no longer exerts any force once it reaches 100.2% volume (since water's compressibility is ~0.2% per 1000 psi), and the internal pressure (ignore gravity for a bit) is now ambient, so "zero". In this example, the density ratio of water was 1.02 @ 1000psi, but air has a density ratio over 69.1 at the same pressure (see https://www.engineeringtoolbox.com/air-temperature-pressure-density-d_771.html -- I was incorrect with the 5% factor, that was the compression ratio at the bottom of the ocean lol >>24990 ) (Formal explanation of it: https://www.sciencedirect.com/topics/earth-and-planetary-sciences/compressibility)

>>24882 Sorry man, I missed this first read-through. Staggered mounting holes/"Staggered Connections" are intended to maximize the Fracture Path length (or, for the forces we're working with, it's more "deformation length", especially with materials like thin aluminum) while minimizing Sheer Lag. I know, that means jack shit unless you're already an ME (and I'm not), and I'll spare you the textbook copypasta, so have a couple diagrams that do a pretty good job explaining it intuitively. Basically, maximize the amount of material that has to be sheered/displaced for your mount ("connection") to fail (Fracture Length), and minimize the amount of material that isn't bearing any force (Sheer Lag). >>25190 That link shouldn't end with a ")", oops. real link is: https://www.sciencedirect.com/topics/earth-and-planetary-sciences/compressibility)

>>24990 Thanks for the paper. So I look at this thing and I admit I'm fairly suspicious about this efficiency chart that shows screw pumps as next to useless. So I'm looking at all sorts of web pages on pump efficiency and a lot of them talk about "volumetric efficiency" as being a key performance figure. Look at the chart again and you see one side is efficiency and the other is "change in pressure". Now where that chart came from according to the paper is H. Kauranne, J. Kajaste, M. Vilenius, Hydraulitekniikan perusteet. [Fundamentals in Hydraulic Technology], WSOY, Helsinki, 2006 (in Finnish) now I can't read Finnish so...I'm going to have to guess. Screw pumps DO NOT have a high pressure rise like other pumps. They explicitly talk about staging them to get higher pressure. So I think this chart may be right but it's not measuring what I care about which is mostly actual efficiency as measured by power out based on power in. I think that chart in terms of efficiency is just how much pressure you can get from that type pump in a stage. Not saying it doesn't exist but it's hard to find a graph of total pump efficiency for all different types of pumps on one graph. I suspect it's because all these pumps have different uses and pressures so they feel that it's not correct to show them all under one condition. Now for what I see the paper saying. They are saying to get rid of valves, and yes normal hydraulic valves are totally expensive pieces of shit. They say drive a pump electrically and, I would have to guess, use a electric motor driving each hydraulic pump for each cylinder. For a big machine I expect you might save money like this. Traditional hydraulic valves are last time I checked about $100 USD per valve and have to be precision machined to high tolerances to keep all the fluid from leaking out. In fact mine right now are wore out. But on a waifu...I think this would be a lot of trouble. On second thought...maybe not. Let's say you have a bunch of printed gear type pumps run by switched reluctance motors AND you make a huge system of blood vessel type connections to all the muscles. So you start the motor and it pumps up the muscle.Maybe that would work. So I started out thinking this was horrible but, maybe it would work. Have to think about it. It's also possible some combination of direct pumps going to large muscles, but using one pump for many actuators with valves could be done for stuff like fingers and smaller muscles. Now because it's not really clear and we should think about these things. IF you are driving an electric motor for most every muscle, and you drive a pump for most every muscle, then wouldn't it be more efficient and cheaper just to drive the damn muscle with all electricity in the first place? I mean you now have all this wasted power going through pumps. (read the paper on all the losses you quoted) The only advantage I can see is maybe you could have a pressure accumulator, basically a solid tank, with bladder that has gas on one side and fluid to be used on the other. The gas is just a spring to accumulate power. For myself the only reason I would consider hydraulics is if I could make some sort of accumulator and could come up with a cheap valve. I know a pump can be made cheaply and be run by one or a few motors. Having large castible bladders for all the power systems where fluid runs through castible pipes and bladders could save a lot of time and money compared to building a lot of electric motors for each actuator. PS after looking around I do not think the graph above is the actual total efficiency of a screw pump. Screw pumps they say are very efficient and that they do not have the turbulence problems (turbulence causes inefficiencies) of other pumps. https://www.linquip.com/blog/parts-of-screw-pump/#Advantages_and_Disadvantages_of_Screw_Pumps

>>25190 >Bottom line with "incompressible" stuff This is probably of little use but...it's not true that fluids are ncompressable. Mostly, but not all. There's a guy who invented all kinds of stuff based on this and has a lot of damn near incomprehensible books and papers on this. Very interesting fellow though. https://en.wikipedia.org/wiki/George_Constantinescu http://www.rexresearch.com/constantinesco2/constantinesco.htm

>>25184 This link here on fiber optic sensors for encoders is likely one of the cheapest and most brilliant ideas ever. >>22032

A repeat from prototypes that belongs here. Low cost printable gear train for motors. I talked about George CONSTANTINESCO. The guy has one of the most brilliant transmissions ever invented and it would be very suitable for 3D printing. Here's two links on it. http://www.rexresearch.com/constantinesco2/constantinesco.htm http://www.rexresearch.com/constran/1constran.htm It's difficult to describe. It uses two long levers. Basically in parallel. They are mounted on sprags. A sprag is gears that allow you to pedal a bicycle then coast. Hear the ticking when you coast. That's the sprags rotating. When you pedal they lock up and force the wheel to move. So imagine two long bars that are mounted on a wheel like the petals and main gear on a bicycle. The size of the gear and in this case where the bar is grabbed determines the stroke of the bar. So if you want high power the drive bar is gripped low and the driven bar higher. So the drive bar moves a small distance pushing the driven bar forward but only a little. When it goes backwards the sprag lets it coast. So it drives a small amount on the long post output. Meaning lots of power. If the driven gear is farther up it will drive the output driven post/gear very far for each rotation, so higher speed output. It's really ingenious. The guy was a major genius. Since the force is over a broad area, it's just like a combustion engine crank and piston rod, you can have a sturdy system with light parts. And best of all it's continuously variable. So if the waifu wants to move fast, but with low force, then it can be adjusted so. If lots of force needed it can also do this but at a lower speed. Here's a random 3D printed sprag video. https://www.youtube.com/watch?v=UIhCPl8eb7s random link to files. I have no financial interest in this it's just a link for show. https://www.makersmuse.com/roller-clutch

>>25197 Looks promising, especially part where you use single camera to replace all your encoders, but for now it is out of scope of my project. I'd rather use some magnetic encoder that tells movement & position and gives me output i can use in normalcircuitry. Some thing like AS5600.

Under the category of brushless DC motors, what seems to have been a completely overlooked source of great robot parts: !!!HOVERBOARDS!!! My searches didn't turn up a single mention of them on this board. It's not just anons here, even the Masiro Project could have saved a lot of money if they built their robot's bases using old hoverboards. They were built to carry adults so they are way overbuilt for what they and we want to do. Each board comes with 2 hub motors, 2 gyros, the motor control motherboard and a battery. The motherboard can easily be hacked to flash it with new firmware. The motors can also be run by a hobby esc. Use controllers for cars that have the same speed in reverse as forward, and a braking function and you guys who like tachicomas are halfway home. So if you are like me and planning to run my waifu around on a base until she can walk you could do a lot worse than to investigate the use of hoverboards. Here are some links to a few of the hundreds of videos about hacking hoverboards for many uses including robots, and a couple of threads from the Synthiam site. (Disassembling) what's inside Hoverboard https://www.youtube.com/watch?v=xnCp-DJUCt8&t=15s Hoverboard parts explained for robotics use https://www.youtube.com/watch?v=-VIRlurVht4&t=63s Hoverboard Robotics Tutorial Part 1: Hardware Overview and Hacking https://www.youtube.com/watch?v=bIHjRTG7ilU hoverboard driver hack https://www.youtube.com/watch?v=IL4xcSQ6_rU Driving hoverboard wheel-motor with RC model ESC https://www.youtube.com/watch?v=LJzoy2TJXFk RoboDog Part 3 | Using Hoverboard Motors for RoboDog https://www.youtube.com/watch?v=_VUtfK4bCl8 https://synthiam.com/Community/Robots/Untitled-Hoverboard-Project-17760/comments https://synthiam.com/Community/Questions/Hoverboard-control-18833

>25783 Crap, I forgot the picrel.

>>25783 Interesting, but they guy disassembling it seems to use it for making a regular robodog, not a humanoid on a hoverboard. Also, recycling old devices isn't something that can be shared easily, and I don't have a "hooverboard" (which is a misleading name anyways).

>>25790 >Interesting, but they guy disassembling it seems to use it for making a regular robodog, not a humanoid on a hoverboard I included that one for the anons who are interested in ponies or 4-legged monster girls.

>>25783 Good resources, though huge, they make excellent drive motors as they provide tremendous torque. 36 to 48 volt systems are recommended to ensure a highly responsivr motor.These motors come in many sizes but, usually are around 6.5 inches with 16kV. For hacking, you want to use old hoverboards. I have a new hoverboard which uses a cheap Chinese chip that no one has figured out how to hack yet. >>25790 Please try to open your mind. Information from a tutorial can have benefits outside of explicitly replicating their work. In this case, the tutorial provides beneficial information on the general characteristics of hoverboard motors and controlling them. >>25791 >Quadraped monsta grills Only as the end drive motors.(picrel) These motors are not built to maintain high holding forces. Using them as the base for servos will create a very hot system. You'd need to implement a gearbox which would end up with a gigantic and heavy servo.

>>25792 > a cheap Chinese chip that no one has figured out how to hack yet That's good to know, I'll have to check my boards to see if any are the same. While it would be best to be able to use the controller that they came with, in the worst case they can still be used with RC gear. I hope mine are all good, all of my ESCs are for aircraft- one direction and no brakes.

These seem promising (if not a scam): https://www.alibaba.com/product-detail/Special-price-Limited-Quantity-New-RMD_1600938534322.html I took a chance on these cheap motors so you don't have to. Normal price for about the same motor is 200+$. They'll arrive in the mail by the end of the month, and they can be used for a limb actuator.

>>25910 Thanks Anon! :^)

>>25923 Yw, expect me to post a guide if I figure out how to make them work or dissapointment otherwise. They got held up in shipping because the US needed an importer information form.

>>25910 >>25956 8 motors came in, they all turn so at least it's not a complete scam? I'll test them whenever.

>>26063 Woot woot! Thanks for the good clear shots Anon. Any chance you could give us a side profile pic too?

>>26064 canbus v2 arduino hat on the left, motor is as large as a coffee cup. >=== -minor edit

>>26081 Great scale information Anon. Looking forward to seeing what all you accomplish with them! Cheers. :^)

>>26081 this one works as it is supposed to with the debugging tool on the website: >https://files.catbox.moe/8piz07.mp4 does this tickle the fancy for any of you robot enjoyers?

>>26107 This video is kinda strange. What's going on there? The camera moves away...

>>26113 I tried recording a 10 rpm test but flinched because the motor started sliding on the desk. Stuff like this is dangerous and motors have fallen and broken before when I didn't secure them.

>>26115 Okay, now it's clear, and also a good reminder to be more careful.

3/8 motors are making a clacking noise every couple of revolutions when they go very fast, but it's more or less quiet, so probably fine for my purposes. One of the motors is a bit more clicky than the others. As long as they all move the robot it should be fine? >https://files.catbox.moe/7qze8r.mp4

>>26118 yeah balancing the rotor is ultra high precision stuff the smallest difference of weight on one side will make it vibrate, its fine if you havent attached it to anything when it has to actually work against something its not going to be able to go that fast

>>24328 >Isotropic Meshing Gear https://www.youtu.be/AZQcHOJwhKU

>>12810 Where can you buy a actuators for robowaifu movement?

>>26124 You mean bldc motors, servos in general or specialized robot servos? There are websites for this, I'm just buying cheap electric motors on AliExpress.

>>26131 >https://www.alibaba.com/product-detail/Special-price-Limited-Quantity-New-RMD_1600938534322.html 6/8 of these aren't bad and they were only 100$

>>26123 Excellent, thanks Anon. I see a bright future for these as Elbow/Knee joints.

>>26135 >6/8 of these aren't bad What do you mean? Two of eight of those you bought had issues?

>>26155 Yeah like 3/8. I'm pretty sure it's a scam but it's still cheaper than buying normal price 200/300$ motors so I'm still making the correct move on these. Also they're kind of being nebulous on if they want me to return the bad ones or not.

> How Does a Fluid Coupler Work? I 3D Printed one that Uses Vegetable Oil! https://www.youtu.be/wGw9sXhYEc8 >Today, I bring you my 3D printed fluid coupler. Out of curiosity, I wanted to know if I could 3D print a working fluid coupler that works on the same principals as one you would find in a car. The fluid coupler uses vegetable oil as its fluid much like how a transmission uses transmission fluid. I explain how a fluid coupler works in case you are curious yourself. https://drive.google.com/drive/folders/1KF00u7qUKaTI0XaQ2U0TaewtBtsG-btX?usp=sharing Not sure if it's going to be useful and the current design made in plastic will have heat issues after a while. >ChatGPT: Fluid couplers and gears serve different functions and typically cannot directly replace one another in most mechanical systems. Gears are mechanical devices that transmit power by meshing with one another, transferring rotational motion and torque between shafts. They provide a fixed speed ratio between the input and output shafts. On the other hand, a fluid coupling is a type of hydrodynamic device used to transmit rotational power from one shaft to another. It works by using hydraulic fluid to transfer torque. A fluid coupling doesn't offer the same precise speed control or fixed speed ratios that gears provide. In some applications, a fluid coupling might be used in conjunction with gears in a transmission system, particularly in situations where you need a smoother start or if you want to limit shock loads when starting or stopping machinery. However, in terms of directly replacing a gear with a fluid coupling or vice versa, it's not a straightforward swap as their functionalities and modes of operation are fundamentally different.

>>26280 It seems to me that we'd benefit if we could actually use fluid coupling because of the good resillience/wear dynamics. The problem is that the high rotational speeds needed to get good high-viscoscity transcoupling effects is prohibitive for our needs, I think.

Interesting actuators breakdown analysis video : https://www.youtube.com/watch?v=5MNz0YyqgLY of the Teslabot actuators promotional video : https://www.youtube.com/watch?v=1xChD-gv_pc

>>26780 Thanks, this might be very useful. Maybe rather long-term but nevertheless. These actuators look like one's without neodymium magnets, which could be interesting to be independent from that kind of supply. Unrelated Cute: https://www.youtube.com/shorts/HSDNFXXOThE

>>26782 >These actuators look like one's without neodymium magnets, which could be interesting to be independent from that kind of supply. Yeah, I've been thinking about that exact issue for a while now NoidoDev. Obviously China intends to restrict their hard-won environmentally so so-called 'Rare' Earth metals. It would behoove all robowaifuists to find/devise systems that are much less -dependent on them. >Unrelated Cute: https://www.youtube.com/shorts/HSDNFXXOThE Daww! I see the cousins are still visiting. I wonder if they have something special planned for Christmas & New Year? :D

Differential Swerve Drive for rolling platforms, it uses both motors for driving forward so it is more efficient: https://youtu.be/8RQmpUEgoYU Might be useful for a Gardevoir suit, waifu car or a waifu with wheels. Swerve drive means it can drive sideways, which is more similar to a bipedal creature that can move out of the way easily.

>>26857 >Might be useful for a Gardevoir suit, waifu car or a waifu with wheels. Very cool! I'm actually in the middle of a smol robotic cart->Tachikoma design for Sumomo-chan. This is very timely for me Anon, thanks! :^)

>>26118 >Cheap motor update: I cannot figure out how to use the CAN bus with the arduino can hat v2 and a program to move the motor with a joystick. I'm feeling extremely lazy so I think I need to cut all unnecessary idle tasks and try again.

OP, seems it's time for a new thread please These are some good ones, Anon. Thanks! :^)

NEW THREAD NEW THREAD NEW THREAD (>>27021) (>>27021) (>>27021) (>>27021) (>>27021) NEW THREAD NEW THREAD NEW THREAD