/robowaifu/ - DIY Robot Wives

Advancing robotics to a point where anime catgrill meidos in tiny miniskirts are a reality.

Downtime was caused by the hosting service's network going down. Should be OK now.

An issue with the Webring addon was causing Lynxchan to intermittently crash. The issue has been fixed.

Max message length: 6144

Drag files to upload or
click here to select them

Maximum 5 files / Maximum size: 20.00 MB

More

(used to delete files and postings)


Vera stayed by Anon's side, continuing to support him in building new programs, but their primary goal was no longer work or money or fame.


Open file (428.51 KB 1500x1063 general_engineering-01.jpeg)
Open file (150.45 KB 1024x747 tools_(resized).jpg)
R&D General NoidoDev ##eCt7e4 07/21/2023 (Fri) 15:25:47 No.24152
This is a thread to discuss smaller or general waifu building problems, solutions, proposals and questions that don't warrant a thread or touch on more than one topic. In a way this is a technical meta, minus news. Keep it technical. A lot of topics in the old thread here >>83 have a thread on their own by now. The main topics in the old thread with the link to the related dedicated threads are listed here - it was mostly about actuation at the beginning: Topics in the old OP: - liquid battery and cooling in one (flow batteries) >>5080 - artificial muscles (related to actuators >>12810) - high level and low level intelligence emulation (AI) (related to AI >>77 >>22 >>250 >>27 >>201) - wear and maintenance, including repairs - sanitation >>1627 (related to actuators >>12810) > cheap hydraulic and pneumatic muscles > woven sleeves out of strong nylon fishing line > exhaust excess heat by breathing and panting (related to thermal management >>234) >>1635 (related to energy systems >>5080) > sitting in her 'recharging chair' > swapable battery packs, air tanks > sleeve and other 'fabric' weaving TOOLS >>1691 -TOOLS- > automatic winding >>1698 -TOOLS- > looming mechanism >>1711 (related to actuators >>12810) > belt to transfer the mechanical energy to joints >>1715 (related to actuators >>12810) > not uncommon to use wires >>1716 (related to actuators >>12810) > double-end style towing sock cables (for hydraulics/pneumatics) >>1716 (related to actuators >>12810) > synchromesh drive systems >>1757 (related to actuators >>12810) > wires that themselves contract when heated >>1758 (related to actuators >>12810) > air muscles ... using an origami folding approach >>1779 (related to skin >>242) > self healing stretchable electronic skin >>1792 (related to actuators >>12810) > cable driven design >>1793 > klann linkages >>1799 > Jansen's Linkage >>1879 > software for waifu mechanical development >>1993 > hands (related to hand development >>4577) >>2032 >>2048 > decentralized computing of motion >>2377 (related to AI >>77 >>22 >>250 >>27 >>201) > AI and memory related, fuzzy NAND >>2670 --LIQUID-- > algae living in the internal water tanks >>2754 > copper for internal tanks >>2756 (related to vagoo / punani >>419) > slime onahole >>3182 (related to AI >>77 >>22 >>250 >>27 >>201) > combines MuZero, the Intrinsic Curiosity Module, Go-Explore, Hindsight Experience Replay and Divide-and-Conquer MCTS > pretty tough AI questions to solve >>3223 (related to AI >>77 >>22 >>250 >>27 >>201) > generally AI related thread >>3240 (related to AI >>77 >>22 >>250 >>27 >>201) > Flowtron ... autoregressive flow-based generative network for textto-speech synthesis >>3249 (related to AI >>77 >>22 >>250 >>27 >>201) > Tacotron, transformers >>3930 > intellectual humility >>4827 > most common methods to glue (3d printed) plastics together >>4831 > micro RC car, assembly of small stuff >>4861 > 90 facts about the human body >>5038 > DIY and inventors thread in the dollforum >>5448 > 3D printers to produce tensegrity structures >>5457 > silicone ball jointed doll ... held together by wire or string. >>5457 (related to skin >>242) > silicone skin covering the body with two layers of thin plastic in between ... skin feels like actual skin moving around >>5479 > Polyurethane foam to add plastic to the outside >>5487 > OpenLOCK >>5674 (related to actuators >>12810) > free spring calculator >>5928 (related to vision >>97) > lidar sensors for less than $100 >>5942 (related to organisation >>4143) > summarize papers and abstracts ... research sandbox >>5944 (related to AI >>77 >>22 >>250 >>27 >>201) > AI simulators >>6420 (related to energy systems >>5080) > alternative energy sources >>6503 (related to actuators >>12810) > compressor-free McKibben muscle using high-pressure electrolysis > hydrogen gas muscle >>6578 (related to AI >>77 >>22 >>250 >>27 >>201) > 'understanding'/real world context of the concept itself > quick mental images , AI, NLP >>6587 (related to AI >>77 >>22 >>250 >>27 >>201) > learning from artificial imagination > world models >>6845 > direct ethanol fuel cells >>6845 (related to AI >>77 >>22 >>250 >>27 >>201) > graph databases >>8311 (related to skin >>242) > fabric skin without silicone rubber > Mylar coated with aluminum > muscles should be fabric and a pliant closed cell foam (>>12810) > touch sensitive foam through carbon black and pressure changing resistance >>8312 (related to vagoo / punani >>419) > private parts could be like a pair of shorts with internal pockets for private parts. >>8314 >>8328 >>8361 (related to skin >>242) > Lycra, spandex, and elastane > full body suit >>8364 (related to actuators >>12810) > which bearing should be used in printed drives or in joints >>8369 > bearing types ....
... >>8502 (related to actuators >>12810) > electically activated polymers, dielectric elastomers >>8504 (related to skin >>242) > aluminized mylar film or space blankets for electrodes >>8522 (related to AI >>77 >>22 >>250 >>27 >>201) > VHDL, Hardware Description Language, FPGAs >>8523 > millimeter wave length for power and communication to the muscles > salt water solution in conduits to carry power. (energy systems >>5080) >>8546 > FPGA / computing in elastomers >>8547 (related to actuators >>12810) > mini-hydraulic systems / elastomer motors >>8548 > automating the hardware >>8728 > lists of mechanical solutions >>8729 (related to actuators >>12810) > double-wishbone universal joint ... for transferring rotary motion >>8730 > more human-like design vs more robotic >>8955 (related to AI >>77 >>22 >>250 >>27 >>201) > ReZero, AI, machine learning >>8975 (related to actuators >>12810) > Muscles, Reflexes, and Locomotion, library links >>8976 (related to AI >>77 >>22 >>250 >>27 >>201, SBCs >>16) > XNOR.ai >>8977 (related to programming >>128 >>86) > NIM - lang >>9080 (related to humanoid projects >>374) > Glados (animatronic or robot design) >>9184 > make our own circuit boards, voltera.io >>9265 (related to AI >>77 >>22 >>250 >>27 >>201) > SQLite, serverless databases >>9277 (related to AI >>77 >>22 >>250 >>27 >>201) > ACID transaction functionality > filesystems and (generally) text files for my 'database' >>9319 (related to AI >>77 >>22 >>250 >>27 >>201) > torch.from_file, which read and writes tensors directly to files and works extremely fast >>9325 (related to decentralized AI training >>8958) > BOINC system, decentralized ML training >>9332 (related to AI >>77 >>22 >>250 >>27 >>201) > external servers at home (for AI) >>9958 (related to actuators >>12810) > planetary roller screws >>9984 (related to actuators >>12810) > ball screws > solenoid-actuated 'shuttle' >>16707 > crush ribs for inserting ball bearings into plastic parts >>16750 > scents men prefer relating to women >>16766 (related to thermal management >>234) > elf/fox/cat ears or wings as cooling structures >>16766 > structures could be used to help with balance >>16767 > wings with wheels >>17204 (related to skin >>242) > multifunctional flexible and stretchable graphite-silicone rubber >>17468 (related to business >>3119 >>1642 >>18572) > early models ... geared towards the elderly or PTSD victims > business / profit >>18088 (related to skin >>242) > self-healing light guides for dynamic sensing (SHeaLDS) >>18276 > easy to print 20202 t-slot compatible bar >>18688 > Uncanny Valley >>18744 (related to actuators >>12810) > Open Emag - Make your own electronecks mechanisms (to look at and take inspiration from) >>19009 > functional tattoos ... visible under a special light >>19340 (related to actuators >>12810) > DaVinci's knight >>19390 > micro-robotics, biomimicry >>19392 (related to actuators >>12810) > Simple Field Oriented Control (FOC) >>19398 > rollers as axis, shaft or pivot. Or in some cases to join parts together without needing to add screws and nuts >>19408 (related to actuators >>12810) > printed clamps to ensure a tight fit >>19442 > links to the topics in the MaidCom Development thread >>19513 (related to actuators >>12810) > making gears by gluing timing belts onto 3D prints >>20119 (related to skin >>242) > flexible display films ... as skin >>20122 (related to skin >>242) > thermochromic powders in the skin >>20123 (related to skin >>242) > LEDs shining through the skin >>20327 (related to skin >>242) > impregnation of textiles with silicone (skin related) >>20344 (related to skin >>242 and armatures >>200) > silk paper shell designs >>20775 (related to skin >>242) > custom (open source) microphones >>20789 (related to actuators >>12810) > 'Parallel-Spherical' joint mechanism > torso >>21494 (related to actuators >>12810) (related to skin >>242) > sensing angular rotation >>22032 (related to vision >>97) > 3D printed optical sensor framework >>22129 (related to skin >>242) > fuzzy skin (3D printing) >>22284 (related to thermal management >>234) > heart that pumps coolant using a nechanical pace maker >>22287 (related to actuators >>12810) > creative gears >>22438 (related to actuators >>12810) -TOOLS- > self winding system Things which are left over and belong into this thread are mostly mechanics, the topic of skin (without sensors), tools, general design questions, and things which are really special.
Literally the most comprehensive OP I've ever seen NoidoDev. Nice work!
Open file (171.49 KB 765x1200 general_engineering-02.jpg)
Open file (995.67 KB 1280x870 general_engineering-03.jpg)
A channel around building tools to build other things: https://www.youtube.com/@MrInnovative e.g. DIY Arduino based Wire harness wrapping machine: https://www.youtube.com/watch?v=s923C0_tkjw >>24155 Yeah, thanks, took me some time. And I still forgot the pictures in the second posting. But okay.
>>24157 >e.g. DIY Arduino based Wire harness wrapping machine: If any anon finds any information about DIY actuator wire-wrapping machines, it would be highly-appreciated if you'd post it ITT! :^) >And I still forgot the pictures in the second posting. Oh, I think the OP is just right. And now you've got some add'l resources available for thread #3. BTW, R&D General has always been one of my favorite threads going all the way to back in the day. Thanks again for reviving it! Forward!
>>24158 Did you look into this? >>18744 > Open Emag - Make your own electronics mechanisms (to look at and take inspiration from) I wrote "Make your own electronecks", lol. That MrInnovative channel I just mentioned >>24157 also should have something like that, I didn't look through all the videos yet.
>>24160 >Did you look into this? I 'looked into it' insofar as reasoning my own OC regarding the processes involved, and I think I have settled on a rational basis for it. Unfortunately I don't have the time currently to begin exploring the practical prototyping of this work. If I receive the funding I'm going for r/n, this is near the very top of the stack -- since actuation is effectively fundamental to robotics entire! :^) >tl;dr We need to 'wind our own', if we're ever going to have thousands of customizable electrical actuator systems. Regardless, for our Model A Robowaifu designs, any such actuator-construction systems will be opensauce, and freely-given to the world, such as they are. Hopefully hundreds of competitors can soon be producing custom actuators and innovation will explode in this area.
Thoughts on 3d printing the skin with TPE filaments?
Open file (5.29 MB 2560x1920 chesthip.jpg)
>>24163 Works well enough, be mindful that you must print very slowly with retraction settings of at least 2 mm. Internal details will be inaccurate. TPU is extremely forgiving as out bends and compresses, you still need to be careful and mindful of its properties. Pixel is printed too fast and at 1 meter scale.
>>24164 I think TPE is not the same as TPU. I'm hoping its less hard and more flexible than what you showed on the picture.
>>24165 I'm pretty sure that Grommet & other anons here have discussed TPE filament before (eg, >>23370, ...), so I'm confident you can get some good feedback ITT. Please be patient, we're a slow board at this time. In the meantime, can you share what ideas you have so far about your robowaifu's skin, Anon? For instance, I presume you want many areas that are soft since you want to print in TPE. But does everything need to be so, or would you be satisfied if she has other areas that are more firm or even hard? Regardless, looking forward to seeing your progress with your robowaifu Anon! Cheers. >=== -prose edit
Edited last time by Chobitsu on 07/22/2023 (Sat) 00:54:20.
>>24171 Yes the boobs and butt would probably have to be latex filled with silicon jelly probably. Also the fact that the 3d printer is limited to a certain size means everything needs to be put together in pieces and the skin would would have to be glued piece by piece if its 3d printed.
>>24172 On second thought never mind the latex. I'm of the idea of taking as many shortcuts as possible.
>>24172 >Yes the boobs and butt would probably have to be latex filled with silicon jelly probably. Makes sense. One of the ironic twists to all this is that in their pursuit of beauty, 3DPDs have financed the R&D of medical-grade, fake bewb prosthetics which we here will all benefit from the use of heh. :^) >Also the fact that the 3d printer is limited to a certain size means everything needs to be put together in pieces and the skin would would have to be glued piece by piece if its 3d printed. Yes, that's a complex problem. In my own case, I want to go for commercial production, so my primary thinking is to use something like big rolls of spandex textiles impregnated with some type of near-silicone-like material that can be a) washed, b) cut, stitched & fashioned together into a type of 'custom bodysuit', which can then be c) stretched over and onto the robowaifu's external exoskeleton/shell and snugged down tight to fit. Being forced to use only smol consumer-oriented 3D printers is a much more challenging design requirement than mine is, IMO. >=== -prose edit
Edited last time by Chobitsu on 07/22/2023 (Sat) 01:07:04.
>>24173 I just thought of something else. There are TPE dolls out there that have no exoskeleton really. The skin and the exo skeleton are one in the same if we go that route. >>24174 Yeah I thought about the commercial aspect as well but there's just so much friction. The patent itself is a huge hassle and you have to be careful about borrowing stuff and make everything from scratch so it can be patented. I'll just share the stuff and hope for some collabotion I guess. Like the guy who made reprap and didn't profit from makerbot someone is going to have to take one for the team. There is also the fact that I don't really have any connections. Might make a video about that.
>>24175 >The skin and the exo skeleton are one in the same if we go that route. As mentioned in the 5 principles thread (>>24047), economizing & combining things together is usually a good idea economically, when it's feasible.
>>24152 OMG, I tried to get ChatGPT to sort this stuff by topic and now I'm full of anger. What a piece of garbage.
Open file (14.71 KB 281x741 waifuSkin.jpg)
>>24174 It would be simple to take a women's unitard and impregnate it with silicone. From there you could zip her skin on and off as needed. Just need to tailor the robot to fit. Fast, cheap, and you could make some parts touch sensitive by mixing in conductive powder.
>>24171 >TPE filament and here's TPU. I must say this plastic ranger guy is like the best 100% most excellent guy on the whole entire planet on general knowledge of plastics. He writes very informative general purpose articles on plastics. https://plasticranger.com/what-is-tpu-material/ and to have them together https://plasticranger.com/what-is-tpe-material/
>>24152 This really nice. Thanks.
Oops, my apologies. Somehow got on the end of this threads part 1 and commented. >>24214 It was about fiber optic sensors which is brilliant, here, >>22032
Using fiber optics for pressure sensors I "think" you could also use them for position sensors. Look at this article on gray's code. It's a way of making electrical signals and optical sensors more accurate. It allows low resolution and noisy signals to be determined accurately. https://en.wikipedia.org/wiki/Gray_code The picture you get through fiber is distorted but...we don't care. It may be crap but as long as you program whatever comes back to be whatever position it doesn't matter if it's not optically correct. Only that the pattern changes with movement. So one fiber carrying the light from a paper picture connected to the rotating appendage fed through fiber to a small section of the camera that reads it could do almost instant position sensor determination. This fiber optic sensor thing has got to be one of the best ideas ever that someone thought up.
I had an idea for a super cheap easy way to make electric coils. Especially for actuators. Use Low temperature metal alloys. This is after all what solders are basically. There's all sorts of them at all sorts of temperatures. You could use normal 3D printer filament to print the channels. Which would become wires when the metal is melted into the printed form. You could print PVA, melt the metal into it, dissolve the PVA with water, then mix ferrite with liquid plastic, pour into the dissolved PVA spaces and now you have a custom transformer and you're not stuck with traditional transformer shapes. The beauty of this is most of these metals you can melt over and over. If you get high enough temperature metal you could use them as actual conductors on finished products and have custom solenoid coils to fit where you need. Chobitsu was asking about coil winders but here's a way to avoid them altogether. After I had this idea I found someone was doing something sorta like it. They were making inductors in silicon, melting the metal and then stretching the silicon to change the inductance, then solidifying that value inductance by letting it solidify. Abstract https://www.semanticscholar.org/paper/Remoldable-inductors-based-on-self-heating-fusible-Lazarus-Bedair/b481fe3c8dd2e44b4f6a5e8b47fd57adef61a6e4 BTW here's a book on building a "Universal Coil Winding Machine" http://libgen.rs/book/index.php?md5=1DCAE7F26AA7CCF6D67810D54F3A619E
>>24216 This is actually a pretty fascinating idea Grommet. And yes, I think that as long as all the 'sensors' have the same basic dynamic range of signal, that the deflection off the registered baseline should be fully effective as a 'change-in-position' sensor array. There will be a lot of software maths involved, but nothing particularly difficult in general. As an added bonus, it should calculable on GPU or FPGA -style processors really quickly. >>24218 Can't say I'm really super-clear yet on your idea here Anon tbh. BTW, thanks for the book link. I also found a couple of other related texts that are applicable to our interests here. Cheers.
>>24188 I hope you feel a little better now NoidoDev. Thanks for all your hard work! :^) >>24197 >Fast, cheap, and you could make some parts touch sensitive by mixing in conductive powder. Thanks Kiwi, great idea!
>>24165 TPU is a TPE. I think no one in the printing channels I watch uses the term TPE.
>>24218 >>24221 >[low melting metal windings]Can't say I'm really super-clear yet on your idea here Anon I can more easily explain with a picture The first is a regular motor winding, the second a linear motor winding. I'm only using the pictures as a "general" illustrative purpose. Many different architectures are possible. As you can imagine winding these tiny ass coils if your actuator motor is say one centimeter long or around, [eye lid motor as an example], would be difficult. Just getting the wire to wind it with would likely be difficult. But you could 3d print the form to "hold" the winding, then melt in low temperature metal winding. The low temperature metal "is" the conductor, "is" your coil. The low temp. metals melting point is below the melting point of the filament you printed. So you heat the form and the metal and pour it in or just submerge it into the metal to fill. Searching...here is the first "iron" filament I came up with. It's ferromagnetic, so that's the field concentrating part of the motor. https://www.proto-pasta.com/products/magnetic-iron-pla Other methods to add the ferromagnetic part is to print in PLA, melt in metal conductors, wash out PLA then fill spaces left with ferromagnetic powder and then add plastic, like super glue, to hold the particles together. Even more important than these typical shapes are the odd shapes you could come up with. You could 3D print little stepper motors, all sorts of magnetically operated solenoids, switches, etc. in all sorts of shapes. Shapes that would be impossible with typical wound metal coil motors. https://en.wikipedia.org/wiki/Fusible_alloy https://www.xometry.com/resources/3d-printing/magnetic-iron-pla-3d-printing-filament/
>>24216 >Using fiber optics for pressure sensors I "think" you could also use them for position sensors. I remember now: My idea once was to communicate with light, through the body, but not using uninterrupted fibers. Think about the problem of cables in moving parts breaking over time. We should use a hinge out of metal to transmit electricity, then using light in different colors and frequencies to send the data from joint to joint.
>>24218 >super cheap easy way to make electric coils. Are regular cables too expensive? >There's all sorts of them at all sorts of temperatures. You could use normal 3D printer filament to print the channels. Which would become wires when the metal is melted into the printed form. Creativity has it's limits. >They were making inductors in silicon, melting the metal and then stretching the silicon to change the inductance, I hope I mentioned something like this before and linked the video somewhere. Wrapping a wire around a silicone part and stretching it, should do the same trick. Some female scientist invented or discovered that a few years ago.
>>24229 > fibers. Think about the problem of cables in moving parts breaking over time I'm not seeing that. Plastic fibers are really flexible. True some Chinese brands might be cheap junk so you would have to know what type plastic fiber optic cable you bought. What are we talking about? Breaks after 10 years??? Maybe longer. No. The utility and super cheap cost outweighs other solutions by more than a order of magnitude. I looked up the cost of plastic fiber optics and saw you could get a hundred meters for under $20 USD! Better fiber can't be too much more. Wires, well wires have plastic coating so...no better. Where the utility lies is in the interface. If you use capacitive, magnetic, any other sort of electrical type sensor then you have to add all sorts of parts to the input of a microcontroller. I've been looking at this and MC interfaces are complicated and they cost. You need another board, a bunch of resistors, capacitors (usually) and buffer IC circuits to make this work. You will have to have this for the actuators but adding this complication to the sensors is a whole mass of extra work. Just compensating for stray capacitance,[if you use capacitive sensors], in the pressure senors will be a total pain in the ass and will require constant calibration. Look it up you will see this is true. A lot of design work, more PC boards to make for the interface components. Lots of complication. Lots of cost. But fibers from all the joints, point it at a $9 ESP32 with camera chip and you are done. A thought experiment. Don't parse this too tight. It's just a back of the envelope calculation. Let's say you have 300 actuators of various sorts to be sensed for location. Most you could have 8 bits resolution or 256 units of movement. A few you would need 10 bits. Major limbs. To get 10 bits, I used arm movement. So call the arm movement 2' distance or 609.6mm. Now I say half that to 1218 units of distance so now you have a resolution of 1/2 mm over the whole travel of 609.6mm. Close enough is 10 bits resolution at 1024 units distance for the whole travel of the arm, leg, thigh, forearm. Since the waifu is likely to be smaller this would likely be good enough for 1/2 mm or less resolution in limb, appendage, whatever movement location. Now I speculate that you could use a gray code wheel picture. The one I uploaded has only three bits. You need more depending on which joint it is. One fiber lights up the disc, a lens pics up the picture, sends it through the fiber and then it's displayed onto the camera. A ESP32 CAM Camera Module has 1622×1200 pixels or sensor channels. That's 1,946,400 total. Even if you cut that by 4x for alignment troubles it's still 486,600. In fact to get 10-bit resolution on all 300 muscles for a full function waifu you only need 3,000 pixels. So you're way ahead here. There's a possible glitch in using one fiber to carry the whole picture. "If" moving the fiber screws up the picture, even if the ends are fixed, or changes it as the waifu moves then you can't use one fiber. But that's not really a problem, as we just use one fiber for light and ten fibers to carry the light reflection from the gray code wheel. We have plenty of pixels in the camera to pick this up. Even better. This would be so fast to put together. You make a standard bracket to hold the wheel and place the fibers. The brackets, wheel holders, etc. could all be the same. You place the fibers in and snap closed. Each limb will have different length fibers. So you do just like car manufacturers make wire looms for cars. They have a big board with channels. They thread the wires, in our case fibers, then cut them to length. Add on connectors and wrap and you have the whole sensor system in one nice package. And even better we don't care what order the fibers for each individual sensor are in. After the waifu is built you give it commands to move one actuator at a time. The various fibers will light up as it moves and the processor then makes a map of each location for each fiber. Saves you a lot of time individually tracing each fiber from one fixed location to another, No need. It calibrates itself in software. That guy that thought up fibergrids, pure genius. One of the best ideas I've seen in my life. https://hackaday.io/project/167317-fibergrid#menu-details
>>24221 >deflection off the registered baseline should be fully effective as a 'change-in-position' sensor array. There will be a lot of software maths involved I'm shit for programming but I think the math would be simple, and fast. For movement of the limbs. You have two arrays of data. You need,roughly, 300 actuator positions at 8 bits each. So 2,400 bytes. Next to nothing. Let's say you have 3,000 sensor points, 24,000 bytes. So you make one array that holds and one that is made of constant poling from the sensor camera. Each 37,000 bytes. Compare the one held to the newest and if it changed, it moved or a sensor was triggered. Such a small array of numbers compared has got to be near instantaneous even with a simple microcontroller. The actuator set is only 2,400 bytes. So it could monitor what it "thinks" the movement should be with another set of arrays that is compared to the actual close to real time read sensor value. By using only on-off type values for the camera and maybe less resolution surely even the ESP-32 could do 1/100 second tracking with a little computing time left over. Maybe a lot of computing time left over. https://lastminuteengineers.com/getting-started-with-esp32-cam/
>>24233 Brilliant concept. Thanks for the Hackaday link, Anon. Helps clarify the concepts. --- >update: I had a quick glance at the code, too. https://github.com/rand3289/FiberGrid/tree/master This is mostly good modern C++. This style code is usually more comprehensible (and therefore more maintainable), and in many cases also runs somewhat faster than older C++ or C counterparts do. GG. >=== -add 'update' cmnt
Edited last time by Chobitsu on 07/25/2023 (Tue) 05:14:03.
>>24228 I'm still having a hard time imagining how the molten alloy will remain as separate 'wires' in your concept. Won't they all just flow together into a single block of alloy? If so, then the idea of 'how many windings a coil has' disappears wouldn't it?
>>24234 Fair enough, it may be simpler than I imagine at first glance. Thanks Grommet! :^)
>>24230 >Are regular cables too expensive? No. The main point is the versatility of manufacturing complicated coils, which means motors of all types, solenoids, actuators, etc. , they all have coils, without building winding machines or not being able to build some complicated coil at all. Resistance will likely be higher in the low melt part but, so what. How much more than copper or Aluminium, I don't know but would surprised if it were over 10%. If 10% then you just make the coil 10% bigger to carry the same current. You would also likely have to make it bigger to handle heat so it doesn't melt. And even if it does, if it's contained, it will not drain out and will harden right back in it's channel when it cools. After I wrote the above I decided to look for the resistance. I find a chart of different low melt solder alloys and the highest compared to copper was 5%. https://concorde-electronics.com/blog/low-temperature-alloy-solder/
>>24236 >hard time imagining how the molten alloy will remain as separate 'wires' in your concept Imagine a solid block of 3D printed material. Now imagine a coil "negative", not printed inside the block. The negative, not printed, part is put in [merged in the CAD]the block and printed. So now you have a solid block with a "not printed", empty, tube in it. The NOT printed tube is where you pour the melted low temp. alloy in to, when it hardens, makes the coil which is a wire of hardened low temperature alloy. Remember most of these low melt alloys have a far lower melting point than the plastic so the plastic is just a mold. Note the metal hardens so it's a wire when cooled, if shaped like a wire. Going further to optimize it, this is assuming you do not print ferromagnetic packed plastic filament which would do this in one step. Let's say you do the above but print in dissolvable PVA, pour in the low melt metal coil, melt out the PVA with water then surround the coil with ferromagnetic material. Just so you understand ferromagnetic material. It's iron oxide [there are others but it's commonly iron oxide] and attracted to magnets or electromagnets. If you dig in the ground and separate the soil. Dirt. Then take a magnet and sift through the dirt you will get a bunch of iron oxide [ferromagnetic material] stuck to the magnet. So you can do this, wash it in water then use the magnet again and get something to work with from regular soil, dirt. Why use ferromagnetic material? The ferromagnetic material bends all the magnetic force towards the coils and so strengthens the magnetic field. It's why they have ferrites or steel sheets around solenoids, motor, transformer, etc. It confines the magnetic field and makes it stronger "where" you want it instead of it spreading out around the motor, solenoid, whatever. If you are channeling water you need a tube to hold it. Ferrites act somewhat like a tube channeling magnetic fields. Here's something that will blow your mind, or it should it does mine. Watch this short video. https://www.youtube.com/watch?v=eWSAcMoxITw So this business of ferrites and magnetic fields is NOT simple. It's complicated. I've seen no explanation for the video above. Maybe some physics guy will throw a bunch of equations at you but they really do not explain what is going on. It appears to me the field in the metal acts like an analogy to a superconducting coil except it's superconducting magnetic fields instead of electrons and that somehow holds the metal blocks together. And this is not fake it's real.
I printed this comment in the wrong place, Oops, but it belongs here so I'll repeat it. You could use a cheap ESP32 CAM Camera Module Kit and monitor fiber. Do them in a sort of X'Y' grid and then you have what part of a finger, or any other appendage, and how far up. You could use this for position sensors also. Have a rotating bump or wheel press on an array of fibers. You could have 10-bit resolution with 10 fibers that are pressed by bumps or wheels that press on the fibers as the actuator rotates or moves. Just like pressure changes the fiber light when used for pressure sensors. In fact this may be even easier and cheaper than making gray code wheels. So with a little work you could have every single touch sensor and position sensor in one $10 camera. Some ideas are so good they are just...stupendous and this is one. The guy that thought of this was really doing some top notch thinking.
>>24235 I was overenthusiastic about this. Great for position sensors, still very good, but not for pressure or touch sensors. I think I got carried away because it solved SO many problems that I became caught up in wishful thinking. Another reason I think the pressure part worked was I have actually heard of this and thought he had found some simple way to do it. He started out stating that he felt he could which I missed and read that he had. Sigh. The idea still has merit but it's not solved.
Maybe all is not lost. Maybe this guy could not make fiber pressure sensors work but others are sure working on it. Abstract Low Cost Plastic Optical Fiber Pressure Sensor Embedded in Mattress for Vital Signal Monitoring https://www.mdpi.com/1424-8220/17/12/2900
>>24246 Reading further,"...The POFs have been sensitized to pressure, cutting part of the jacket, the cladding and the core (Figure 1a). A fiber holder with a hemicylindrical groove with the same diameter of the fiber was used together with a razor blade to obtain cuts with approximately the same depth. As described in Figure 1b, any applied pressure on the cut point changes the light intensity at the fiber end..." So this can be done and without too much difficulty. The paper.
Found another paper. A much better one. Seems there's not much commercial fiber that has the right properties. These guys found a polymer, "Geniomer 200 (Wacker AG, Germany), a polysiloxane-urea-copolymer with a polysiloxane content of > 90%. The polymer was used as supplied with no further treatment. The melt flow index (MFI) has measured 17.78 ± 0.52 g/10min (n = 3) at 1 kg and 170°C (Wacker datasheet: 170°C, 2.16 kg, 110-150 g/10min), and a melting temperature (Tm) of 171.31..." So tentatively it looks very easy to 3D print with normal 3D printers after making into filament. They then dipped in a refractive index polymer, "...A low refractive index coating material (THV 220, n = 1.36; Dyneon, Germany) was later applied by dip coating the fibers in 5 wt% acetone solution (23°C). The coating thickness was calculated to 16 ± 3 µm by gravimetric analysis (n = 3). After the dip coating with THV polymer they lost their sticky behavior...." The response in these, if they are not lying, is very good. 0-40 Newtons pressure. So fiber optic sensors not outmore complicated than just weaving a few off of amazon. It may well be that you could still use some cheap amazon fibers but...likely too stiff.
>>24242 Thanks for the instruction Grommet, that's helpful! :^) >>24243 >>24245 >>24246 >>24247 >>24257 Fiber has a metric boatload of advantages within a robowaifu. I'm sure they will play many important roles for us in the future, including involvement with sensors.
>>24246 >but others are sure working on it. Yeah, it would be surprise if no one would. It's too obvious of a good idea. In theory it should be sufficient to have a little ball joint, where the ball can move in different directions, holding a fiber. Then find a way to add changing light into the fiber. Maybe the ball could be somewhat transparent, or some parts transparent others not. Then LEDs with different colors would shine onto it. Bending the fiber makes the ball move into another angle, changing the light spectrum. >>24247 >>24257 Thanks, I'll look deeper into it when I'm at a point actually needing something like this.
Imagine an entire robowaifu shell of this stuff, under a soft, transparent skin.
>>24227 I'm going by what bard says and it says that TPE is 20% softer than TPU and cheaper. I'm going with TPE because it seems like the path of least resistance to me.
Open file (539.29 KB 2500x1138 Shore-Hardness-Chart-01.jpg)
>>24227 You are correct. TPU (thermoplastic polyurethane) is indeed a TPE (thermoplastic elastomer) a type of copolymer which mixes a plastic and rubber to achieve thermoplasticity and a rubbers softness/malleability. What really matters is the shore hardness of the filament. Picrel is a chart that can be helpful for getting the gist. Note how there's multiple scales, it's important to note the letter when checking the scale. I personally use 98A TPU which feels somwhere between skin and leather in terms of flexibility and apparent softness.
>>24268 >Shore-Hardness-Chart-01 Great chart, thanks.Very useful
I think the elephant in the room is the legs. We'd want to be able to spread them for the sexy time too. How to go about this.
>>24287 I think you need look no further than the ball-and-socket design that God created for the hip joint. Artists like Sukabu do a rather good job of rendering this basic approach into a robowaifu-centric set of studies.
I can't get my 3d printer to work and I've been thinking. Wouldn't slush casting be better anyways? Although 3d printing could aid slush casting by making a 3d printed mold though.
>>24297 >Although 3d printing could aid slush casting by making a 3d printed mold though. I think several anons would probably agree with you on this point Anon, good thinking!
Open file (273.70 KB 2480x3508 legs.jpg)
>>24293 This thing is never going to get solved unless you guys get on krita or ms paint. Here is my attempt.
Open file (19.73 KB 392x423 RevoltechJoint.png)
>>24300 Did you use one of those shuffling robots that pretend to walk on ratcheting wheels as inspiration? >Joints I'm still going to be using Revoltech inspired joint system. Currently experiencing with methods of implementation.
>>24298 For everyone who wants to get into casting, I highly recommend Smooth-On silicone to start out with. This channel is great for beginners, many tutorials that are clearly articulated. Always remember to use a release agent too. https://m.youtube.com/@RobertTolone
>>24302 what about the top though? What makes the sphere at the top move?
Open file (46.07 KB 523x866 robots.gif)
>>24304 An actuator of course. It's a sphere with two attachment points that could be rotated. I would recommend looking into articulated figures and robots walking on video sites. This website is a good resource for researching robotic motion in general. http://www.robot-welding.com/robots.htm
>>24287 Related (walking): >>243 I shared an partial sketch of my idea about this here: >>23962 That said, I didn't think rigorous enough about every situation. Which is really something we should be doing in regards to every movement of the body. Technically the problem is close to back-driveability. A outside force to the legs should make them move, but they also need to function with actuators. If it's just about security, as in getting out of the way, then some spring or rubber string mechanism that can overwrite the current state would be good enough. It would snap back after the outside force is gone. I don't know if my idea is going to work, but I want to use strings, springs, keep the servos back-driveable and also adjust automatically, if the situation indicates that the new position of the legs is meant to be the new state. General remark: Female walking is closer to how models walk, and we at least want that as an option. The legs do not just move forward from the side of the body. I also plan to take the hip moving while walking into account.
>>24300 While i drew that quickly I think it'd fit the design pretty well. The vertical gear with two side gear arrangment is called a bevel gear train. Each side would be offset by 180 degrees, so while one side is up the other one is down. The wheels could server for the robot to turn to different angles. So the waifu rotates say 45 degrees with the wheels and then walks without the wheels. A regular dc motor with high torque could be used for this. I'm not sure but I think regular dc motors are cheap than stepper or servo motors and never mind actuators. Slap in a couple of pneumatic cylinders to spread the legs and you got something that's good enough. The pneumatic cylinders would be used to limit the spreading and would not need to be powered by anything.
>>24300 >>24309 Thanks for the work Anon. Good luck. >>24302 >>24306 Neat. Please provide more Revoltech information if you will Anon. >>24303 > I highly recommend Smooth-On silicone to start out with. This. It's quite expensive, but I've seen it demo'd with flawless results multiple times. >>24308 >Female walking is closer to how models walk, and we at least want that as an option. The legs do not just move forward from the side of the body. I also plan to take the hip moving while walking into account. Great, biomemetic thinking NoidoDev. But this is a rather difficult set of kinematics problems for us to solve, heh. :^)
Open file (252.95 KB 501x414 Screenshot_47.png)
Open file (105.94 KB 848x1199 internals-001.jpg)
Open file (86.66 KB 1024x558 HRP-2-Body_(resized).jpg)
>>24309 >gear arrangment is called a bevel gear train Reminds me of this: https://youtu.be/W6xzBVVuml8 - I thought about finding a way to make this work without drilling into metal. >regular dc motor Might be a good idea if you don't plan to have an enclosure which you can't remove later. I think dc motors have coals which will need to be replaced after a while. >The pneumatic cylinders would be used to limit the spreading and would not need to be powered by anything. Excellent idea. We should also keep HRP-4 approach as option in mind (picrel 3).
Open file (3.28 MB 3000x4000 IMG_20230730_102839.jpg)
>>24316 I found this toy laying around the house. I've opened it before to put a resistor in it to make it quieter. It also comes with a motor. It might be good enough to make a test for the legs on a small scale. I'll also be going to the hardware store tomorrow so I might be able to get some other stuff as well.
>>24319 Oh forgot name handle
>>24297 >3d printing could aid slush casting by making a 3d printed mold That's definitely my way of thinking also.
Open file (293.63 KB 463x460 Screenshot_77.png)
Open file (589.66 KB 464x832 Screenshot_78.png)
Stringing piece, 3D printed: https://youtube.com/shorts/EzQtoJlunwM
Open file (209.20 KB 460x311 Screenshot_79.png)
Hannah from @DaveMakes got additional emotions to change the skin around her eye cavity when she moves her eyes up and down. Never seen that anywhere before: https://www.youtube.com/shorts/lbZZIK33T1I
>>24749 Very pretty. I'd expect that Kiwi and others can utilize similar approaches for winding tensegrity suspensions, as well. :^) >>24750 He's made lots of progress over the year or so. Thanks for the update, Anon! Cheers. :^)
>What's the difference between a CPU and GPU? And what the heck is a TPU, DPU, or QPU? Learn the how computers actually compute things in this quick lesson. https://youtu.be/r5NQecwZs1A >Curious exactly what happens when you run a program on your computer? Read this article to learn how multiprocessing works, what system calls really are, how computers manage memory with hardware interrupts, and how Linux loads executables: cpu.land
Open file (498.57 KB 1159x1006 ClipboardImage.png)
>>24749 Does anyone know of a technique to automatically generate those strings? It should be possible, Cura does similar with infill patterns but, none that I know of can replicate this. If I could edit the "lines" infill pattern to have layer between, it would be close enough, Would be very useful for print in place tensegrity.
>>24950 Didn't he explain it in the video or in the comments?
>>24950 >Would be very useful for print in place tensegrity. Ahh, the old 'necessity is the mother of invention' driving-force of nature, ehh? :^) Very innovative thinking Anon. Godspeed.
>>24952 I looked, it actually isn't. I could try in OpenScad... <the rabbit hole teases you, some call it they Abyss>
>>24749 and >>24994 Related: >>24996
Wraith Madoka's true form has a fascinating design that looks similar to a tensegrity structure. This concept could be modified into a cute and functional robot
Open file (42.06 KB 640x232 FourBarMechanism.png)
Four bar mechanisms can implement almost any desired motion in very simple and elegant designs. Hackaday has an insightful starter. https://hackaday.com/2017/03/29/marvelous-mechanisms-the-ubiquitous-four-bar-linkage/ Wiki; https://en.wikipedia.org/wiki/Four-bar_linkage Valuable deep dive; https://dynref.engr.illinois.edu/aml.html Here's a video; https://www.youtube.com/watch?v=XvYNqj4awM8
Are there any resources for robot spines?
>>25824 Kiwi's link was very good and has a lot of links that link from it that are very comprehensive. One I found might be useful if not too academic.(most of it is not directly related to robowaifus but some good data points are in it relating to power needed, etc.) Animal as a Machine and a Prime Motor and the Laws of Energetics https://ecommons.cornell.edu/items/51558415-c151-49a3-8a0f-3dcc27a1e2e9 Large set of books on kinematics at the same site https://ecommons.cornell.edu/collections/78104a5e-7f84-41ea-929a-1fc694f75ba2
Hypoid vs. Worm Gears: A More Cost Effective Right-Angle Reducer: https://www.orientalmotor.com/ac-motors-gear-motors/technology/hypoid-vs-worm-gears.html
There are some STEMMA-QT connectors which make connecting a Raspi to I2P easier, there are even hubs for this. https://www.youtu.be/-d0rOHjntAo >Yes you CAN install STEMMA-QT / qwiic connectors on a Raspberry Pi! Several popular alternatives are shown, including hats, a shim, and directly wiring a port to the board. Skip breadboarding for I2C components, and build faster, more fool-proof projects.
Does she need a bellybutton? Making a more feminine design for clothing.
>>26330 Looking good Kiwi! No, I wouldn't think she particularly needs a bellybutton (particularly during this early prototype phase). I suggest doing things that simplify printing/assembly of the mesh flats at this stage tbh. BTW, it strikes me that her Lats are a bit too masculine, but admittedly it's hard to tell w/o a 360' turntable animation. Keep up the good work, Anon! :^) >=== -sp edit
Edited last time by Chobitsu on 11/11/2023 (Sat) 18:56:01.
>>26330 A question which comes to my mind is: Why are you still working on a human-like body design since now we know that Make Human works for that?
Open file (255.78 KB 825x1123 ClipboardImage.png)
Open file (257.55 KB 825x1123 ClipboardImage.png)
>>26331 I feel the same way, I favor a very feminine and petite frame. I'll try to make her lats softer. Optimizing for print-ability while having her able to wear clothes is a delicate balancing act. >>26332 I prefer a more stylized body than I can get from MakeHuman. I fully support anyone using it. I agree that most should use it instead of investing time to make their own mesh. I'm just a weirdo who needs things just right.
>>26337 >I'll try to make her lats softer. Good thinking IMO. Also, now that I can see her backside, again, I think her gluteus maximus are too masculine. I'd suggest giving her more of a rump. >I'm just a weirdo who needs things just right. Haha. Autism isn't being weird Kiwi. It just shows you have a real vision you're following! Cheers. :^)
Open file (207.06 KB 714x1000 ClipboardImage.png)
Open file (215.29 KB 714x1025 ClipboardImage.png)
>>26340 Smoothed things out. Her new lats are the same size, just much rounder. It should look better while still being appropriate for clothes. Made her shoulders smaller to be slightly girlier. Smoothed her booty over, same size, just rounder. More rump like :^) You can see the original and new as halves of the mesh. Added a slight bellybutton. I remembered I made a promise, long ago. Because raisins (I also have a thing for navels :^) (Her arms and legs are random kitbashed addons for the sake anatomy.)
>>26350 >More rump like :^) Heh. Looking forward to seeing the final outcome, Kiwi. :^)

Report/Delete/Moderation Forms
Delete
Report