/robowaifu/ - DIY Robot Wives

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

Build Back Better

Sorry for the delays in the BBB plan. An update will be issued in the thread soon in late August. -r

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)


“Nothing in this world can take the place of persistence. Talent will not; nothing is more common than unsuccessful people with talent. Genius will not; unrewarded genius is almost a proverb. Education will not; the world is full of educated derelicts. Persistence and determination alone are omnipotent. The slogan “press on” has solved and always will solve the problems of the human race” -t. Calvin Coolidge


Prototypes and Failures #4 Noido Dev ##pTGTWW 01/23/2024 (Tue) 03:17:26 No.28715
Post your prototypes and failures. We fail until we win. Don't forget looking through the old threads here >>418, >>18800 and >>21647 to understand how we got to where we are now.
>>29910 My plan is to use this as a simulation for the control software. After I finish the fleshlight roomba with two motors, a contact sensor and a camera on top I want to share the simulation with you guys. I will probably make a small tutorial in python on how to use the data and send commands to the simulation in Unreal. I think in it's first iteration it will only run in Windows. My goal is to make the simulation available in your local network. This way you can use any computer you want to receive data and send commands.
>>29720 Unreal is such a pain in the ass when it comes to animation blueprints and retargeting skeleton animations. I still can't wrap my head around them. Can do just about anything else but not that.
>>29720 Fuck that bed looks so nice.
Started a new attempt to create some skull, starting with a mandible. Using a human model and then extract a pattern from it, which I can work with. Still early, but I wanted to post something, since it isn't much going on.
>>30464 Why are there missing teeth?
Open file (12.09 KB 467x419 kneeT01v01a-001.png)
Open file (12.12 KB 467x419 kneeT01v01a-002.png)
>>30465 Idk, might be from an old women. I just try to use it as a model to get something vaguely similar, of course with teeth. It's about the bone in this case. The teeth are useful for orientation. I already hit a roadblock with this, since the way I wanted it work doesn't work for now. I need to slice the stl file into more pieces and running into errors while trying that. >Knee I also started to sketch out a knee. Again, still very early. But I'm on it.
Open file (18.84 KB 467x419 kneeT01v01b-001.png)
Open file (20.81 KB 467x419 kneeT01v01b-002.png)
Open file (89.33 KB 1024x1024 knee_anotomy.jpeg)
Open file (275.41 KB 1536x2048 knee_replacement.jpeg)
>>30485 New flavor, closer to some artificial knee. Not sure where to go with this exactly, so far. But I'd like to have a kneecap that moves like a human one and also a quite similar look and feel of the knee. It doesn't need to be the same internally. I also downloaded some foot bone model. I think this isn't only important for realism, but will also help with walking. I plan to put some sensors between the bones, so if these move related to each other this will be one way to feel the surface of the ground. Though, I'm not sure yet since these might not move enough. Maybe it would be easier in this case to make a more massive TPU part, where there are normally different bones.
>>30493 There has been efforts to make synthetic joints out of a special hydrogel https://pubs.acs.org/doi/abs/10.1021/acsbiomaterials.7b00094
>>30501 Okay, thanks, I put it in here into 3D printing: >>30506
Open file (18.21 KB 467x419 kneeT01v01b-003.png)
>>30493 Not sure if I an pull it off, but I guess it should be connected with a pole that can rotate and move with it, but only within a certain range. On the front there would of course be a string with the kneecap.
not a designer / engineer / techhead, but I have some thoughts/questions for those who are 1) what are you using for reference/inspiration? Personally, just got back from disney and what they do with their robotics/animatronics is pretty decent. 2) from a concept-to-product design standpoint, what are considered midpoints in overall design? The 'AI' aspect of a nandroid/maidbot seems to my ignorant mind the first 'easy' step. A midpoint towards a fully humanoid housed robot might be a form of 'projection' robot that has core tech components in a rugged casing and can basically follow/roll along after you to a greater or lesser degree but then also allows for an image 'projector' to generate an image on a flat surface that you can interact with to some degree (AR/VR component as tech evolves would come next) 3) best structure/musculature material? This gets into areas that are probably going through a massive ramp-up in terms of development from a layperson/off-the-shelf availability. Basic research shows carbon fiber nylon or polycarbonate structural parts. what about wet-tech style 3d printing? slightly outside our reach at a commercial level for the time being?
>>30514 If you want to discuss what robots should be build, please do this in /meta. Don't come at me with corporate lingo like "concept-to-product design standpoint" and "midpoints in overall design". But trying to answer it shortly: I'm the one who goes for something quite human like, while leaving the more robotic variants to other people. I have another more simple project going on myself, but this is currently stalled.
>>30514 >2) see >>240
Open file (72.56 KB 467x419 footT02v1a.png)
>>30511 Started sketching some foot internals. I'm going from design to design to test how far I get within a relatively short period of time. The idea here is to have something like a internal case, maybe out of TPU which can hold some servos and sensors. The surrounding material would be silicon rubber of course. My previous approach was trying to model something from the outside in Solvespace, but this turned out to be difficult: >>16751 >>16831 >>16855
mandatory message
lurk a year
Edited last time by Chobitsu on 03/26/2024 (Tue) 21:48:33.
>>30522 Idk, you seem to be able to post. The more expensive dolls are made with silicone rubber, so we will need to master this. There are plenty of tutorials on Youtube and forums like the inventors corner in the Dollforum. Using other materials where it makes sense it not something I'm opposed to, but I'm certainly going to use silicone rubber for the skin. Maybe powermesh for cheaper variants, and of course combinations if possible.
Open file (12.33 KB 467x419 hingeT01v01a-001.png)
>>30520 This is my design for a hinge that can carry electricity from one side to the other. The idea is to cut down on the need of cables which would bend and could break. The rod in the middle should be made out of an insulating material, but might still have a steel rod inside. Each pair of discs should be out of metal, while there's a layer of insulation in between. This doesn't necessarily need to carry the whole load alone, but in places with not much space it might.
>>30545 Wouldnt friction cause oxidation reducing the electrical conductivity? You should use graphite powder lubricant.
Open file (707.90 KB 467x387 animatedGIF.gif)
>>30548 >Wouldnt friction cause oxidation reducing the electrical conductivity? I don't know, thanks for the tip. Didn't know that friction does that, I'll look into it. If the part itself isn't used to carry load then the friction might not need to be as high. So this would be worth to keep in mind. >You should use graphite powder lubricant. Good idea.
>>30548 >Wouldnt friction cause oxidation I looked into it for a minute and didn't find anything. If you claim this to be true, then provide a source please.
>>30558 stainless steel should be fine, ironically youre already running a wire through it so iccp (impressed current cathodic protection) might be interesting
>>30559 Friction produces heat for one thing. Or were you looking for other causes? > This study has demonstrated that oxidation does take place at a temperature as low as 200 °C. The oxygen can diffuse into copper to a distance greater than 100 nm after oxidation at 200 °C in air for only 1 minute https://www.researchgate.net/publication/293013781_Oxidation_Behavior_of_Copper_at_a_Temperature_below_300_C_and_the_Methodology_for_Passivation Actually thinking of it more a graphite powder lubricant can be good or bad depending on how many connections are there and if you isolate them where the graphite wont make a short circuit.
Open file (795.30 KB 467x387 hingeT01v01a-002.gif)
I'm getting better at designing parts in a way that I can make animations out of it. It requires a different way of doing things, and always keeping that in mind. >>30567 >take place at a temperature as low as 200 °C I don't get it, this here would be at room temperature or whatever temperature small cables have. >if you isolate them where the graphite wont make a short circuit For adding a lubricant I would need to add a casing, which I will probably do, at least as an option.
>>30573 > I don't get it, this here would be at room temperature or whatever temperature small cables have. Higher the temperatures the more rapid oxidation and it even starts at around 200°C. Also metal expands when heated so uneven wear may occur over time causing rough spots that in theory could lead to small arcing which also oxidizes. Oxidized copper is not electrically conductive. The risk is likely small just worth keeping in mind in design. > For adding a lubricant I would need to add a casing, which I will probably do, at least as an option. I meant all the wire connections presumably pass through same hinge which I didnt consider originally so these need isolated from eachother within the hinge.
>>30574 You somehow seem to not have understand the whole concept. Overlooked something or whatever the reason is. Again: - It doesn't get anywhere near 200ºC, there's no heat involved or not much, or I don't know about it yet. - The different pairs of disk would be insulated, I wrote that in my first posting on this design: "Each pair of discs should be out of metal, while there's a layer of insulation in between." It's also somewhat obvious from the design >>30545 that the light blue disks are for insulation.
>>30573 This a type of slip ring. A component that everyone should endeavor to become formiliar with. https://www.uea-inc.com/blog/what-is-a-slip-ring-and-how-does-it-work
>>30589 -Well yes, I was just giving a possibility. I don't know what kind of temperatures can be generated from a little friction but sure it likely isn't that high. -Yes I can see the insulation in your design. I was suggesting if you used a graphite lubricant it would need the insulation to be small walls to prevent the lubricant from bridging connections If you were to use it.
>>30590 >This a type of slip ring Thanks, I thought that someone must have come up with that idea before and I think I came across the term before or I even asked about my idea and got told about the term but forgot it. We really need to keep this in mind, since we can't just replace bending cables in a human-like robot body all the time, and also it needs to be reliable during operation. Think especially about the hands, how much they move and bend. I can't imagine a cable lasting long in there.
Open file (54.83 KB 467x429 eyesT1v1a-001.gif)
Open file (69.43 KB 467x429 eyesT1v1a-002.gif)
>>30573 Early sketch of my eye design, I only had a very crude one here >>8034 which took me way longer. The goal is to let it move without motors, only using magnets. - The darkgrey rails are there to prevent it rolling sideways, and need to be designed in a way to prevent unwanted movement. This here is just a sketch. - In the holding sphere and on that backside of the eyeball as well, there should be magnets or alternatively metal on one side and magnets on the other. Then the magnets are supposed to move it, without making as much noise as a servo would. The sphere will also need to be electroplated to protect the camera from the magnetism. The hole in the back is obviously for the cables.
>>30723 You dont have to reinvent the wheel. Just use a cable or string pull system like this but with magnets. There is a few ways to accomplish that. Also unless you want ability to go cross eyed you possibly can control both eyes together as a pair instead of individually to save energy usage and a small bit of cost. You also wouldnt need ugly rails to prevent rotation of the eyes which wouldnt have really worked well anyway. https://www.youtube.com/watch?v=6jqvu_MQ5pc Reason being if you place a camera in the eyes it likely is better to avoid magnetic forces that may in theory interfere with image quality and also if anyone wants to use magnet based touch sensors it is better to avoid being near the surface for an eye mechanism.
>>30732 Thanks for the tips. I'll take it into consideration. The problem is, part of the whole idea is to make the head waterproof and make the eye easily removable. I think this will be difficult to do with a string pull system. Idk yet, but it strikes me as more complex. >You dont have to reinvent the wheel. ... YT video Avoiding the common animatronic eye mechanism is exactly one of the goals. It's not water proof. >unless you want ability to go cross eyed Yes, ideally this should work. >You also wouldnt need ugly rails There's really a misunderstanding. Maybe I should've explained it better, but thought it was clear: This above is the backside of the eyeball. I wrote "The hole in the back is obviously for the cables." so I thought it was clear, but you probably got confused with me showing it from the back as well in the first pic. Then it looks like the eye with an opening for the camera. But no, so far this is all the backside. These rails would be internal. >use magnet based touch sensors I didn't plan to do so, or don't know yet, but thanks for the hint.
>>30737 >waterproof >backside of the eye Okay I think I understand better. I'm more visual focused so the images overrode anything I read. So does that mean you would have an eye socket that is sealed with the head with the ace just plop in there though water can flow around the eyes it cant flow into the head? I actually have seen little water pumps that had the propellor controlled by magnets so I could see how that would work I suppose. Why physical cables over using a very short distance wireless signal if you're going that route? It would make it more durable and easier to remove the eyes. Potential interference and image quality loss and the issue of batteries for the camera(s)? What made you think of that way over simply putting a stationary transparent dome that is flush to the that moves under it? Transparent plastic does yellow over time though so that would have to be replaced after some years though.
>>30739 >eye socket that is sealed with the head with the ace just plop in there though water can flow around the eyes it cant flow into the head? Yes, if I can make it work. The eyeball would be in a cavity, the electro magnets would be on the other side of the "skin" inside of the cavity. The cables would be sealed with silicone, with a connector underneath and longer than necessary. Of course the cavity would need to hold them as well. It could be stretch cables or whatever, idk yet. So if I wanted to switch out the eyeball, the "bone" around the eye socket should be able to extend, like a ring that can be stretched to increase the diameter with one hand. Then I could pull the eyeball out, cut through the silicone around the connector and plug them out. New ones plugged in and then sealed and put back. The main goals and advantages over the common animatronic eye mechanism would be: - Head can be water proof - The eye movement would be more quiet Optionally there could be a servo under each eye but still under the other side of the boundary of the holding cavity, the servo would have a magnet attacked, rotating the eyes sideways. This might be an option, if they're supposed to only move a little bit. I'm concerned that they would move to fast and only in a few steps instead gradually. Not sure if this looks well. Alternatively, under the eyeball there could be a (permanent) magnet with polarity to make the eyeballs a bit floating, so the other magnets could move them easier. > though water can flow around the eyes it cant flow into the head? No and kinda yes. Only the cavities could fill up, but for that reason there would probably be a little tube opening on the bottom where the water could flow into the body in a safe way. Probably into the esophagus (/gullet). This would also allow for putting in a cleaning liquid. Or having a mechanism for tears coming from a little tank in the head, without having the eye sockets being filled permanently. >I actually have seen little water pumps that had the propellor controlled by magnets so I could see how that would work I suppose. Okay, but this is then likely a very week mechanism. >Why physical cables over using a very short distance wireless signal if you're going that route? Idk if I can trust wireless signals, also magnetism would interfere with it as well, I guess. Especially since the antennas would need to be exposed. > It would make it more durable and easier to remove the eyes. Maybe, but I don't thing by much. I would rather make the connector between the eyeballs and cables into something that can move around on it's own. >Potential interference and image quality loss and the issue of batteries for the camera(s)? I hope this can be mitigated by electroplating these parts and maybe adding some addition metal mesh inside. >What made you think of that way over simply putting a stationary transparent dome that is flush to the that moves under it? Idk, because my approach is closer to human eyes? This is actually worth contemplating, good idea. I don't think I want or could create transparent plastic spheres, but they might be available from some sources. Then the border would need to be sealed really well, and the more traditional mechanism could work inside. Though, not sure how much this could be trusted and how it would look like. Thinking about how to switch them out if they're old or there's a scratch would also necessary. >Transparent plastic does yellow over time though so that would have to be replaced after some years though. Yes.
Open file (181.00 KB 1570x987 eyesT1v1a-001.png)
Open file (225.74 KB 1048x706 eyesT1v1a-002.png)
Open file (180.64 KB 1048x706 eyesT1v1a-003.png)
Open file (21.56 KB 467x461 eyesT1v1a-004.png)
>>30747 I made a negative of how the backside of they eyeball is supposed to be allowed to move in the cavity. Creating this is quite compute intensive and slow on my laptop. I'm also not sure if I made a mistake, since it looks like there are areas which would be blocking. But it's just a test anyways. I had the idea to put little pivot spheres onto the eyeball (backside). On top and bottom. This is to reduce friction. Sideways looking would just require to rotate the eyeball on these, the "rails" are just to constrain it from rolling, but these might not even be necessary and I'll probably remove them. Of course the eye cavity part in which it moves will also need to be altered. Additionally this part also needs a whole for the cables.
Open file (30.15 KB 467x419 eyesT1v1a-005.png)
Open file (46.61 KB 467x419 eyesT1v1a-006.png)
Open file (60.21 KB 467x419 eyesT1v1a-008.png)
Open file (36.71 KB 467x419 eyesT1v1a-009.png)
Open file (69.60 KB 467x419 eyesT1v1a-007.png)
>>30758 >backside of an eyeball and eye sockets with cutouts to tilt or rotate them around without falling out I changed the design towards only using some small spheres as holders and pivots. The white parts in the picrels would be what's cut out from the holder behind, so it can move using 4 small spheres around the backside of the eyeball. Picrel one and two are with fewer cuts, just to show the principle. Not sure if it's going to work that way, I still have to try it out and maybe think more about it. But I can adjust the code relatively easy now if I think it's necessary.
I really like where your design research is going here, NoidoDev! Looking forward to seeing your results, Anon. Cheers. :^)
Open file (48.69 KB 666x822 skullT01v01a_001.png)
Open file (83.71 KB 666x822 skullT01v01a_003.png)
Open file (126.97 KB 666x822 skullT01v01a_002.png)
Started working on a skull in OpenScad with the aim of making it easily configurable based on the eye size and later maybe other features, by just changing a number like the size of the eyes and the rest should adapt to that change. Not sure if this is going to work. Such attempts tend to somewhat fail, since everything has to relate to each other. Either way, I'll need a human-like waifu skull. The model I'm using for reference is a female skull from Thingyverse. I will first make it relatively similar with as few parts as possible, and then increase the size of the eye sockets. It already looks kinda interesting.
Open file (53.36 KB 585x599 skullT01v01a_004.png)
Open file (111.16 KB 585x599 skullT01v01a_005.png)
Open file (116.90 KB 585x599 skullT01v01a_006.png)
Open file (118.50 KB 585x599 skullT01v01a_007.png)
>>31383 Working on it, one step at the time, but I wasted too much time with other stuff. Anyways, here the update: - Are we building bodies for aliens? - The nose is still to big. - Cheek bones still missing. - Chin is maybe still to big. - Generally a lot of things are necessary to do, but I can always use this as a placeholder.
Open file (56.40 KB 585x599 skullT01v01b_001.png)
Open file (120.57 KB 585x599 skullT01v01a_008.png)
Open file (130.53 KB 585x599 skullT01v01a_009.png)
Open file (69.23 KB 585x599 skullT01v01b_002.png)
Open file (114.54 KB 585x599 skullT01v01b_004.png)
>>31414 Further optimizations.
Open file (55.29 KB 685x599 skullT01v01c_001.png)
Open file (40.49 KB 649x641 skullT01v01c_003.png)
Open file (140.98 KB 649x641 skullT01v01c_002.png)
>>31415 Still on it.
Very cool progress, NoidoDev! Please keep up the great work. Cheers. :^)
>>31422 An early, fast and simple sketch to make a head fitting to a 3D face, admittingly not a good face in this case. The face is from 2D to 3D rendering, described here >>32770 and in the linked comments. Part of the idea is to make the head smaller while the 2D to 3D AI isn't willing to do that. Ideally we will also be able to use that to make an automatic face modelling AI, which would creating the right sized heads for the face and combine the parts.
>>31422 >>32793 I also worked a bit on a lower jaw with teeth, but this was just a test and I'm not sure if this is going anywhere. Might not be the right way, but could still be a placeholder.
>>32794 >Might not be the right way, but could still be a placeholder. I think it's fine at this stage, NoidoDev. Character modelling by hand like this is almost always a 'process of refinements', is it not? Cheers. :^)
Open file (3.28 MB 4624x3468 TummySwatches.jpg)
Prototyping soft TPU panels for a flexible tummy. Tests show that best aesthetics are achieved printing it upright. That it should be as thin as possible, and that a back which is an inversion of the front leads to the best detail and hand feel. Overall, I am not sure how to viably continue this research. It seems that the maximum height before significant distortion is 10cm. This is rather tiny.
>>33724 >It seems that the maximum height before significant distortion is 10cm. This is rather tiny. May be effectively a blessing in disguise, Kiwi. If we can keep our component swatch sizes smol, I think that will likely increase the number of Anons in the future (particularly low-income places) that can 'follow along in the fun!' with us here. Ideas? >=== -sp, minor edit
Edited last time by Chobitsu on 09/22/2024 (Sun) 04:04:27.
>>33726 Any ideas for connecting these pieces together? I've tried M3 screws but, TPU can stretch around the head.
>>33727 Yes. Print a series of interlocking tabs/flats (for tabs) on the backs of each panel. Print a base ring around her waist (for support) that the first row of interlocked panels go down into. Then progressively build up the next row of interlocked panels, etc. Make sense Kiwi?
>>33728 Could you provide some visual to go with your idea? What I imagine based on your response would still have the issue of the TPU parts stretching out of the connectors.
>>33730 >What I imagine based on your response would still have the issue of the TPU parts stretching out of the connectors. Yes, that would be an issue. I'd suggest experimenting with some non-panel test strips of tabs/flats, with a running series of matched sets that get progressively longer. Then you can just fiddle with them by hand to get a feel of what length will stay secure. Also, you might decide that each flat could use a leading 'guide slot' (kind of like a belt going through it's attached loop to stay snug) the tab could slide down into to keep it nicely secured into that matching flat. Also, you might decide to have a series of rings running up the (torso, say) to provide supports for the rows of panels both above and below. Kind of like Victorian hoop dresses were. As for the basic concept, there are lots of storage bins that have a series of interlocking tabs/flats on the two lid halves. When you close the two halves of the lid, they interlock together (one tab to one flat) and it provides passive rigidity. Also I can't post images while Robi has file posting blocked for Torfags (this is also why I can't bake new breads rn). Sorry. :P >=== -add 'guide slot' cmnt
Edited last time by Chobitsu on 09/22/2024 (Sun) 07:26:20.

Report/Delete/Moderation Forms
Delete
Report