/robowaifu/ - DIY Robot Wives

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

Roadmap: file restoration script within a few days, Final Solution alpha in a couple weeks.

Sorry for not being around for so long, will start getting back to it soon.

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)


Robot skeletons and armatures Robowaifu Technician 09/13/2019 (Fri) 11:26:51 No.200
What are the best designs and materials for creating a skeleton/framework for a mobile, life-sized gynoid robot?
Guy is a master at creating hinges using wooden dowelling and cardstock! Particularly impressive are the photos where some of her joints appear to be holding a pose. Floppy, loose joints are a big challenge to overcome when designing a doll...and to pull it off using such materials is amazing! (The movable "breasts" are a nice idea...although I can't say I'm a fan of those spikey nipples).
>>8071 >Guy is a master at creating hinges using wooden dowelling and cardstock More details? I'm not quite sure what you are referring to Anon.
>>8070 She definitely needs a can of spray-paint, but I think it would look quite beautiful after a little preparation. >>8071 is right about those nipples, though. What about the design looks unreasonable? It has three distinct layers, according to the author, and offers great posing options and articulation. It's great for the people designing it because the parts are easily sorted, and it's great as an end product because of the relatively few resources required to make it. I'm not saying we should use paper or anything, but a plastic equivalent should be quite cheap. Honestly, I'm thinking about picking up a 3D Printing software just to try to make these parts on my own. Not that I would know where to start...
>>8071 I don't think he even used dowelling. If you look close, the rods appear to be made of rolled up paper hardened into a plastic-like substance. I bet he used a doll of some sort in order to shape his pieces. He did make some improvements to the base shape. Look at the feet, for example. The bottommost layer of her sole extends past the foot, restricting the movement of the doll to a reasonable degree. Just brilliant!
>>8083 >Not that I would know where to start... If I am ever struggling with how to start designing a part then I literally break the complicated shape up into smaller, simpler 3D shapes. Like a hand becomes a load of cuboids and cylinders. Any fancy or complicated rounding, bevelling and chamfering I leave until last.
>>8083 >I'm not saying we should use paper or anything, Actually, not only do I think that's not a bad idea as an underlying form (along with thin foam) upon which to build up layers of hard shell over it >For a real life example of this design look at how modern light weight bike helmets are made. >>7980 but I'm actually using cardstock in my preliminary design work now. Over a lightweight tubular tensegrity space-framed chassis, paper makes an excellent type of light, cheap, and easily modifiable underlying substrate. Very quick, cheap and easy to mock up design shapes, and then to use it to build up the hard shell over. >(BTW, we have a papercraft waifu thread >>271 , though it hasn't been repopulated yet after the migration here).
If you are okay how this design looks (>>8062), then you should consider yourselves glad, since designing a body shape like this is probably much easier than designing human-like bones and a mold for a silicone body. On the other hand, if you want to use cardboard or paper because you can't afford a printer and the 12-14$ for a kilo of plastics, then that's another story. >>8083 >>414 is the thread for CAD. That basics are easier to learn than you might think, just a bit of frustration tolerance might be necessary, since some errors and problems might be confusing at the beginning. It also does not need a great computer, Solvespace for example runs on a Raspi3 and FreeCAD is also in the repository. Sculpting needs more resources, which is one reason more not to confuse those two. You''ll probably going to need that for the face.
>>8093 >On the other hand, if you want to use cardboard or paper because you can't afford a printer and the 12-14$ for a kilo of plastics, then that's another story. While I myself can manage that I have both already, I'm explicitly spending time years now, lol thinking through and working on kit designs that will be highly inexpensive and easy to assemble even for kids. This is how robowaifus will jailbreak the system, and escape the globohomo corporate stranglehold and escape into the free world. Everyone here do as you see fit, ofc, but I'd advise all of us to at the least consider these issues. Thanks for the good advice btw, Anon.
>>8096 >kit designs that will be highly inexpensive and easy to assemble even for kids I do consider this, nearly put it into the other comment, but left it out at the end. I also think that it would be great to have something in case our technological society has some breakdown or something for all the very poor men in poor countries. The later doesn't even need to be driven by compassion for those men, but by the hope to lower the birthrates in such places (while "middle-class people" in wealthier countries might even have more kids again, thanks to more advanced robowaifus and -nannies). However, I think it is highly unlikely that plastics and mass production of parts will go away anytime soon, currently even poor people in poor countries are flooded with them.
>>8098 Good points Anon. I have to agree with them all tbh.
Open file (54.62 KB 510x680 Eqr7t-tVoAEgBDF.jpg)
>>7727 >>7729 On his page https://twpf.jp/hamcat_mqq he writes in Japanese that he might sell the project one day maybe, so I don't think it is open source and probably never will be. (I copied the text into Google Translate). He also uses zBrush for his doll molds, not sure if he made the skeleton also with it. It would make it much less interesting anyways, since this program is for sculpting for all I know. I think it's better to use the project as inspiration, we don't need the files but the skills to make our own in parametric way. He also wrote that he need circa three years from zero to making his doll.
>>8156 Thank you for taking the time to look into these questions Anon. While I can't blame him for wanting to keep this work proprietary, I had rather hoped he would make it opensource. Oh well. Yes, you're right that we all need to work on our own skills. And that was 3 years well-spent developing his skills. I'd like to see us do the same here.
Open file (355.92 KB 1064x554 Selection_266.png)
Open file (143.86 KB 889x670 3DPrintRobotPage3.jpg)
I really like the space-frame design of this guy's pretty large hexapod. I'm personally convinced that this type of low-density good-rigidity structural approach (similar to tensegrity designs) will be vital to creating relatively inexpensive robowaifus in the end. https://invidious.snopyta.org/watch?v=SXGR5NzBNu8
>>8158 I dont see it. Maybe as a pattern for TPU (flexible plastics), to hold muscles and silicone parts in place under the skin. Bones should be on the inside, and feel as human as possibe. Also, even metal rods are quite cheap.
>>8166 >quite cheap Well, cost is only part of the full calculus (not that I pretend I understand all that just yet) but rather to my thinking the basic issue is simply the laws of physics. Mass is everything. I went into this in detail here: >>4313 But honestly, it's only one consideration among others, even if it's the most important one. Structural rigidity, speed & agility, graceful motion. All of these also come into play for highly appealing bipedal robowaifus. Structural frames bring the same benefits to the interior frames of a robowaifu that they do to architectural designs or box-girder bridges or other spanned structures. Namely, an efficient combination of strength and mass. These things all matter, and low mass (ie, low density) frames are the key to everything else.
Open file (528.99 KB 2048x1481 humanoid-skating.jpg)
>>8167 BTW, the designers of the poppy project also understand this basic point too. >>4436
>>8167 Thin metal tubes or rods are quit light. The material around could be light, e.g. foam and the frame could be fiber enforced silicone rubber. Holding it together, but not too rigid. It's probably more a matter of preference, more doll- and human-like or not. >>8168 Low-cost and Poppy-Project... It's rather about 3d print some parts and then spend all the money on servos instead printing the gearboxes and using cheap motors.
>>8169 Sure ofc, preference in design and aesthetics certainly plays an important part in any design engineer's work. Wouldn't have it any other way, right? :^) Yes, cost of servos is certainly an issue, thus why I asked in the other post if anyone had played around with that design using cheaper actuators yet. There's a really delicate balancing act between bulk/low-cost & compact/high-cost. This is true in practically all engineering however, so it's nothing new to our efforts. Smaller generally means more expensive, power & other constraints being nearly equal. It's a fine line we're all trying to walk here to manage to create systems that are both capable, pleasing aesthetically, and inexpensive all in one go. Some might be skeptical because it sounds like we're trying to have our cake and eat it to, and it's well-established that all engineering is a series of trade-offs. Quality/Performance/Price, choose any two.
Open file (63.92 KB 1080x1782 IMG_20210123_023213.jpg)
Open file (22.97 KB 508x508 IMG_20210123_023236.jpg)
Open file (72.36 KB 576x1024 IMG_20210202_063843.jpg)
Close shots from Alita's design. I like that some parts have layers. I imagine this to be a good idea, e.g. to run current though some parts instead of using cables. Then the layers need to be separated. Same, if the metal is for rigidity, but other layers contain electronics e.g. sensors.
Open file (71.29 KB 576x1024 IMG_20210202_063833.jpg)
Open file (239.20 KB 1152x2048 IMG_20210202_063900.jpg)
Open file (261.49 KB 1152x2048 IMG_20210202_063856.jpg)
>>8401 >>8402 Neat! Thanks Anon I've always been looking forward to some detailed images of the design work that went into the more recent Alita movie. Please dump everything you find somewhere here.
Open file (112.59 KB 576x1024 IMG_20210202_111218.jpg)
Open file (103.08 KB 576x1024 IMG_20210202_111223.jpg)
Open file (173.82 KB 1152x2048 IMG_20210202_111227.jpg)
Open file (49.17 KB 576x1024 IMG_20210202_111232.jpg)
>>8406 There actually was more. Only found this bc I went back to take a peek into Twitter, which I try not to do.
Open file (101.53 KB 576x1024 IMG_20210202_111240.jpg)
Open file (80.94 KB 576x1024 IMG_20210202_111244.jpg)
Open file (74.33 KB 576x1024 IMG_20210202_111248.jpg)
Open file (74.73 KB 576x1024 IMG_20210202_111252.jpg)
>>8412 Ah, that's fine! Thanks Anon, appreciated. I'd really like to try my hand at papercrafting a shell for my robowaifu that has some similar characteristics. Once I get it right then I imagine I should be able to use that to create some hard resin forms from it. It still has to be super lightweight to work well however, or everything else involved in moving and supporting the robowaifu becomes heavier and more expensive. But I really like some of these designs so it's worth the extra effort to try something nicer.
i'm only starting to get into the technical side of the robowaifu idea, but this design (https://www.youtube.com/watch?v=aLaqMreVj9o) should be fairly simple to integrate especially when driving it most of the motors don't have to be built into the joints themselves
>>8458 Thanks Anon! Yes, that is a really clever set of designs that team are using. Keeping the mass of the actuators themselves out of the limbs themselves may prove to be the key to giving us all fluid, smoothly-moving robowaifus. Please drop more interesting things here for us.
>>8461 gladly :3 of course everything i'll find will be shared as to further the good cause! a friend of mine, who works with orthopedics showed this to me, he has a project idea i'd like to help him with to make affordable opensource prosthetics, i'm sure i'll learn a good deal to make those useful for this kind of robotics >>8461
>>8466 Sounds good. BTW, we have a prosthetics thread here >>417 . Please post any good things you know of there, as that area definitely can provide good inspiration for ever more lifelike robowaifus, etc.
I post the link here, to the files for a mount (connector) to a motor which has none and a tubeholder for a round skeleton e.g. steel or aluminium tube. I wasn't sure if here was the right place or if the actuator thread was better. However, this is more about integration of a motor into the skeleton than about the motor itself. Not sure if the PM35S even works for this project here, since it's AC and needs a special driver if it even takes DC power. I only bought my two ones spontaneous because they were very cheap and I wanted to have something for playing around. However: The whole construction could be adapted to similar designs, of course. Basically the motor has a cogwheel, but if something is connected to it, it can fall off sideways. This is meant to solve it. Might come up with other motors as well. So here it is: https://files.catbox.moe/76t3vo.gz - nothing special, pics related. I used tar.gz for compression, not sure if zip would be better (more common).
>>8584 >and a tubeholder for a round skeleton e.g. steel or aluminium tube. What about other 'tubes' like wooden dowels, or something like PVC tubing sections? Do you think your tubeholder joints would work OK in that way? >not sure if zip would be better I think any choice is fine, do what's more convenient for yourself. I'd guess that every regular here is proficient enough to use any type of compression approach. Everything looks fine on my end, BTW. >
Open file (170.41 KB 1920x1016 tubeholder_mount.png)
>>8589 >wooden dowels, or something like PVC tubing sections Yes, sure. I had plastic tubing in mind first. Wood might be a bit heavy. The hole in the upper part might need to be fitted, which is one value in the Solvespace file, more if the whole construction has to change some sizes like thickness of the walls. This is just the current state, e.g. I forgot that I wanted some holes on the side to put in some screws. Right now, only friction or other improvisations would hold it in place. Generally, the idea is to use either tubes, which should work for fast first prototyping of a skeleton. We could also use the approach beyond that, especially for combing metal with plastics. With tubes we can also use the inside, but adding printed parts around. Wherever one can feel the bone through the skin, these parts could be designed human-like to make it feel like it. In other places where the bone is more hidden under tissue, it could be bigger than a human one. Alternatively, thinner metal rods could replace the tubes later, leaving more space for printed parts. >PVC Try to use another plastic if possible. This stuff is dangerous when heated. >attached picture This is not how to print it! Just for showing how it is supposed to fit.
>>8596 >Wherever one can feel the bone through the skin, these parts could be designed human-like to make it feel like it. In other places where the bone is more hidden under tissue, it could be bigger than a human one. Great ideas. >>PVC >Try to use another plastic if possible. This stuff is dangerous when heated. Thanks for the warning. I think there is some around the house unused, but I was more about the idea rather than specifics. I'll look for alternate resources Anon.
This here, by hamcat_mqq is of course skeleton related: >>7709
>>8602 I'm looking into the topic of ball-joints right now. Including how to design and print them >>8585. I would also never rule out buying stock parts which are used somewhere, like cars or smaller vehicles. We're her to build robots, not BJ-dolls, but in some places this might be useful, starting with fast prototyping of a skeleton and then having a talking waifu standing around for motivation. Jumping into the rabitt hole of joints, made me find much more which could be useful, some familiar, but often I also forgot about them. Regular ball-joints out of steel are used in every automobile, and in many other places, so these might even be available for cheap. Maybe such ones will be useful around the shoulders or the neck, for example. https://en.wikipedia.org/wiki/Ball_joint - site includes a description of spherical rolling joint from picture one, which looks cool but might not be very useful. >rod end bearing, aka heim joint (N. America) or rose joint (U.K. and elsewhere): https://en.wikipedia.org/wiki/Rod_end_bearing >The heim joint's advantage is that the ball insert permits the rod or bolt passing through it to be misaligned to a limited degree (an angle other than 90 degrees) >If spacing is critical, female heim joints are able to be threaded on, instead of welding inserts to the shaft. When dealing with aluminium shafts, the easiest way to use heim joints is to use the female heim joint. >Light weight is a key factors when building competitive robots, so using aluminum rods and female heim joints can be key. >shifter of motorcycles. The shifting mechanism allows forces to be applied linear, but still be able to work at angles when in different gears. universal joint (universal coupling, U-joint, Cardan joint, Spicer or Hardy Spicer joint, or Hooke's joint): https://en.wikipedia.org/wiki/Universal_joint >commonly used in shafts that transmit rotary motion. Then we get to the more human stuff, which was already mentioned here before (pic 5): > 1: Ball and socket joint; 2: Condyloid joint (Ellipsoid); 3: Saddle joint; 4 Hinge joint; 5: Pivot joint; or in German: 1 Kugelgelenk, 2 Eigelenk, 3 Sattelgelenk, 4 Scharniergelenk, 5 Zapfengelenk > The ball and socket joint (or spheroid joint) is a type of synovial joint in which the ball-shaped surface of one rounded bone fits into the cup-like depression of another bone. The distal bone is capable of motion around an indefinite number of axes, which have one common center. This enables the joint to move in many directions. >An enarthrosis is a special kind of spheroidal joint in which the socket covers the sphere beyond its equator. Related: >>7588 and answers, >>7625 and answers How coupling is boing done in other machines should also be a topic of study here, since this might be useful, e.g. making sytems more flexible or safer: https://en.wikipedia.org/wiki/Coupling
>>8611 Of course I had to forget something: Link to Ball-and-socket joint (related to pic 5): https://en.wikipedia.org/wiki/Ball-and-socket_joint
>>8611 >starting with fast prototyping of a skeleton and then having a talking waifu standing around for motivation. I think this is a valid approach actually. Inspiration and motivation are important to us here. An articulated, life-sized BJD skeleton (especially one you are designing and building yourself) would be a good learning experience at the very least. And couple that with speech it would be a nice start at things. Thanks for all the joint images Anon. It's an important engineering topic obviously.
>>8584 Just realized, the folder in this file is named Feb2020 like this was a year old. No, my mind is just stuck in 2020.
>>8611 I've been trying to think of ways to mimic human hip joints using inexpensive and 3D-printed parts. I really like the MIT Mini Cheetah form-factor actuators >>4890 (>>8415) . They seems to have a good combination of power efficiency & mass, as well as being in a particularly well-suited form factor for locating down in the lower pelvis edges to drive hip joints. I wonder if anyone has any good suggestion about the kind of coupling/joints that would allow for something the Anon's dream of a ballroom dancing/ballerina robowaifu? >>7889 I think it would be absolutely revolutionary if we could build lithe & limber, dancing robowaifus someday.
>>8641 Here's a search, maybe it will help >>8642 .
>>8641 Make them walk and dance is one step after the other. Making them dance while holding a pole or something with both feet on the ground will already be something. The hip-joints are ball-and-socket-joints like here https://en.wikipedia.org/wiki/Ball-and-socket_joint - I once saw some kind of ball out of ceramics, which could be screwed onto something. Doorknob or something alike. This might be a method. On the other hand, we need a way to move current and maybe light between the extremities, for the electronics, but also communication. So we'll need to think about putting this into th joint. We'll need to add some contacts in a way that it can transmit while moving.
>>8644 >We'll need to add some contacts in a way that it can transmit while moving What I reckon would work would be to have 2 or 3 small-diameter, flexible conduits fastened through grommets in the hip 'bone', and then also down on the thighbone. Something similar up at the shoulders as well. Nothing that sticks out too much, and also flexes and bends with the ball-and-socket joint's movements. The wiring, Bowden cables, optical fibres and suchlike would ride down through these conduits around the joint areas.
Open file (237.75 KB 187x272 Hexapod_general_Anim.gif)
>>8644 Found this through your links Anon. Seems like it might be a useful approach where we can print out several linear actuators, then use them to manipulate plate-force areas for different angling degrees. During swaying/rocking motion while our robowaifus are walking or dancing with us, for instance. > Seems like a lower plate the Cheetah motor was affixed to on the inside for example, could be tilted to different basic angles during her gait by this type mechanism. The hip/pelvis volume could have two sets of these mechanisms operating independently. We'd have to be careful to leave volume for the robowaifu's vagoo down there ofc. https://en.wikipedia.org/wiki/Stewart_platform
>>8648 That looks like it would be useful for a waist (with a powerful motor on the top disk to allow the upper body to rotate 360 degrees). Although it's length may prevent legs being attached. But that's not a problem if your robowaifu is modular and can swap between bases - wheels and legs.
>>8650 Good observation, you're right Anon.
>>8650 Hmm, you know it occurs to me you could shape both plates to be more oval that the one in that example .gif. If you sort of designed both parts to be more like thin slice cross-sections of a female abdomen then you could have a Stewart platform that could serve pretty nicely as the base for a robowaifu's upper body. One like our Sophie here, instance. Regardless of what someone does for the hips downwards, fixed-base, wheeled buggy, actual legs, w/e, this section in her middle would work pretty nicely. Bright idea, lad. :^)
>>8656 >One like our Sophie here, for instance.
>>8646 Thanks, considering. But my hope is to integrate as much as possible into the bone, so there would be no bending parts. Not sure if I'll get there, had some thoughts on it today, while playing around with my printed ball-joints. >>8648 Interesting. I didn't really think about it in detail. I want to use motors for some muscles, soft muscles for others. I also thought about using springs or bumbers to make certain repetitive movements using low amounts of energy. Currently I dont know how to insert a bit of randomness there. The body should go down automatically if relaxed, and bumbing back if the spring is unlocked. I'll need a mechanism to change the traits of the spring, probably some mechanism that blocks it gradually. >>8656 Yes, seems to be the way. I didn't think of it as steward plattform, but this seems to be a good way to have a mental model of it. However, all of this is more for the hard shell body type like Sophie. I'd like to try move it via tha spine, more like a human.
>>8660 >I'd like to try move it via tha spine, more like a human. Yep, I think that ultimately that would be the most lifelike ideal for a robowaifu. We'll all get there with eventually I'm sure.
I posted something about DARA project, a guy developing a hand for his robot here >>8700. It's a male bot (like Data), but we might learn from it. He is also working on other parts, like feet, which is rather rare. Could have gone into the thread for bipedal locomotion >>237, but I think its more general so I put it in here: Foot, short clip: https://youtu.be/OBsXwv7m1hY[Embed] Tilt slider for leg: https://youtu.be/R9Wdc0H4iPs[Embed] Same, but more info: https://youtu.be/DW6PW5yQr4o[Embed] Toes, short clip: https://youtu.be/AbhPPy1ioXk[Embed] Middle toes V2: https://youtu.be/i7DZusCGz7k[Embed] There's more: https://youtube.com/c/DARARobot
>>8704 That's fine, I'll just xlink it here.

Report/Delete/Moderation Forms
Delete
Report

Captcha (required for reports)

no cookies?