/robowaifu/ - DIY Robot Wives

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

LynxChan updated to 2.5.7, let me know whether there are any issues (admin at j dot w).


Reports of my death have been greatly overestimiste.

Still trying to get done with some IRL work, but should be able to update some stuff soon.

#WEALWAYSWIN

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)


Welcome to /robowaifu/, the exotic AI tavern where intrepid adventurers gather to swap loot & old war stories...


Open file (1.12 MB 2391x2701 20210710_233134.jpg)
Open Simple Robot Maid (OSRM) Robowaifu Technician 07/11/2021 (Sun) 06:40:52 No.11446
Basic design for an open source low cost robowaifu maid. Currently attempting to make a maid that looks like Ilulu from Miss Kobayashi's Dragon Maid. Right now she's an RC car with two servo steering. Will share designs when they're a tad better. Ultimate goal is cute dragon maid waifu that rolls around and gently listens to you while holding things for you.
>>12033 Trouble is, to work with such small motors and fit all kind of things into such a small body. Think of the SBC and electronics alone. But good luck. One meter is more realistic.
>>12061 Not him, but you make a good point. OTOH, all engineering is a series of tradeoffs. Overall, situating components in a tight space is an easier problem to solve, particularly using CAD and 3D. But the laws of physics (and their indirectly corollaries such as the Square-Cube Law) are absolutely immovable. Extra volume is a more-achievable luxury one can enjoy and take advantage of, with the more focus that's spent on reducing robowaifu mass overall. OSMR-Dev Anon is already on a good path in this area, as he is trying to make a fusion of 3D-printed parts + lightweight & strong organic bamboo structural spars. Good combination, you ask me.
>>12064 >Overall, situating components in a tight space is an easier problem to solve, particularly using CAD and 3D. If there's no space, none of that helps. Very small means smaler options of movement, bc no space for motors. If the chest would be smaller than a Raspi Zero, it would also be the question what kind of computer would be used. Sumomo can go everywhere. In reality she needs at least a small SBC which connect to others, or she alternativly would only have electronics relaying data to some external system.
>>12074 >If there's no space, none of that helps. Fair enough, but I'd suggest an approach to solutions that doesn't involve anon going "It's impossible! Can't be done!11" It's can be fairly easy sometimes to jump to that kind of conclusion, particularly for something as monumentally complex as robowaifus. The primary points are: A) All engineering is a series of tradeoffs. Find a combination that is sufficient along several axis-of-need's simultaneously. Pick any two (but only two): cheap -- fast -- good B) The laws of physics cannot be violated. Period, end of story. Our job here is to try and find the ways to work alongside them and still build cool and useful robowaifus.
>>12087 Since we're a robotics concern here, I'll just clarify that that fast means 'fast time-to-market', not 'high-velocity'. That triangle is a business notion for engineers and managers.
Open file (1008.75 KB 4000x3000 IMG_20210801_201404.jpg)
Open file (946.16 KB 4000x3000 IMG_20210801_201414.jpg)
Open file (1.25 MB 4000x3000 IMG_20210801_201426.jpg)
>>12033 >>12046 Sumomo scale is an interesting challenge. Both >>12061 and >>12064 are correct depending on functionality. A fully functioning Sumomo is impossible. A minimally functioning robowaifu that resembles Sumomo is completely possible. Once a 1 m robowaifu is made, a 1.5 will come next but, a handheld waifu would be a fun future challenge. Currently the 1 m fox girl has a new chest and waist. They're designed so that a 10 mm cube will function as modular connectors. They are far from perfect but, pleasant in the hand. I've decided to shoot for good enough until an initial full body prototype is finished. I've wasted too much time and plastic trying for perfect. Good enough is enough for a prototype.
>>12095 It looks like you're already on your way with re-scaled parts. >I've decided to shoot for good enough until an initial full body prototype is finished >Good enough is enough for a prototype. That's the spirit that will lead you to success in the end Anon. Keep it up!
>>12095 Will those parts be mounted onto her wheeled base? Hopefully she wil still balance while moving about on her original wheels. But if you run into problems, I imagine at that scale a standard R/C car could be used as a mobile platform.
Open file (4.99 MB 4128x3096 20210803_162708.jpg)
>>12102 Thanks for your encouragement >>12108 She's being built with with legs that act as suspension, her waist is her base. I forgot her support tail so, for now she just topples over. Atleast the triangle support spars work.
>>12165 Excellent work Anon. Might I suggest you add two additional bamboo spars to your design? I recommend that you add two new ones + 5 connectors into the same spaceframe you've already built. One of these would cross fore-to-aft and affixed to the very center of the the two triangle 'bases' (that is, it starts from the top-front, and travels downwards and to the rear). And conversely, another one does just the opposite (that is, it starts from the top-rear, and travels downwards and to the fore), and affix it to the center of the two triangle 'points'. They will now form an 'X' shape where they cross in the center. Affix them each securely to these 4 new positions, and also to each other at their common intersection point in the very middle. To recap, that's five fixation points, for the two additional spars. I believe if you'll make this modification with these new, additional triangles in the spaceframe, you'll discover it will be well-rigidized, strengthened, and enhanced against torsional and bending stresses with very little mass addition. The tendency of fore & aft shear, in particular, will be significantly reduced. Look forward to seeing all the progress with your foxgrill robowaifu Anon!
>>12165 Great, but try using Imagepipe or something else to make your pics smaller.
>>12169 I'm definitely going to use that cross spar idea for the full size waifu. There will be 4 sets of spars and two sets of cross spars for the full scale waifu. I've remade her bust and butt to be easier and faster to print while being stronger. Good enough to scale up now. I'm moving forward to the full scale now and will make videos. I'm not good at writing so, I hope that will help spread the knowledge efficiently. I was thinking of posting to bitchute and YouTube. >>12046 I'll try to post some files to catbox soon, can't really say when exactly. >>12239 Is this size better?
>>12242 This looks wonderful Anon. I think you're going on the correct pathway to success in a number of ways at once. Primarily low mass, but also inexpensive construction techniques, and simplicity of construction. All these (and more) will turn out to be vital for widespread robowaifu kits in the end. Videos are a good idea thanks for paying attention to image size now, btw in my opinion. Be aware you'll probably eventually cancelled by Google for it though. Also, I'd suggest you don't use accounts either there or elsewhere that can easily be associated with your real identity by stronk, independynt amateur """investigators""". BTW, I think your writing is just fine Anon. Just apply yourself further. Look forward to your robowaifu getting her tail and motors soon! Good luck.
>>12242 This is shaping up to be a very original robowaifu, anon. The lightweight construction should make it easy to have her zipping about at high speed! With the wheels for feet she reminds me a bit of Motorball Alita.
>>12298 >With the wheels for feet she reminds me a bit of Motorball Alita. Neat thinking. Now, will Motorball have a division for cute foxgrill meido waifus I wonder?
>>12243 Through the use of a .8 mm nozzle, I was actually able to print 1.5 m scale chest and waist parts at the same mass as 1 m scaled parts. Thicker walls enable huge gains in strength to weight ratios in prints. Unfortunately, the support structure of cross spars didn't work because my bamboo is too springy. Would have worked if the spars were supported in the middle per this anons suggestion >>12169. Laying on it showed it to be surprisingly strong and resilient given the sub 300 gram mass for the entire structure. >>12298 >>12308 >Foxgrill meido Motorball Kek, once we make DIY instructions and kits, it would be great to have anons run their own robowaifu Motorball competitions.
>>12428 So, mind spelling out what's going on here lad? Is this an 'after' shot of a 'test-to-failure' strength test or what? How did it do BTW? Do you have a 'before' shot? Kind of hard to tell the exact design before the test. Looks like you have entirely new system with entirely new parts? Regardless, sounds like you are making steady progress, and that generally it's good news overall. Hope so lad, keep at it!
>>12444 >>12444 Unfortunately there is no before shot. The inserts that held the bamboo couldn't take take the pressure when tested. I've redesigned it to be simpler with less spars that are more strategically placed for a balance of strength and weight. I redesigned her chest and waist slightly to be easier to print faster. I want to fail faster to reach the goal faster. Unfortunately, it still takes a whole day to print a 1.5 m scale chest and waist. Now it's onward to designing suspension legs that work.
>>12463 Thanks for the status update. Good luck with the legs. We've talked before about those swing-arm lamps as a possible basic frame design for the legs, maybe you can have a look at those if you haven't yet? (>>961, ...)
>>12027 >Screw it, I'm building a 1 m fox girl.
>>12464 Another step closer. With a head, she would be close to 1.5 m. My bamboo isn't long enough for properly scaled legs. Unfortunate, she's super floppy and hard to balance. She needs rigidity at this scale that 3 mm thick bamboo just won't provide unless many are used, which nullifies their weight advantage while increasing construction complexity. Any low cost alternative ideas?
>>12485 Excellent progress Anon. >Any low cost alternative ideas? Use triangles. For example run two cross-braces on the thighs each (one on the front, one the other angle on the back). Repeat for the tail's joints, and the lower leg's joints. So for the front thighs, you'd run one diagonally from the inner crotch, down the thigh, then attached to the outer knee. Do the exact opposite angle (outer butt to inner knee) on the backside. Mirror on the opposite side (the other way round). Do the same thing for the other 4 remaining 'limb' parts. That should help with individual limb parts rigidity. You now can think about stability (not the same issue). For that, you can think about angling the two legs outward slightly to form more of a 'bow-legged' tripod form in space. Slightly awkward ATM, but in the future as you begin to turn her into a real robowaifu, she'll have active balancing mechanisms available -- same as we humans do. Those stabilizing gyroscope cubes we've talked about here for instance, might be a feasible way to improve her balance and stability. (>>5645, >>9057) Lastly, as you move forward with her design, attempt to locate as many high-mass items (batteries & motors for example) inside her pelvic volume. Not in the higher, upper volume of the chest/head. I believe you'll quickly discover this engineering approach will assist you in many ways with your designs, not the least of which will be the balancing your robowaifu.
>>12485 BTW, any chance you can place some 'common' item within the shots of your robowaifu to assist us in judging her scale Anon. A box of cereal, or a gallon milk jug for instance, might be helpful to improving our ability to perceive her size. Or whatever else you have handy that you think would look better than either of these two choices. Regardless, you're making solid progress for your stated goals ITT. Good luck with her!
>>12486 The double cross does help but, even with added stabilizers, it's just not enough. It's a big let down given how well bamboo skewers work on very small scales. I actually do plan to use leg geometry to help with balance and stability down the line. Because she's designed to roll around, I'm going to try to localize as much mass as possible into her legs to keep her naturally stable. Active stability is a waste of watts if the problem can be solved by design. I'm obsessed with efficiency and reducing mass and energy consumption. Of course she'll need active stabilization but, the less it's needed, the better. >>12487 Cheerios for scale, thanks for the luck!
>>12495 As the old saying goes, 'Success is 99% percent perspiration, and 1% genius'. Please don't let a minor design setback discourage you. You're on the right track, I'd say Anon.
>>12485 >Any low cost alternative ideas? Would some thin ropes, additional to the bamboo, help? Tensigrity? The other thing which was discussed here before were using a bundle of drinking staws. In that case you could also consider flexible plastic or silicone tubes for water. Then there's also the approach Poppy Project uses, a mesh on the outside, which could be flexible to pressure if you want soft thighs. In general, I was assuming you would add something in between the bamboo anyways, so think about using that to create tensigrity and as a constraint for the bamboo moving around. In that regard, what about trying to use rubber bands between the bamboo, to constrain the movements? Or it's generally the wrong approach: Even thin metal tubes for broom sticks or thinner ones from a hardware store, or plastic water tubes, the aren't really very heavy.
Open file (193.48 KB 960x540 01.jpg)
Open file (190.58 KB 960x540 02.jpg)
Open file (142.85 KB 960x540 03.jpg)
Open file (370.04 KB 960x540 04.jpg)
>>12495 With all due respect Anon, and again, I applaud your efforts; but quite frankly, you aren't following my advice just yet. I mentioned using triangles. So far, you aren't really doing so. Compare: > #1 #2 with > #3 #4 There's a difference. Triangles always have 3 edges, and their common vertices are always localized together into common points. Triangles have a unique geometric characteristic that allows them to form effective space-frames. Try using using as many strictly-defined triangles as possible, and I'm sure you'll see a marked improvement in your waifu's frame rigidity. OTOH, you did produce one example that was composed of 4 interlocking triangles (>>11613). I bet that one wasn't floppy or wobbly, right? Good luck Anon. You have some great ideas, and especially >"I'm obsessed with efficiency and reducing mass and energy consumption. Of course she'll need active stabilization but, the less it's needed, the better." is right on track. Keep going Anon, you can do it! :^)
>>12517 Ah, right. Like the Poppy Project I mentioned, but of course he could do it further inside instead around the whole outside of the leg.
>>12517 >and their common vertices are always localized together into three (3) common points. * I felt I should probably clarify that, since it might not be perfectly clear. So, 3 edges, three vertices (two end-points each) forming 3 points in space. Note that you can obtain a couple of triangles for 'free' with two cross-spars, but only if you rigidly fix their common intersection into a fixed point. Otherwise, you only have two cross-spars and no triangles.
>>12517 The cross spar arrangement makes double triangles under tension. Having two at 90 degrees with overlap actually produces greater rigidity then pure triangles. The bamboo is strong enough in this configuration for 1 m scale. At 1.5 m scale, the bamboos just becomes too flimsy. Pure triangles are good at lower scales like .7 m or below, which the first chest and waist were. I'm moving towards cedar dowels at 6.25 mm diameter for greater rigidity. Triangles should be ideal for those rods and the next 1.5 m scale prototype will use a space frame with tensegrity principles per >>12513 's suggestion. >>12506 Thanks for the encouragement Chobitsu, I'm probably going to fail a lot more but, that Tesla bot really lit a fire. We need to beat Tesla to market.
Open file (3.58 MB 4128x3096 20210821_101849.jpg)
>>12519 I actually did rigidly the intersections but, the braces break and pop off when put under any strain. PLA couldn't handle it and string just moves around. It feels incredibly sturdy until they pop
>>12518 Yep. The Poppy-Project team (wisely) adopted a variation of the so-called 'box girder' trussed space frame for Poppy's armature. It's a good & lightweight design that we Anon's would be smart to steal mimic...
>>12521 Then strengthen the joint, obviously. This is how engineering goes Anon, but 'keep throwing stuff against the wall'. Eventually something will stick.
>>12520 You're quite welcome. You have our heartfelt encouragement for your project Anon. Godspeed OSRM, may she give all of us good insight and wise design councils. I added a Tensegrity waifusearch to the Library on your thread's behalf OP. >>12524 >=== -minor prose edit
Edited last time by Chobitsu on 08/21/2021 (Sat) 18:44:21.
Open file (600.13 KB 2138x1318 g.png)
>>12522 >Poppy Looks interesting, thanks for the heads-up >>12525 Neat, thanks Any thoughts on the Tesla Bot design? I'm considering a cloth outer covering like Tesla is using. Feels like a waste of mass and cost but, the aesthetic value may be worthwhile
>>12529 6.3 mm (1/4 inch) birch dowels have phenomenal strength and rigidity for their mass. With these, I am very confident in the future of this project! KitsuMeido unit 1 as soon as I can build her. Also, really need help with the face
>>12547 Sweet looks good Anon, sounds like you found a good stock material for your goals! >Also, really need help with the face Can you spell out in more details what you mean please?
Open file (17.96 KB 710x473 v1.jpg)
>>12548 I have the basic design down and somewhat tested, ready to go for the actual design. The major problem is, she doesn't have a head. I'm not sure how to give her a face that will work. I originally used a print out of Ilulu but, it feels weird.
>>12549 I see. SophieDev just used an exported game asset mesh for his beloved Elfdroid Sophie robowaifu Anon. He's freely posted the STL files, etc. for Sophie, why not just start there? She's very cute and at the very least it's a good place to start I'd say. >"I am currently printing a bunch of smaller bits and pieces for the new head - ears, new bolting plates, electronics mounts. Once that's all assembled I think it will be time for me to upload her parts onto some 3D printing websites and try to spread the robo-love ;D" (>>9863) SophieDev are you around? What do you yourself think of Anon using Elfdroid Sophie's face?
>>12554 Using Sophie's face is certainly one option. Ofc if you can learn to use Noesis and Ninjaripper then you can export a much wider variety of heads from either Honey Select or Honey Select 2. One thing that is especially good nowadays is that those games come in ready-to-install packages (check out 'ScrewThisNoise' repacks). Originally I spent hours setting up Japanese Honey Select, downloading fixes, expansions, mods, user tools, translations etc. But now it's all-in-one! If you don't want to go the game asset-rip to 3D-printing route, I suggest purchasing a Dollfie head (or a head from one of the many Volks line of dolls) or maybe the head from another fashion doll line such as Smartdoll? Although many of these will be 1/3 scale at most. That's a big problem with most non 3D-printed doll heads. Many of them are just too small to be useful.
Open file (631.50 KB 3379x1287 20210823_185920.jpg)
>>12566 >rip a game asset for her face That's not a bad idea. I've been doing some doodlengineering and looking through other threads. I think giving her a gravity compensation spring for her legs and a passive mass dampener for her torso could help with efficient stability. Also, I came up with an artificial muscle idea, a dual ended differential windlass that uses pulleys for tendons, the pulleys would push and pull, just like agonist and protagonist muscles. This give great gear ratios without the weight or cost of gears.
>>12521 TPU? Nylon might also work with such low hights and no enclosure.
>>12589 Belts can certainly work if the right size tensioner is used (see James Bruton's "Open Dog" project and many of his others). The main worry I have with any kind of pulley system is keeping it from collapsing or slipping. Since a pulley usually relies on some kind of rope or belt and the mass of the pulled object + gravity for tension. Initially I tried using a belt system in my robot arms, but it kept slipping just with the weight of the forearm and hand attached. Which is why I eventually opted for pair of fairly large helical gears at the elbows, instead. Still, I wish you eventual success with your chosen mechanism. Your efforts to push the boundaries of practical robowaifu development are admirable!
Open file (154.75 KB 1024x768 Blog.jpg)
>>12614 >belts Nah mate, it's two coaxial differential windlasses. It functions by letting out string equal to the difference in drum circumference on one end while taking in the same length from the other end. This allows for cheap high ratio linear actuators. They're also very efficient and light weight. This demo actuator though not very useful, has a sub 10 gram total mass. I've been considering between motors right now I'm currently on 130m motors.. Normal brushed are cheap as dirt. Can get large quantities for 30 to 60 cents each easily, their design lends them to easy integration into designs (lowers manufacturing complexity, time, and reduces faulty parts). The main problem is the abysmal 40 percent efficiency.
>>12654 This is a good idea. I'd suggest you add a small & simple gearbox between the motor and the spindle to increase torque. My windlass designs are involving 'silk' ribbons (usually a synthetic instead) as they are particularly good for this. Good luck anon.
>>12656 Please do share your design. I'm still actively developing the windlass design to optimize mass cost and manufactuability
>>12549 Get a Vroid Studio model, import it into Blender, and just get the head? If you want the most anime head without the trouble of modeling from scratch.
>>13007 Sounds like a good idea. We also have a thread for faces >>9 and looking in Youtube gets you videos for how to model an anime head in Blender: https://youtu.be/fQZ7TEdFdMU also you could probably use Sofies head. In all cases printing might be difficult.
Open file (3.25 MB 4128x3096 20210911_133952.jpg)
>>13007 I've tried but, the internal geometry of vroid studio model heads means they'd take a day to print and would be heavy. >>13020 That's a great video tutorial. Hopefully I can figure out a good head that's printable. I've been working on her drive. I simplified things down to two drive motors because servo based steering didn't work well. Unfortunately, her legs drove off her body so, work still needs to be done there.
>>13020 I'm surprised by that video Anon. His approach is very methodical, but it's not bad as a beginning and he got to it in a short timeframe. Also, focusing on low-poly at the front of a project as he is doing I consider to be vitally important for success. Thanks!

Report/Delete/Moderation Forms
Delete
Report

no cookies?