/robowaifu/ - DIY Robot Wives

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

Server and LynxChan upgrade done. Canary update and a general address soon. -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)


Remember, /agdg/ 's Game Jam is still going on. Go Team! :D


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.
>>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!
After many failures trying to get her to balance, I'm going to try wider bigger wheels.
Open file (95.42 KB 1000x750 _FRNLZLakBt.jpg)
>>13569 Have you tried a wider pyramid shape? .
>>13570 Thanks for the suggestion. I'll try increasing her grounded surface area.
Open file (48.61 KB 752x564 6hxqa8t4y0831.jpg)
>>13582 >Also, since you're having problems too, want to team up on OSRM? I already have a decent chest and pelvis. I've been stuck for months since and you seem to have good ideas. 1: where are you stuck? 2: is this like pic related? I guess we could start with a clear mission statement. What functions does it need to perform, what are the aesthetic and design limitations or standards?
Open file (360.07 KB 641x483 grwm1a9i30131.png)
>>13608 The biggest problem is the lack of a clear mission statement. I started out just wanting Iruru then wanting Senko San and modeling her chest and pelvis. I'm stuck on making the rest. Working on a mission statement together would help. I want her to be able to talk to me, follow me around, and snu snu. Her ideal design is basically Senko San
MeidoDev I want to help you, this will help me to focus on doing something rather than scattering my attention all over the place. Looks like a fun and reasonably feasible project. I'm assuming we're ok with wheels, and you want her/it to look like Senko San. Must - talk, and I assume process what is said to it as well, understand basic commands, doesn't need to be intelligent, but "just work" - follows you, so it needs to have a way to distinguish you from say, another person, dog or roombah. Cheap method: gps tracker fob you carry. More difficult: Facial / sillhouette recognition - Ambulatory but wheels or treads are acceptable, seems simple, you mentioned balance was an issue, how about angled wheels and gyroscopic stabilization? - Snu Snu: well enough said, if you're going to get intimate with this Meidobot then it depends what level of interaction you want and how tall does she need to be? I mean I'm assuming you want more than a fleshlight on wheels. I'm downloading Fusion 360 and I'm going to teach myself how to use it. Initially though I can provide pencil/paper sketches, I have large sheets of graph paper for this so it will be a notch above just scribbles on notebook paper hopefully. I'll update you with progress ITT.
Open file (22.92 KB 300x344 f.jpg)
Open file (37.84 KB 670x377 images-1.jpg)
Open file (141.48 KB 960x700 images-2.jpg)
>>13691 Thanks and welcome to the project Meta Ronin. Senko is the pretty much the goal. >talk I also want her to say "kon" at the end of sentences to emphasize her kitsune more >following GPS is a great idea >movement Angling her wheels for camber is a good idea. Any examples of it helping with balancing? I know it helps cars stay stable. >snu snu Hm, it would be nice if she said things like "daisuke" and other cute and loving things during and moved around a little. Senko Sans official height is 125cm with 10cm ears so, that should be fine. I suppose she'd be a skater in the moe fox danger zone to go by the classifications I saw in a thread a bit ago. Notebook sketches will be appreciated. Wishing you luck with Fusion 360, I use Blender and Meshmixer. Sk8o and Ascento look like good designs we could learn from.
>>13692 well given that its a prototype I don't think anyone's gonna be breaking doors down for making a 4' tall robot on wheels lol >I also want her to say "kon" at the end of sentences to emphasize her kitsune more probably the easiest thing to do out of any of it. Good examples there, I like the idea of separating the wheels so they can climb. I think bigger is better for stability and traction. I'm picturing something like chunky Megaman/Roll feet but with a wheel "inside" partially but I'm going to play with that concept a bit and see if I can innovate something "outside of the box". I'll post some sketches.
>>13693 I'm also hoping her being like a fox will help her be more acceptable in conventions and faires I agree that bigger is better. Chunky roll fox would be perfect. I bought her a tail. Wanted a normal one but, they had really faded color so got a silver one.
>>13071 >internal geometry of vroid studio model heads means they'd take a day to print and would be heavy. Please elaborate on that. Maybe in the thread for 3D design software or in the one for faces. Or upload some example. Thanks.
>>13718 I guess it's just optimized for creating avatars.
>>13715 Posted about it in the 3D modeling/printing thread. >>13721 >>13719 This is correct. The models were made specifically for animation and not for printing.
>>13722 > the 3D modeling/printing thread. What? I only realized that now. How is it the thread for a specific build or development but also the general thread for two other topics?
>>13743 I have no Idea. Chobitsu just decided it was that way.
>>13744 I'm going to crosslink every general comment in the other theads related to these topics.
>>13743 The 3D printing thread should naturally include modeling for printing. All of those general posts belong in. >>94
Open file (190.28 KB 771x1036 monster-1.jpg)
>>13693 Found cute inspirations
>>13895 Simplified faves may be the way to go
>>13896 Those big dopey eyes are cute.
>>13895 Thanks! Sorry for my recent absence. I'd like to continue work on this on the robowaifu.club until it can be determined what happened to the BO/Chobitsu or if this place is dead in the water.
>>14081 I'll make a new thread over there and all updates will be made over there until moderation returns.
>>14082 Posted the new thread
Open file (712.91 KB 724x1024 r.png)
This is exactly what I want. A beautiful dedicated maid filled with love and affection
>>14093 Hope you are doing well MeidoDev. It would be nice to see what you're up lately. BTW, that waifu is wonderful. I hope you make her someday! :^)

Report/Delete/Moderation Forms
Delete
Report