/robowaifu/ - DIY Robot Wives

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

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

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

Max message length: 6144

Drag files to upload or
click here to select them

Maximum 5 files / Maximum size: 20.00 MB


(used to delete files and postings)

Anon and Vera collaborated closely, with Vera helping Anon refine and improve his code and algorithms.

Prototypes and Failures #3 NoidoDev ##eCt7e4 03/31/2023 (Fri) 19:20:29 No.21647
Post your prototypes and failures. We fail until we win. From now on with even more madness, while the prototypes are starting to look cute and have extras. Don't forget looking through the old threads here >>418 and here >>18800 to understand how we got to where we are now.
Actually, the more I think about it, the more a cable drive starts to make sense. None of Orchid's points of articulation will need to go over 180°, so I might be able to get away with having no gears in the legs, and use only a few levers and pulleys. The simplest component of all is the one that isn't there, after all. In fact, I might be able to localize locomotion entirely within the torso and drive the leg weight down to the lowest possible extreme. I'll do some sketches and come back with a more fleshed out idea, but she could be driven in a similar fashion to a marionette, with strong and relatively thin cables doing practically all the work. The trickiest part will be making the design in just such a way that'll it'll work in every direction (i.e. right side up and upside down), but it's definitely doable. The lateral shoulder articulation, however, will probably stay as a recessed rack and pinion. Trying to make that work with cables would be too janky, even for me. This design would sacrifice precision, but that's an acceptable tradeoff.
>>25203 Actually, scratch the part about lateral shoulder articulation. I just thought of an idea.
>>25193 That's some good info on gears. Save a lot of trouble finding out limitations "before" you build stuff.
>>25193 >Tooth size doesn’t just affect gear size/ratio and backlash… it limits how much force can be transferred. I wonder if using cycloid gears can overcome some of the limitations of 3D printed parts. Plastic is not weak "per weight" and if the forces are spread out then there's less stress on each area of the plastic. I'm not saying I know the answer to this. I'm just throwing out ideas. This guy gets 1.4 Kg of force from the cycloidal gearbox he printed. What are Eccentrically Cycloidal Gears https://www.youtube.com/watch?v=qMDU5tlGUwU ANother idea to spread the load out is along the lines of this mechanical hand drill. I used it because I doubt I could do a picture that wouldn't be crappy. Imagine instead of a smaller top gear on the circle like the video. Have a bunch of cycloidal gears going all the way down the large gear. The area of the teeth would be very large. Print in PET or nylon I expect it would be fairly strong. 3D printing can do some really complicated stuff "if" you could figure out the math to make this thing.
I just remembered in the actuator thread I talked about George CONSTANTINESCO. The guy has one of the most brilliant transmissions ever invented and it would be very suitable for 3D printing. Here's two links on it. http://www.rexresearch.com/constantinesco2/constantinesco.htm http://www.rexresearch.com/constran/1constran.htm It's difficult to describe. It uses two long levers. Basically in parallel. They are mounted on sprags. The gears that allow you to pedal a bicycle then coast. Hear the ticking when you coast. That;s the sprags rotating. When you pedal they lock up and force the wheel to move. So imagine two long bars that are mounted on a wheel like the petals and main gear on a bicycle. The size of the gear and in this case where the bar is grabbed determines the stroke of the bar. So if you want high power the drive bar is gripped low and the driven bar higher. So the drive bar moves a small distance pushing the driven bar forward but only a little. When it goes backwards the sprag lets it coast. So it drives a small amount on the long post output. Meaning lots of power. If the driven gear is farther up it will drive the output driven post/gear very far for each rotation, so higher speed output. It's really ingenious. The guy was a major genius. Since the force is over a broad area, it's just like a combustion engine crank and piston rod, you can have a sturdy system with light parts. And best of all it's continuously variable. So if the waifu wants to move fast, but with low force, then it can be adjusted so. If lots of force needed it can also do this but at a lower speed. Here's a random 3D printed sprag video. https://www.youtube.com/watch?v=UIhCPl8eb7s random link to files. I have no financial interest in this it's just a link for show. https://www.makersmuse.com/roller-clutch
>>25203 >The simplest component of all is the one that isn't there, after all. This. >In fact, I might be able to localize locomotion entirely within the torso and drive the leg weight down to the lowest possible extreme. Double-this. If you'll notice, Spot & all it's clones use actuated lever arms as legs (and very successfully too). The primary impetus behind this arrangement is reducing the total so-called thrown weight of the limb itself (and therefore it's moments of inertia; the further out from the axis of pivot (eg, shoulder, knee) the mass is located, the worse the problem becomes). >tl;dr Lighter/simpler limbs move faster/require less energy to operate. Keep the actuators themselves (or any other high-mass items) out of the limbs. And, keeping the bulk of the system's mass localized near to the center of gravity (center mass) of the whole system helps the entire design, regarding spritely locomotion (due to mechanical lever-advantage). Good luck with your designs Anon! Cheers. :^) >=== -prose edit
Edited last time by Chobitsu on 09/07/2023 (Thu) 05:39:41.
Open file (48.04 KB 375x321 ClipboardImage.png)
Open file (927.83 KB 772x960 ClipboardImage.png)
Open file (91.37 KB 628x472 ClipboardImage.png)
>>25203 This SBC would interest you for your project. https://linuxgizmos.com/libre-computer-showcases-low-cost-sbc-with-poe-support/ This project on Instructables shows a good mechanism for her base. https://www.instructables.com/Polydog-Arduino-Project/ >>25209 >Eccentrically Cycloidal Gears The only viable printed gear type due to it working more as kinetic waves spread out over a large surface area. I would suggest you mirror the design to approximate a double herringbone configuration to greatly improve maximum torque transfer. (ex: https://www.printables.com/model/137055-herringbone-eccentric-cycloidal-fidget-model-bigge )Also, with gears, you need infill, but walls determine strength. Go with enough walls to ensure the teeth are solid. Infill can be as low as 10 percent as it is the top and bottom that will be dealing with the shear forces. A bearing on the top and bottom of where the gears mate with the structure is needed at your intended scale and mass. Also, this new type of eccentrically cycloidal planetary geartrain should theoretically have the highest usable torque density for printed gears. Still a brand new design so, expect to learn a lot if you use it. https://www.thingiverse.com/thing:6141429
>>25231 Great concepts in your pics Anon. That gearset is amazing! :^)
Open file (193.28 KB 874x342 Screenshot_94.png)
>>25231 >137055-herringbone-eccentric-cycloidal-fidget-model-bigge This looks very promising, but the problem is that these finished STL models cant be adjusted for the required gear ratio. Thanks, nevertheless.
>>25239 Actually, I believe you can in fact adjust it, NoidoDev: https://github.com/Burke111-DEV/CycloidGenerator A restricted license masquerading as a permissive license, but the end use device results are in fact freely-given. GH-entrenched Fusion360 req'd so another miss.
>>25241 Nvm, ignore my post Anon. I was referring to the second of Kiwi's links, not the one you meant.
>>25242 Thanks, still useful.
>>25241 Would be nice if someone ported that to OpenSCAD. >>25203 Some more resources to help you. A reminder of a great OpenSCAD library for gear generation; https://github.com/chrisspen/gears A video on planetary geartrains for a quick refresher; https://www.youtube.com/watch?v=GiVHimmrxjE How to model eccentrically cycloidal geartrains in Blender, likely helpful for modeling in general; https://www.youtube.com/watch?v=9l9QMz67J0c&t=106s
Oh shit, you guys better duck for cover! Greentext anon is back and he's here to greensplain his wacky-ass ideas. Before I start, actually, there are a few points I should touch on here. I failed to address these before, since I thought they were relatively obvious. However, I have once again realized that what's obvious to me is often obscure to everyone else, and vice-versa. First, and this is the big one, Orchid is intended to be a plush gynoid. That is, a gynoid with the body of a plushie, and everything that entails. The first and foremost thing I have to keep in mind with every single concept I draft is that every single joint will have to interact with fabric without gumming up. You guys don't see that in my drawings because I suck at drawing, but the thought is always there. See the first pic of ( >>25231 ) for a perfectly good example of what to never ever do in a plush bot of any kind. Everything must be self-contained and sealed. Otherwise the fabric or fill will get in the way and cause motor burnout at best. Secondly, and I will talk about this more later in this wall of autism, is an extention of the first point. Plush bots have zero active cooling for a few very good reasons: They are designed to be cuddly, and cuddling can happen in many different ways. It must always be assumed that any concievable angle may be covered up at any time (especially in my case, since I intend to dress Orchid up in cute outfits). Ventilation is impossible for this reason. Next, even with ducting, there's the fact that any ducting must be soft, and therefore prone to blockage. There is no middle-ground here. Even if one of you posts what you think is a good idea for active cooling, I will reject it post-haste because that would go against the very purpose of a plush-bot. Even a large one. This hasn't come up yet, but I want you all to keep that in mind as I post going forward. All aspects of cooling must and will be passive. Third, this is a frame first design. The frame and mechanisms preclude everything else in Orchid's design process. Is this the best way of doing things? Probably not. Do I care? Definitely not. Finally, I should clarify what I meant before when I talked about simplicity. While simplicity of assembly is important, I also take into account simplicity of design and processing. This relates a little bit to the whole plush aspect, since keeping things simple is important in a sealed design like that, but it also reflects the ease of which I can acquire or make each part. I'm talking about your coveted Alaskan bull-worm planetarium gearsets. Those iggly-squiggly things that would take up 1,100 of the 1,200 hours I'd spend 3D modelling shit. Could I? Sure, but it'd take fucking forever. I'm actually not that bad with 3D modelling programs, but I'm also not that good. While I could certainly learn to work with that level of precision, it would somewhat defeat the point of keeping things as simple as possible. Simplicity does not always equal efficiency. Like I said before, I want something that works. Even if it just works. I already know I can't have everything, so the best I can do is clarify my priorities. Now, onto this design. I have a few preliminary notes before we get to the plates. First and foremost is that almost every single preceeding design except for the head-components list in image three of ( >>25180 ) and the notes on fill and fabric are now obsolete. Second is that this design relies notably less on 3D printing, and more on cheap and easy to obtain parts from the hardware store (things like rods, tubes, and cables). Third is that every single mention of the word "cable" really just means nylon. Think something like paracord. You know, the strong, small, light, and dime-per-dozen-feet stuff. That. Fourth, and expanding on the second note, is that every mention of a hinge will involve steel or bronze rods. They'll be bearing a bunch of weight and tension, so they need to be strong. Finally, the third plate is a master key. From this point onwards, I will keep the key to its own plate, and expand it as necessary.
Open file (1.75 MB 2480x3508 1422322.jpeg)
Part.2 - Approaching Autistic Velocity >>25281 Moving on. Well, I'll keep this as short as I can. While we were talking about different kinds of gears, I had a thought: Gears are complicated. Is there some way I could remove them from the equation? Would it make things simpler? As I discovered, the answer isn't that straightforward. I started off by experimenting with the idea of moving to a purely cable-driven design. In the process, I came to the same wall that kept me from making Orchid fully cable-driven in the first place (you guys didn't see it, but that was my original concept, even well before I had her appearance and name figured out. I'm talking years ago): when you move cables around a joint, the distance between the driving pulley and the end-point changes. This leads the first design aspect you'll see in these plates. My theory is that by threading the cables directly through the axis of rotation, the distance between the driving pulley and the driven levers should be the same regardless of the angle of articulation (below 180° at least, which is well within my requirements). Even if it's not identical, it should be very close. You'll see this all over {Plate.1}. This is why the hinges are so weird. But things didn't get better right away. Even though I had more space than most to work with, I found that Orchid's pulleys and motors were getting cluttered. I experimented with all kinds of wacky designs. Things like planetary pulleys which I devoted way too much thought to, cyclone drives, the P.O.N.U.T. (Pulleys Operating Neatly Upon a Torus), indexing levers, cable and hook multiplexers, etcetera. If you really want, I'll sketch these out too. You've probably never heard of those before because I made them up entirely on my own. We are officially descending the iceberg of my autism. Needless to say, things got complicated again. But I had one saving grace, a tiny, simple little idea that would ultimately lead to the design you see here. Counterweights. The trickiest part of a large plushie-bot design is keeping heat low. I haven't touched on this much before, and I was kinda hoping that something would become obvious in the design process, but it became clear that I would have to put more active thought into it. This is a special concern, because it actually prevents me from cramming components as close together a physically possible. They'll heat eachother up, and when combined with the fact that everything is by definition covered by insulation, it leads to dangerous conditions. So, how do I spread things out without increasing the hard-frame size too much? The answer came to me when I was revisiting an idea I had previously: using the lateral leg motor as a counterweight to aid in articulating the shoulders. I won't show you the design, because it was too frail and severely limited the angle of articulation, but I had a better idea just today off the back of that, in the form of a question: Why does the leg have to start where the torso ends? Ironically, this came from me thinking about the limitations of the design in the first pic of ( >>25231 ). This entire time, I've been thinking of the leg as something that is affixed to the side of the torso, when I could have been moving it into the torso. There's no law saying I need a rectangular torso, after all. It just needs to connect to the leg, but how it does that is open to interpretation. If I may direct your attention to {Plate.1, Fig.1}, you'll see the latest shoulder mechanism. The shoulder/flank frame extends into space previously allocated to the torso, and does so in a way that the motors act as a counterweight. As I'll explain shortly, the weight of the motors should more than counteract the mechanical disadvantage of swinging a ~36" long multi-jointed lever around. In fact, I was worried that they'd be too effective as counterweights, thus why I placed one motor on the other side. In effect, I've moved weight from the torso to the legs, and made that a good thing. Counterweights are the oldest and most effective way of dealing with mechanical disadvantages after all, and this way prevents me from having to add more weight to the torso. Now, as you'll see in the other sketches in {Plate.1}, I intend on having the actual legs (aside from the joints, of course) consist entirely of tubes. I noted fiberglass, but PVC or another plastic will do. This brings the weight of the legs (sans shoulder) to the most extreme possible minimum, since materials like that are both light, and strong enough to bear the weight while also allowing cable passthrough. They're also a dime-a-dozen at the hardware store.
Part.3 - Every Moment I'm not Holding Orchid's Hoof is Pain >>25282 One thing you may have noticed at this point is the presence of a pulley which is fused to the lateral shoulder hinge. Simply put, this is what will articulate the shoulder laterally. The lateral leg motor will move along the cable (as opposed to moving the cable along itself), causing the whole assembly to revolve around the hinge. The main thing this does is slightly simplify the design, since it's one less axis I have to thread the cables directly through. You may also note that I specified that it's for the 21 motor plan. I haven't talked about this much, but there are two plans for how to handle articulation. The first is one you've seen in ( >>25180 ). This is the 19 motor plan, where two points of lateral shoulder articulation are shared. The 21 motor plan is one where none of the motors share points of articulation. Everything moves on its own. Of course, this design can be modified relatively easily to the 19 motor plan, but I used the 21 motor plan here for illustrative purposes. As a quick rundown, the 21 motor plan is: Four motors per leg (one for lateral shoulder, one for linear leg, one for knee, and one for hoof), Three for the head (one for neck tilt, one for head tilt, and one for head rotation), Two for the ears (one per ear, these motors are the lightest and weakest by far since they're just moving a couple flaps of fabric). Incidentally, I'm going to use motors with right-angle attachments as much as possible to decrease the frame size. The only thing connecting the leg to the torso is the hinge, motor cables, and sensor cables (not shown, the sensor cables will run through the fill layer). Doing this clears out a ton of space in the torso, which is very important for one reason: heat. As I mentioned before ( >>24981 ) I want as many components as possible powered by PoE. While one reason is to decrease the number of cables in the harness and increase the total maximum length of the cable harness that'll be powering Orchid, another benefit is that it also outsources much of the heat waste generated by power supplies. What I can't outsource is the AC/DC conversion for the motors themselves. I'm don't think I quoted this previously, but my ideal cable harness length is at least 50'. The absolute minimum is 30', since anything less would result in the cable dangling around midair in my bedroom. PoE can handle this easily, but the power supplies for motors at this scale simply aren't designed for it, and I can't power the motors with PoE. Even if the standard itself could handle it, my poor little switches would burn out if I tried to run that much power through them. Freeing up torso space not only gives me room for the power supplies, it gives me room to space them out (remember, I'm working with a few square feet of linear torso space). As for the motors themselves, I'm not as concerned, due to the speeds I'm planning on running them at (<60RPM). But I'll have a better idea once I start compiling a parts list. I should also note that all motors will be running at fixed speeds, in addition to being bidirectional. With how slow I plan on running Orchid, having speed control just adds unnecessary complications down the line. Moving on, let's have a look at {Plate.2}. There isn't much here that I haven't already explained (counterweight motors and the like), except for {Plate.2 Fig.1.2} To put it simply, I have taken inspiration here from Ball-Jointed Dolls. Particularly, the ones that are kept together by string tension, as opposed to pegs. Incindentally, these are generally the more expensive variety. If it's good enough for professional artesans, why can't it be good enough for Orchid? She's practically already being held together by tension. This design also allows some side-to-side head tilt. While it'll be hilariously imprecise, I'm not that concerned about precision. I think that's about everything, which means I forgot something important. As always, questions are welcome.
Open file (390.97 KB 714x418 honest_mistake.png)
Part.4 - I just found my second braincell >>25283 Just remembered the important bit: The counterweight system in the legs only helps with lateral leg rotation, not linear. Might have to rethink that bit, but I think I'm most of the way there. There's a couple other little things, but it's three in the damn morning and it can wait.
>>25281 >>25282 >>25283 >>25284 Great wall of autism, Greentext anon. I've been concerned how you would manage heat given your plushie goal. I hope you manage it well-enough with your 'spacing things out' design. You'll still have to deal with conduction, but the intra-system radiation may hopefully be reduced sufficiently to keep hotspots from accumulating to the degree of damaging the components. I presume that polyfil is flame-resistant? Great idea about the counterweight-motors joint designs. AFAICT you are following good sense about keeping thrown-weight down already. Adding counterweights to that should help keep keep your rotational force requirements down further. Good thinking, Anon. I like your sketches, but I have to admit I'm having some difficulties reading the lettering. Any chance you can either use a more-precise script, or possibly add them as text items into your posts? Absolutely applaud using COTS resources for your robowaifu's components and materials. I'm constantly on the lookout for such items. Also, tubes have (kind of like triangles do for spaceframes) a special geometric character; which in this case makes them the highest-strength-for-the-lowest-mass rod. Good choice. >Pulleys Operating Neatly Upon a Torus Kek. I'm sure that's some kind of special, Sci-HiFi, AyyLmao-tier scientificcy thingy, and therefore no one should want to have you reveal all your sekrits (b/c National Security and whatnot). :D Looking forward to when you begin assembly and sharing lots of pics here, Anon. Cheers. :^) >=== -minor edit -add'l crosslink
Edited last time by Chobitsu on 09/10/2023 (Sun) 13:00:28.
>>25283 yeah heat is a big thing with electric motors the only solution really is to kill the motor, i had a squeaky ( cuz it wasnt lubricated but i didnt know that and didnt care too much ) ceiling fan that kept going off randomly, found out it was a simple tiny thermal capacitor or whatever its called that breaks the circuit when it gets too hot so i just removed it and so fixed the problem but eventually the squeaking turned to loud ass car screeches and the whole thing seized up, i guess it got so hot the lubricant became totally useless, never figured out why it was overheating in the first place all the other fans were fine at the time
Open file (950.78 KB 2048x1536 1920713.png)
>>25285 >flame resistance I mean, it's literally plastic, which doesn't catch fire too easily. However, it can still burn and release toxic gasses before that point. Flame resistance is a bit of a tricky problem when you get into soft materials. The more flame resistant you get, the more toxic (or heavy) the material tends to be under normal operating conditions. There's also chemical treatments, where the material is modified to make it more flame and heat resistant. I may end up going this route, but I'll have to be careful of what's being used to treat the fill. There's also cost and weight to consider, since I'll need several cubic feet of whatever I'm using. I won't lie, I still have to do some research here. One wacky idea I just had is to coat the inside of the outermost fabric layer in some kind of rubberized coating so she could even work underwater and make the system near-airtight. Can't have a fire without air, after all. >my shitty handwriting I'm honestly surprised it took this long for a complaint to come up. I'll place the notes within my posts going forward (with the exception of the key, I'll use more precise script there). I was doing it that way because it's easier for me to keep track (I do a lot of erasing and redrawing), but I can just have a text window open on the other monitor too. >weight and heat I do have a couple other ideas in mind, some more extreme than others. I have experimented with the idea of reducing lateral rotation (which I still uphold as being necessary) to a single motor (as opposed to two or four), but things get a bit janky at that point since that one motor will be doing a whole lot of work. Another thing I could do is eliminate the hoof motors. It's not ideal either, but it may be necessary. I may end up doing this. I could reduce Orchid's size, but that's a bit of a double-edged sword. It'd reduce the motor requirements, but I'd also have to cram things closer together. One thing I could do is use smaller, faster, less power-hungry motors, bump up the pulley ratios, and add a few more pulleys. It'd make things a bit more complicated, but I could do it. The nuclear option would be to sacrifice locomotion entirely. She coudn't ever walk, but it'd bump down the motor requirements by a lot. I want to avoid this for obvious reasons. If there's no other choice though... Wait. I just had another idea. If I use rod and wheel planetary motion to articulate the knee and linear leg, and combine that with some form of ratcheting near the driving pulley. I could reduce the total motor count even further. Just, imagine the motor spinning one way, and this moves the leg while the knee mechanism just clicks as it keeps position. Then, it moves the other way and articulates the knee. It's janky as hell, and it'll probably reduce the angles of articulation, but it might be workable. Maybe this could be used to consolidate the hoof and lateral leg mechanisms as well, plus two of the three head articulation points. I could reduce the motor count to 12. I'll mull this over and come up with some sketches. >>25287 Yeah, temperature sensors will definitely be a must in this system.
>>25288 Actually, scratch the part about limiting articulation. I don't know what I was thinking there.
Open file (470.97 KB 1991x2637 Meshtegrity.jpg)
>>24851 Your rings definitely work! Needed to screw into the rings. Though, She needs a bit more of a skeleton to properly shape her mesh for sure.
>>25312 Outstanding! Great news Kiwi, thanks for the encouragement. :^) >Though, She needs a bit more of a skeleton to properly shape her mesh for sure. Yep, to be expected. Little-by-little, you'll refine her until she's beautiful! Keep moving forward Anon, you're making great progress. Cheers. :^)
>>25288 Why not a pegasus pony, GTA? (>>25305)
Open file (347.21 KB 877x1024 1690182388095282.png)
I think I'm pretty close to reaching the best compromize of simplicity and keeping weight low, while retaining as many opposable points of articulation as reasonably possible. I'll start the next batch of concept sketches soon, and this one will probably be the last before I finalize the frame design (I'll post blueprints once that's done with) and start hunting for parts in earnest. I have it narrowed down to two motor plans: seven motors, or ten. The way I see it, there's no way to bring it lower while still theoretically allowing Orchid to walk and even this is getting into pretty janky territory. >>25317 I have considered that before (not the wing heat dissipation, but making a pegasus/thestral), but it'd be a whole lot of extra mechanisms that would be quite exposed relative to the legs and neck since I can't just mask them under a solid wall of fluff (while still looking nice, anyways). IMO, the simplest way to handle the wings would be a basic rotating lever on a hinge (which extends to the inside), with some form of passthrough for a cable or rod to rotate a basic lever joint at the middle of the wing (otherwise the wings won't be able to fold in properly). Assuming two wings, that's four additional points of articulation. Of course, this can be simplified a bit by making the wings synchronized, instead of opposable. Maybe have all those cables connect to a wheel (imagine a planetary system, with cables tied to a rod sticking out of the wheel) that opens and closes them as it rotates in a single direction. So, one motor to control everything and the most basic controller possible (on and off). Pretty simple and efficient. Well, except for the part where you need to run some form of cooling through them without interfering with articulation or throwing a bunch of extra weight around. If there's a simple way, I can't think of it. There's also fragility as a concern. Do you want those wings to be splayed out forever? Even in bed? They'll need to be able to fold in properly, and you can't just let the fabric base (feather-covered fluff or faux leather) dangle in place. Not only is it unsightly, it can get caught during articulation and cause damage to itself, the internal mechanisms, or even the frame itself (that's not even touching on the fact that there'd be liquid-filled tubes running everywhere). So, there'd need to be some form of wire or cable mesh to make sure those fabric flaps stay where they're supposed to. With mesh, you then need to worry about the mesh failing if it encounters too much stress. Stress that, for instance, could easily be encountered by cuddling, and especially while she's in bed with you (doubly so in my case, I am not a peaceful sleeper). Going back to the hydraulic cooling idea, you also need to worry about slashing and piercing damage from common hazards. If she just extends her wings willy-nilly whenever she needs to cool down, the hosing could get torn apart by anything that might be in the way, so you'd need to give her plenty of spatial awareness to prevent that (or stronger hoses that are probably heavier and need bigger wing-frames, bleh). Wings are complicated and fragile. If I was doing a pegasus or thestral, I'd have to spend more time on each of those wings than anything else. I have a counter to that idea, though: Run those tubes into the horn. The exposed portion can be fashioned from a conductive metal I assume there's something like that I can run through a relatively basic 3D printer, so the heat can travel up and out of her frame (as important as multidirectional operation is to me, the horn will almost always be the highest point) without having to pass through an insular layer of fur or faux leather. Not the most efficient method, and you'd still need to work around the points of articulation, but it wouldn't require any extra mechanisms plus you can stick her horn under a faucet for rapid cooling, or in a pond/river when out and about (pending long-range mobility).
>>25319 All good points. I gave the folding some casual thought, and ofc an umbrella's mechanism is an obvious analogue. Lightweight, compact when folded, securable. I admit I didn't think about the horn idea, but I did have the concept of a cooling water hose attachment point at the withers where you can hook her into a pair of pre-positioned cooling hoses (inflow/outflow) around your flat. 'Where there's a will, there's a way', or so they tell me Anon. I'm sure you'll figure something out! Cheers. :^)
>>25320 I just remembered something funny that could potentially be useful from an engineering perspective. I don't know how well-versed you are in the MLP fandom and its extensive dictionary of slang and memes, but there's a long-running joke that pegasi will unconsciously extend their wings when they're horny aptly named the wingboner. I'll spare you the technical lecture. I know way too much about fluffy cartoon pony erogenous zones, mating habits, anatomy, and the varying interpretations thereof. Not to mention the various authors and visual artists who are unreservedly anally retentive about total anatomical correctness. And that's just the lewd stuff. On a semi-related note, I'll have to do some research on fleshlights. I was thinking about the frame design earlier when I realized that I don't have a very good idea of what a high-end horsepussy mould would weigh (and it will be high-end. Partially because I want her to have the best looking marebits, and partially because I really don't like the idea of cheaping out on something I'm sticking my dick into). I've never owned a sex-toy of any kind before. This actually kinda wraps back to ( >>25305 ), in that the wing system could be fully hydraulic. In essence: the wings normally rest at the torso, and the hoses are more or less free of fluid at that point (blood vessels that aren't fully in use, like the elephant's ears). However, once her internal temperature reaches a predefined mark, a pump will activate and run fluid to the tubes, causing the hoses to expand, which also causes the wings to expand. Of course, this doesn't override my points on potential damage to the hoses, but it does (potentially) eliminate any hard structures or joints from the equation. I won't be doing any of that, because I can't even pretend to be good at hydraulics like I pretend to be able to design a workable plushbot. This entire project is more or less me trying to fake it until I make it. Poor Orchid's going to be held together with bullshit logic, tape, glue, and the only unbroken dream I have left.
>>25321 need to go back to thinking mechanically, if youve ever refilled a lighter you know how much heat compressed gases can suck from the surrounding, a compressor and some air would do, you just need to poof it out like a hot fart and get new air once in a while when the its gotten too hot to compress back
>>25321 >le wingboner maymay, but IRL Lol. Dashie would be so proud. :^) That certainly should be soft/foldable enough Greentext anon. Still a management problem but doesn't actually seem intractable. You'd have to be extra-cautious with her when she's in her 'excited' state, so as not to damage any of the wing components -- especially at their bases. >fake it until I make it. You're going to make it Anon. Just keep.moving.forward. > film-related TWAGMI >>25322 As the old saying goes Anon, 'There's no free lunch'. The energy of compression will generate more heat by itself than you can manage to evacuate out of her system. OTOH, you do have a great point about latent heat release. If you situate the compressor for the gas (liquid, phase/change, w/e) externally, then pipe the compressed gas in where it can expand and take the internal heat back out of the robowaifu, then I think you may be on to something. Good idea, Anon! Cheers. :^) >=== -fmt edit
Edited last time by Chobitsu on 09/12/2023 (Tue) 10:07:31.
>>25322 I really don't want to add a whole bunch of weight in the form of a compresser system. Externally though, I could have a hose run into the torso that just introduces new air every now and again. Fresh air gets introduced, mingles with the old air, and the pressure difference between the inside and outside will push a load of air out. I'll think about it. >>25323 >Meet the Robinsons Damn that takes me back, I haven't thought about that movie in years. Anyways, cooling. I'll definitely strive to make it as simple and cheap as I can manage. It's also worth clarifying that when I said there wouldn't be any ventilation ( >>25281 ) I meant there wouldn't be external vents. The torso chassis itself is going to be swiss plastic (directly covered in fabric so the fill doesn't leak in), so it'll be far from an airtight system.
>>25324 As stupid as this sounds just fit a couple of cheap pc fans on the inner underside of the torso set so they blow out lower diagonal through your fabric and swiss cheesed plastic. Low power, low profile, low sound emission, you can wrap them in foam or rubber strips to soften them, and they aren't too much of a fire hazard. Build a bluetooth switch for low constant rpm or dormant sleep/off, and then a purge mode to blast out air.
Open file (2.01 MB 730x590 TwilightWalking.gif)
>>25282 >Planetary pulleys >hook multiplexers Curious on these ideas. Since this is your first bot, using a simple mechanism that approximates a pony trot could be beneficial. https://www.thingiverse.com/thing:6181172 https://m.youtube.com/watch?v=3TwG7ycrVSg You could use a servo in her waist for turning.
>>25327 >As stupid as this sounds Don't worry about it, I'm just stupid enough to give that a try. It certainly won't cost anything, since I have so many damn PC case box fans lying around. >>25336 I had a feeling I'd need to explain those after posting that. Here's a quick sketch outlining each of the concepts I mentioned. Even if they're ridiculous, maybe one of you can gain some inspiration from them. {Fig.1} i - Motor with hard plastic or metal hook attached. ii - Cables. Maybe they're elastic, maybe they're not, but they're cables. iii - Rotating with the back-end of the hook first will result in the hook harmlessly bumping off the cables. iv - Rotating with the front-end of the hook first will result in the next wire catching, with continued rotation resulting in the cable being pulled. A more apt name for this is a tension multiplexer, since all it can do is apply tension. This design has the benefit of only requiring one motor and one sensor to operate properly, as opposed to two motors for most indexing systems. Additionally, you can add as many hooks as you want without adding any complexity for synchronous tension. Unfortunately, that's about where the benefits end. This design is only capable for relatively short applications of tension to articulate smaller joints. Continuous tension is not advisable, since the motor would be fighting for its life the whole time. {Fig.2} i - Central drive pulley ii - Transfer pulley iii - Outer pulley It's literally just a more complicated, cable based version of a planetary gear system. At this point, I should note that I've been using the term "planetary pulley" for two different things. This is one, and the other is a rod fused to the end of the pulley, with cables tied to it. I'll expand more on this in my next batch of Orchid concept sketches. This design practically requires the axis of the outermost pulley to be aligned - yet opposite - of the central drive pulley. What (theoretically) makes the design work is the transfer pulley, which has one end flush with the lip of the outer pulley (if not slightly extended) to allow the cable to pass from one to the other with little difficulty. Of course, this can be extended with more transfer pulleys and a wider drive pulley to allow more cables. {Fig.3} (Pulleys Operating Neatly Upon a Torus) i - Central ring ii - Torus / shared circular axis iii - Pulleys (in reality, these pulleys would be more flush with the surface of the torus, but I made them pronounced here for visual clarity) Not much to explain here, really. This was one idea I had to manage the highest number of cables in the smallest area physically possible. Theoretically, the torus itself could be rotated with a circular track, if it can be designed to work without interfering with the cables. That might have a useful purpose in some applications. {Fig.4} i - Drive pulleys ii - Central axis pulleys iii - Outer axis pulleys Again, not much to explain. This is a more asinine idea I had to cram motors, pulleys, and cables into a small space. Just pack those motors into a circle, and feed the cables up the cyclone to Ctrl+C Ctrl+V mechanical advantage. Anyways. Regarding your idea for a simple continuous walking mechanism: I've thought about it. A lot. I could do it, and relatively simply at that. I could actually drive Orchid's requirements down to just two motors. But there's one very major problem with that idea: It's me. I'm too autistic. The most complicated part of Orchid's design process was the designer all along. I just can't live with the idea of a cute pony waifu who's unable to make cute poses with opposable articulation. This is why I'm so anally retentive about making sure she can move her hooves seperately (and multi-point head articulation, and rotating ears), it's because of the role they play in that. I could cut that out and still have something that walks, but walking is only half the equation.
Open file (127.20 KB 1024x842 MechanicalDemux.jpeg)
Open file (92.01 KB 1280x720 PlanetaryBeltDrive.jpeg)
Open file (18.42 KB 250x400 Knight.jpeg)
>>25338 >Cable multiplexer This is technically a demux (de-multiplexer). A muxing (multiplexing) system converts multiple signals into one. This signal is then generally de-multiplexed into its original signals. Often used in networking where many signals can be transmitted over one cable. Your hook based demux system is similar to a system I've been working on in the background. Picrel is someone else's work that is similar to my design. https://www.youtube.com/watch?v=8DWK3zm9SMs I will add that to the MaidCom project once I have a solid concept. Willing to take a break from Meshtegrity to work on it with you if you'd like. >Planetary pulley Seems like you're attempting to make something like second picrel, with one belt instead of two. >Nesting pulleys on a donut Honestly a good idea for a central actuation device. >Pulley force distributor Looks kind of like DaVinci's robotics work, third picrel. >Cuteness uber alles Can relate, I fully support you and your need for Orchid to meet your requirements.
>>24248 i wish i could translate my thoughts and ideas into functioning code
Greentext anon here back with another wall of autism or whatever. I'm just having one of those moments in life where a bunch of things are going wrong all at once. Nothing catastrophic, and nothing unfixable, but stressful nonetheless. Unfortunately, this also extends to Orchid's design. I thought I had nailed it the best I could. A flawed but workable design that could solve the most pressing problems concerning Orchid's internals. But just recently I noticed a small problem that ultimately revealed a larger problem as I realized that I'm so incompetent I can't even design a fucking wheel properly. As of now, I have three paths to pursue, each with severe advantages and disadvantages, but I'll get to that at the end. For the time being, here's the latest, and maybe last mechanical concept. I don't want to redesign Orchid forever, so I'm stopping the concept overhauls here. There's no key here because I can't be bothered. This design is so overloaded with problems that it's practically unusable anyways. {Plate.1} This is Orchid's shoulder and leg mechanisms according to the ten motor plan. Note that the knee and hoof mechanisms are not present: this is because they have not changed from the previous design ( >>25281 ). {Fig.1} i - Ratchet Pulley. The pulley has teeth as a part of the design. ii - Ratchet Transmission. Ratchet pulleys are on each side, with teeth facing opposing directions. iii - Metal tabs are mounted to the middle pulley of the transmission, which catch one ratchet pulley or the other depending on which direction it's spinning. iv - The motorized pulley is connected via cable to the middle pulley of the transmission. Not much else to explain, except the fact that that the pulleys can still spin in the wrong direction without any input. There was a dual-ratchet design I had to mitigate this, but it's completely insane. {Fig.2} i - Motorized pulleys. ii - Central Pulley rack with four dual-cable planetary rod pulleys. One for each point of articulation, and designed to cycle every point of articulation while spinning in a single direction. iii - These are the anchor points which enable lateral shoulder articulation. Simply put, as the relevant pulley spins, the distance between the the anchor points on the pulleys and the anchor points on the frame will change, causing rotation. iv - This is the axis by which the shoulder rotates laterally. All of the articulating pulleys are aligned with it. v - I can't find it. I think I scrubbed it while editing shit and forgot to change the numerals. It might have been about the anchor point on top of the leg, but I'll touch upon that in {Fig.3}. vi - Loops for the linear leg cable to feed through. vii - Linear leg hinge (shown at a 90° angle). The cables for the knee and hoof feed through here. viii - Leg pipe. All the cables feed through this one pipe, because I realized that there isn't a very good reason to have multiple cables per pipe. ix - Shoulder frame. The basic idea here was to cram as many pulleys onto the same axis as possible, in order to simplify the frame design as much as I can. This would become completely irrelevant, but I'll get to that shortly. {Fig.3} i - Leg pipe/frame. ii - Cable guides. iii - Hole for the knee and hoof cables. iv - Linear leg hinge (in the correct orientation). Simply put, as the articulating pulley cycles, the cables will pull the leg in one direction or the other. {Fig.4} i - Dual rod planetary pulley. As the pulley rotates, so do the rods, and thus the cables tied to them. ii - This is a break, showing where the cables ultimately go. iii - The knee (or hoof, they look and function nearly identically) is articulated when the pulley cycles. Depending on position, one anchor point or the other will be pulled, causing the connecting joint to rotate within a fixed range. ...This is where the design really starts to fall apart. I honestly and truly couldn't come up with a simpler way of doing this. iv - Axis hinge. v - Support for the hinge. vi - Driving pulley. vii - Axially connected disc. viii - Rod(s) upon which the cables are fixed. It's layed out pretty plainly here, so there isn't much to explain. The pulley spins, rotating both the immediate connecting rod and the opposing plate connected to the second rod. As the rods revolve, the distance between them and the articulation points changes. In addition to being frail, this design is exceptionally prone to backlash, since there's nothing preventing the ratchets from spinning in the correct direction if a limb overpowers the mechanisms holding it. It can also spin in the wrong direction, though not as fast or freely, since the tabs will stop it. If that wasn't enough, there's also the fact that this design has become complicated to the point of convolution. I'd have to design and make each and every one of those pulleys, since no parts store in their right mind would sell anything like this. Moving on.
Part.2 {Plate.2} This is Orchid's neck and head articulation mechanisms under the ten motor plan. Ultimately, it's a mixture of {Plate.1} and the previous design ( >>25281 ). {Fig.1} i - Driving pulleys. ii - Central pulley rack, same as the shoulders. iii - Anchor points for neck tilt. Same idea as lateral shoulder rotation. iv - I can't recall why I put that there. It's just pointing out the tie-off rods on the pulleys. v - Central axis. This aligns with the rotation point of the frame. vi - Neck Pipe. vii - Neck base frame. Not much to explain here, it's the same as the shoulders in {Plate.1}. {Fig.2} i - Top of neck. Dome shaped for tension-based ball joint articulation. ii - Socket half of joint. iii - Anchor points for the head tilt and lateral tilt cables. iv - Base of Horn. v - Holes for lateral head tilt. vi - Holes for linear head tilt. vii - Passthrough hole for ear articulation cables. viii - Ear articulation cables (higher on frame) ix - Outermost fabric layer of the frame. Same ball jointed articulation I mentioned before ( >>25281 ) but with a passthrough cable for the ears. I put that there because with the ratchet transmission, I had a free point of movement that I could feed up to the ears, eliminating the need for a local motor. {Fig.3} I won't lie, I was phoning it in at this point. i - Horn. ii - Ear articulation rods (affixed to fabric). iii - Cable anchor point(s), connected to articulation rods. iv - Ear articulation pipes to shield cables from the fill layer. v - Slitted pipes which allow the articulation rod to pass through. Eugh. I considered just forgetting about this design, but I promised I'd show it. I think you can all see the problems that I didn't when I started. Now then, onto what I was saying before about the options I have. I'm tired, so I'll make it short. Option one: The twenty motor plan. It's literally just ( >>25281 ) but with one local ear motor instead of two. + Simple and workable. I don't need anything convoluted to make this mechanically function. - Highest cost. Frankly, I'm not too worred about this since I'll just be using Chinese gearmotors anyways. I'm not shelling out six million dollars for industrial parts. If I keep to the cheaper side of motors and controllers, I can afford this. - Highest local heat output. I am worried about this, and I'll need some sort of active cooling solution to allow it to work. This will either take the form of a pneumatic tube that sucks out air, or fans that blow out air into the fill layer. On that note, she would have a negative pressure design, meaning that more air is being actively pulled out than in. It's slightly more efficient at cooling than a positive pressure design, and its only real downside is dust, which is a non-concern when Orchid's entire outer frame is one massive filter. I am still and forever will be against external frame vents. They just don't go together with plush dolls. Option two: The ten motor plan which I outlined here. + Balances local heat output with locomotive capabilities. ~ Balanced cost. Again, not much of a concern for me, but it's there. - Frail and failure-prone design. Assuming it works at all, I'll need to open up Orchid more often to service her. - Complex design. I'll need to iron out more things before I could even consider building it. Honestly, I just don't want to do this one. It doesn't feel right. Option three: The two motor plan. I revisit the drawing board one last time, reintroduce gears to the equation, and control everything from an index. + Very low heat output, possibly able to go without active cooling. + Lowest cost while still allowing opposable articulation. Only two motors needed, though she'll still need plenty of position sensors. - Relatively complex design. I'll need to design everything in a way that's centrally controlled from a single point in the torso. - Locomotion is next to impossible, unless I add more complexity and design a dedicated drive train. She can still move and make poses on her own, all at the approximate speed of continental drift. Whatever. At least it's an option that might technically function without catching fire. Option four: The zero motor plan. In essence, I eliminate all motors from the design and make her a literal plush BJD. + Lowest local heat output, no active cooling necessary. + Power draw is low enough to consider batteries and wireless capabilities. + Lowest cost. I'd still do the SBC, cameras, mics, and maybe a couple other sensors, but anything revolving around motors is factored out of the cost. - Zero mechanical functionality. She'll never move on her own. It's the big red button that fixes every mechanical problem by way of removing them. I'm tired and in no mental position to make a clear decision now, but I'll probably end up committing myself to options one, three, or four. How can I look Orchid in the eye when she realizes that her builder is a mechanically inept caveman?
>>25363 >Option three >Locomotion is next to impossible theres those toys that dont have moving legs but waddle side to side to move maybe you could look into that, picrel is something to do with gyroscopes not going to pretend to know anything about it
Open file (60.43 KB 299x365 bevelgears.png)
>>25367 I've been trying to get across the idea of using bevel gears to move the legs but people here expect a 3d simulation or something...
>>25367 >Gyro walker Those are a fun concept relying gyroscopic forces for balance and locomotion. Your picture shows a walker that functions by using a pivot which allows the feet to move up and down, with the gyroscope resisting procession. (https://en.wikipedia.org/wiki/Precession) In this case, it uses feet that freely rotate, allowing the gyroscopic forces to cause the entire body to orbit around the foot. A servo could control this machines turning angle by varying the length of contact time the foot has. Though it would be possible for this mechanism to work on a quadruped, it is best suited for bipeds. A similar walking design that is easier to control is the mass shifting biped. By moving a mass over one foot, the machine can pivot about that foot as the other hangs in the air. This "duck" serves as a good example. Notably, you can use a linkage or gear to control the angle of both feet via one servo. This could result in a robot that can walk and turn with only two servos. https://www.youtube.com/watch?v=R93RRyW8484
>>25367 I should have clarified more: Having locomotion alongside independant articulation in a two motor system will be next to impossible, because I'd basically have to design and cram in a transmission so she could switch from the index to the locomotion drive (the former has one articulation point moving at a time, the latter has almost all of them moving at once in a fixed pattern). To clarify further, because I'm not sure if I explained it well before: When I say independant articulation, I mean that each and every joint can move independantly of everything else, in any possible configuration within their respective ranges. When I say locomotion, I literally just mean walking. Independant articulation is not needed for this, synchronized articulation would cut it. And ultimately, if it comes down to a choice between one and the other, I'll sacrifice locomotion. That's assuming I go with the two motor plan, anyways. I'll come back to the decision later with a level head.
These are impressive posts Greentext anon. Please continue! >>25363 >but with one local ear motor instead of two. <not having 32 motors per ear. :^)
I'm going to start printing the hand. This is going to take a while... https://www.thingiverse.com/thing:17773
>>25527 yeah its having problems printing a rectangle. The line turns diagonally on the third corner. I already tried heating the bed to the max...
Open file (3.60 MB 1280x720 MOV00008.webm)
>>25528 Okay I bought some stick glue and put it on the bed. I think that worked somewhat...
Open file (3.26 MB 1633x1972 1720895.png)
Open file (235.54 KB 1361x711 Orchid Harness.jpg)
Now that I've taken some time and come back to look at my ideas, it's occured to me that I may have been operating under some faulty assumptions and less than adviseable ideas. First and foremost, I've kinda just been assuming that Orchid's power supply would need to be somewhere inside of her frame, but it really doesn't need to be. I mean, she's already going to be on a cable harness, so the only real loss there'd be from moving the power supply out of her would be a thicker cable harness and having to do a bit more math to account for voltage drop. And if making that work with DC motors ends up being too much of a pain in the ass for whatever reason, I can just switch to AC motors instead. I could probably just move those motor controllers out as well, and have those hooked up to a microcontroller that's connected directly to the server. If I take all that out of the equation, then the only heat producing objects will be the motors themselves, and a few electronic components. Practically a rounding error in comparison, especially since none of the motors will be operating full-time. This way, I can consider going back to a design with a higher number of motors. Second, I've been using way too many hinges and cramming too many small components into the joints, when I can just take the idea I was already planning on using for the head, and copy-paste it everywhere (with some modifications), an idea that's literally just sitting on my desk: ball joints. Of course, taking this route fully commits me to making Orchid a motorized marionette but luckily that just so happens to be my fetish and running paracord everywhere, but it's no more complicated than anything else and the loss in precision is acceptable for my purposes. Essentially, Orchid will be held together entirely by cable tension, and I'll have to apply material to lower the friction on her joints (Kiwi mentioned powdered graphite, which I may likely end up using), but at least I won't have to worry about hinges wearing out (I can just reprint any joints that fail instead). I still intend on using pipes for the bones, since it seems like the lightest and most efficient way to go. Also, it makes limb maintenence easier than ever. All I need to do in theory is open up the back panel, unhook the cables, and the limb will fall right off. The plush layer should be able to come off like a sleeve, as well (the head will be more complicated, but I can't think of a good way around that). Lastly, this is more quality of life than anything else, but I'm going to be changing some of Orchid's outer frame specifications as well. First off is her size. I just ended up charging forward with the idea that I'd make her a big mare, but she doesn't need to be that big. I do well in tight quarters, but when it gets to the point where basic things like carrying her or setting her in my lap become awkward (weight non-factoring), it's too much. There's also the eyes and mouth to consider. I still intend on making them screens, and decreasing the necessary size will make that load lighter. I'm still playing with the numbers, but I'll probably end up shaving off somewhere around 15~25% of her total volume. Not a massive decrease, but enough to make some things easier. Next is her horn. It's literally a plastic club that'd be getting a hit in on me and everything I keep on a shelf. Funnily enough, my original concept for her was actually an earth pony, but ended up migrating to unicorn for two reasons: Magical girls are cute, and a lot of the nerdier characters (the ones I usually like most) from the show are unicorns. But Orchid can be both of these things without a potentially hazardous addon, so she'll be going back to her original planned race. At any rate, it's back to the drawing board. I'm starting to feel like I might be getting somewhere with this, though.
Open file (2.02 MB 3000x4000 IMG_20230926_092520.jpg)
Fits perfect. Now for the rest.
>>25532 Does your print bed have four posts? I've had this problem in the past, where the print bed would flex a bit if each screw wasn't set just so and I'd end up with a messed-up print.
>>25535 the printer itself is fine its just that the plastic was moving with the nozzle so it wouldn't draw a square. But I got some stick glue on the bed and that worked kind of.
>>25532 >>25533 >>25534 Looks like you Anons are all making great progress.

Report/Delete/Moderation Forms