>>7102 >LOL fuck that I'm just a pervert building himself a robowaifu not a Boston Dynamics engineer! Actually I would say you closer to being a pervert building himself a robowaifu and a Boston Dynamics engineer! Or at least closer to than say the average orangutan heh. But seriously, you're much better at this than you give yourself credit for don't let it go to your head Anon :^) and you're making nice progress with your Elfdroid Sophie. It's inspiring to us all, so please don't stop until she's done.

There are not only stepper motors and servos! This is not the only distinction, and we are getting confused here over and over again... https://en.wikipedia.org/wiki/Brushless_DC_electric_motor can be a stepper motor: https://en.wikipedia.org/wiki/Stepper_motor but there are also ones for airplanes (aeromodelling) or radio-controlled cars, which are not (brushless) stepper motors! If I'm understanding it correctly now, steppers have the positioning system build in (encoder), and WP on servos as well, but it a additional system like a optical rotary encoder. WP: "It consists of a suitable motor coupled to a sensor for position feedback. It also requires a relatively sophisticated controller, often a dedicated module designed specifically for use with servomotors." We can use also use simple electromotors, ideally BLDC in most cases, I guess. Then adding the sensors somewher else, and also adding our own gearboxes or drives. This might not always be the best way, but it's more flexible. >>7089 The NEMA motors for printers are quite cheap. Though the NEMA standard refers to the size and format anyways. More expensive motors ones have probably also more power. >>7095 Using BLDC motors with our own drives might be precise enough in many cases. My understanding is that these are like your high-torgue motors, but the control would be external. Steppers and servos are very precise, probably more than we need it. >>7102 > 60kg.cm servos in the elbow Fine, but why does she need to lift ~40kg or more using both of her arms? Could the arms and hands even bear that much? > As for making legs that actually walk though - That's something for later. Most people here always came to the conclusion that it's not that important. Also it's a gradual thing. Standing while doing some works (dishes, cooking) or dancing while standing on the same spot would already be quite useful.

>>7102 Hey this is my first post here, some anon linked me here after I made a thread on /diy/. Anways, are you sure 60kg is enough for the elbow? My estimate for the elbow alone was a 200kg servo. With an arm length of 30cm that leaves us with 6.6kg. Now if the shoulder moves the entire arm, the elbow has to be able to hold the forearm at least steady, if not move it, when it is being moved. For the shoulder itself I guess 300+kg? As stated though this is my first post and I havent been in this field for long. I wanted to share my opinion on the calculatios and I'd like to know aswell how you determined your values :) PS: Other kg servos work aswell, but after gearing them they need to be able to do at least 0.5s / 60 deg to be humanlike, I measured that.

>>7130 >Hey this is my first post here Hello Anon, welcome. Please make yourself at home here on the board and don't be afraid to ask questions. Most everyone here is pretty helpful and will answer you if we can. We have plenty of different robowaifu-related topics on the board and hopefully you can find some that interest you here. Good timing, we're having our 4th birthday coming up tomorrow.

>>7130 Welcome. How did you calculate that? I realized my calculation in >>7108 was rubbish, bc I got confused, but I don't even see where you are coming from. The simple version would be 60kg / 25cm is 2.4kg per arm, or not?

>>7106 > don't let it go to your head Anon This. One of the many reasons I am building a robowaifu is so that I can have a companion who lacks an ego. For all the advances it has driven us to make, the human ego has also caused incalculable damage to everything on this planet (I think it's one of the biggest human flaws). You know how women like to bang on about the "fragile male ego"? Well, why is it then, that the entire Western mass media and advertising seems to be full of characters and scenarios created for the sole reason of pandering to the female ego? Maybe it's this, combined with the bullshit that simps tell women, but it seems to me that many Western females have astronomical-sized egos (at least as bad as the men). I've worked with plenty of women. Four of them had gigantic egos (only one guy I knew was egotistical). One woman was even fired because of it! She sank herself. I didn't have to do or say anything! These are the kind of women who, if you contradict them or disagree with them or correct them on something then they will go off on one. Especially if you refuse to give and dig your heels in. I've witnessed months of spiteful bickering between women over nothing! One of them was ranting on about "professional integrity" or some shit. I don't know. Didn't have the time to listen to all their arguing because I was too busy doing my job. Honestly, the best thing to do is not get involved and robowaifu instead. > please don't stop until she's done Thanks anon! To paraphrase John Connor; "She cannot be bargained with. She can only be reasoned with. She doesn't feel envy or hatred or fear. Sophie absolutely will not stop, ever, until she is Kawaii!" I plan to complete her body although I'm not sure if she'll ever be fully 'done' because there's always things to improve and upgrade! >>7108 > why does she need to lift ~40kg or more using both of her arms? Could the arms and hands even bear that much? TBH the only reason she has 60kg.cm servos in her elbows is because Ryan Gross - the original creator of the Proto1 robot arm - recommends that torque rating. I have seen his arm operational in the following video - https://www.youtube.com/watch?v=O-9RETaBMVw Other robot arms that I have purchased as a kit tend to come with the cheapest, weakest servo motors possible, meaning they can pick up a sponge or a plastic cup and that's about it. You can do very little with them. So I'd rather my robowaifu have overkill servos and be stronk rather than having only just enough power to lift her own forearms. I have been pleasantly surprised so far at how strong PLA with triangular/pyramidal infill is. I literally cannot break some of parts that I've printed with my hands. I'd have to stamp on them. Even then I'd probably hurt my foot. Am still waiting for a pack of servos to ship and I have a couple more parts to 3D print before I can begin testing my redesigned version of the Proto1arm. It will not be as strong structurally as the original because I have removed a lot of plastic, but it will be very light. This means that I am itching to test it because my version of the arm may very well break at certain points and require reinforcing again. I need to run lots of tests to see what I can get away with. (Also, other anons will be able to easily modify the .stls and fit smaller servo motors into the joints if they wish). >>7130 Welcome to the wonderful world of Robowaifus anon! If you want to build your own A.I. companion, with or without a robotic body (and the necessary anime cat-girl upgrades) then this is the place to be! On the subject of robotic bodies, I just found out about "Reachy" the robot torso and arms. They have some interesting resources available: https://www.pollen-robotics.com/opensource/ These may be of some use to any other anons who are crafting a robowaifu. BUT the 'Pollen Robotics' company are using those freaking Robotis Dynamixel servos again, which basically means this "open-source" project is out of reach for the vast majority of people because they only count as "off-the-shelf parts" if you are absolutely loaded and have thousands of dollars in disposable income. They are trying to sell the whole thing for 15,490 Euros ($18,432) and it doesn't have fully articulated hands!

>>7138 Calculate what? Rotation speed i measured and I did some assumptions about arm weight itself and an ideal payload. As I'm new here I'm not sure if we all have the same goal in mind of what she should be able to do. Me personally I dont care about movement atm, Im most interested in humanlike arm movement, along with proper strength.

>>7136 Thanks :) Is there a TLDR of the progess you guys have all made over the 3.999 years? >>7217 Thanks :) Ill let some other anons handle the AI, Ill focus on the mechanics first!

>>7217 >I plan to complete her body although I'm not sure if she'll ever be fully 'done' because there's always things to improve and upgrade! Haha, good point. Well, just keep at Anon. Do it for all the John Conners among us! Cute pic of Sophie btw, she's already becoming pretty kawaii! Happy Thanksgiving /robowaifu/.

>>7217 Yes I see, if you can fit such servos in then it makes perfectly sense. The problem was just my absurd miscalculation and then I was wondering why you would make her so strong. I'll look into the pollen robot website, thanks. 15k is the number I would anticipate for the absolute premium models of a fully functional robotwaifu, while the cheapest with their own body should be under 1k. Their choice of servos might be the reason why they already hitting the premium price point, without having a human-like robot. >>7219 Sadly not that much progress, a lot of people which came here had to start from scratch, some were only dreaming, a lot gave probably up. Others might only be active on other platforms, like Youtube, Thingiverse, Discords like the one of Lex Friedman, some other AI or robot websites, Doll Forum inventor thread, ... Others might have found jobs and doing other things... @all: let't try to keep this thread as much as possible on topic (actuators of all kinds), and for other conversations move over to other threads. Thanks. Meta about the forum and robowaifus >>3108 and Off-Topic >>39

>>1428 >Hydraulics are highly toxic. It's really not. I repair Hydraulic cylinders for a living. It's fine as long as you don't eat it. I also wouldn't stick my dick in hydraulic fluid. It's the pinhole leaks you need to worry about. Hydraulic injection hazard is real and needs to be taken seriously. But you can get great control with them. And a lot of power. But they can be noisy. You don't necessarily need a high pressure system to do the job. In fact I wouldn't want anything using more than 500 psi interacting with a human. My boss had his finger cut off by an air cylinder. and that was at 150 psi. Tldr Hydraulics can be dangerous but not cause they are toxic

>>7288 >> I also wouldn't stick my dick in hydraulic fluid. If we would be using that, then in the arms and even more so in the legs. That might be close, but not too close. Also: >> Hydraulic injection hazard is real Thanks, but there are ideas to use some more harmless liquids like plantoils and replace them more often. If we would use Hydraulics, then we would probably go with that. >>But they can be noisy. This could be a problem, but I guess this depends on various factors. One way to look at it, would be that we would have some strong muscles which are only be used when necessary. In special situations, like lifting something heavy or going up some stairs. >My boss had his finger cut off by an air cylinder. and that was at 150 psi. Oh, thanks for the remainder to be careful. This certainly applies as well when working with Pneumatics or motors.

>>7217 >https://www.pollen-robotics.com/opensource/ >They have some interesting resources available Yes, they have: https://forum.pollen-robotics.com/t/orbita-presentation/20 though the files on Onshape seem only to be available to view them. But I didn't look around very much, maybe I need to register first.

>>7288 Hmm. I see, thanks for the information Anon. Hydraulics could be highly useful for some types of humanoid robotics, but personally I'm pretty skeptical atp about it's utility inside /comfy/ robowaifus you want to snuggle with inside your house. >>7290 >but there are ideas to use some more harmless liquids like plantoils and replace them more often Seems like a good idea if it's workable. That would at least reduce some of the issues involved with hydraulics. >>7354 Yes, that's a pretty interesting joint design IMO Anon.

>>7095 check >>7510

>>7108 Quick fact check. Stepper motors are four phase brushless motors with a many toothed rotor to facilitate high precision in motion. They are commonly used without encoders because their inherently high accuracy and precision leads to not needing positional feedback unless the rotor is prevented from moving. This is known as missing a step and rarely happens in proper implementation. Brushless motors have a higher power density as a smooth rotor leads to a higher magnetic flux density. They typically are three phase but, can have any number of phases in theory. Two phase brushless motors in fact exist and are primarily used in fans where they only need to rotate in one direction with high efficiency and a long life. A servo is any actuator with feedback for control. A linear actuator with switches to for positioning? Servo. A stepper with a rotary encoder? Servo. A balloon that displaces water which triggers open electrical contacts to control volume? Servo. Anything with with feedback for controls is a servo. In short, brushless motors are technically low precision stepper motors and stepper motors are high precision brushless motors but, the term stepper is useful towards differentiating which kind of rotor is implemented. Servos have some sort of feedback for control. Brushless motors are currently the ideal actuator for waifus. They are significantly lighter for their power density then anything comparable. Their main problem is their incredibly high price when compared to DC motors due largely to their controllers needing an order of magnitude greater complexity compared to DC motors simple H-bridge.

>>7951 Very lucid definitions, quite understandable. Thanks.

>related xpost >>7969

Here's another cyclodial drive, really cool: https://youtu.be/vYmF4hZzFhI I might start trying some design soon, probably recreating it in Solvespace. It might be that one. Looks easy enough. But I don't know when to start since I'll need to take it easy for one or two month bc other things to do. I currently don't have the right motors anyways.

>>8170 Good luck Anon. I'm sure you'll figure everything out alright.

>>8170 More on cyclodial drives: More detailed explanation with design process including formulars and such, Solidworks, milled aluminum: https://youtu.be/Nk3aaVcvbpA - he also mentioned at the beginning that these can be made backdrivable and therefore compliant (save, responsive to some resistance). He uses a GYEMS RMD-L-7015 servo. Here the first test with a 3D printed prototype: https://youtu.be/i_AI95lYe0M - 71deg/s and circa 3.2kg with a 28cm arm. Another one, 12:1, small with a TT gearmotor: https://youtu.be/01v10qvjFIA and on Thingy (but only stl): https://www.thingiverse.com/thing:4444667 Roomba board, 20:1, Nema-17, with a lot of thoughts on how to do it with a printed design: https://youtu.be/WvnJjsAkslE and Thingy (stl only): https://www.thingiverse.com/thing:4604783/files Other topics: Someone mentioned mjbots quad project in the meta-thread >>7969, Skyentific made a video on their brushless motor controler: https://youtu.be/R2uuTfuyadk

Reading a bit on brushless motors. Intentionally not ones for robots, since I want to get the basics and understand the differences between the different variants. Quads (aka Drones): https://dronenodes.com/drone-motors-brushless-guide/ RC cars: https://www.rccaraction.com/brushless-motor-tech-everything-you-need-to-know/ Helicopters: http://www.modelaviation.com/brushless-motor-design Bigger ones and a longer interview about the details on motors: http://www.modelaviation.com/brushless-motor-design >RH: Most RC helicopter motors are outrunners, because they provide excellent torque. In an outrunner motor, the motor casing spins and magnets are glued to the inside wall of the motor casing. The extra inertia from the spinning motor casing helps with torque. >For inrunner motors, the motor casing does not spin; the magnets are mounted on a rotor that sits in the center of the motor. Inrunner motors have lower inertia and can spin very fast—even up to 50,000 rpm. They are popular for on-road RC race cars. >High-quality stator material allows the internal molecules to change direction quicker. An iron-nickel (NiFe) alloy is a typical stator material. Currently, Japan manufactures the best treated iron material for motor stator use. Good stator iron is expensive, requires a special manufacturing process, has gone through the right hot-cool cycle, is stable with temperature, and has no oxygen inside. We buy our stator material from Japan then stamp it to the shape we need in Germany.

Of course, I forgot the pictures. Because my browser crashed before and reloading didn't load them. So here they are.

>>8213 Thanks. I like those diagrams, very helpful.

Reeee, I think this is the "gearbox" we really need: https://youtu.be/0-uSUrcRsyw and it's from 2017 but has been ignored. I don't care about the efficiency part so much, till I can take my waifu hiking or so, but it can reverse directions smoothly. This is crucial for security and compliance. https://spectrum.ieee.org/automaton/robotics/robotics-hardware/inception-drive-a-compact-infinitely-variable-transmission-for-robotics > Force control: Apply a constant force at the motor (or spring) and determine the output force by modulating the gear ratio; >Velocity control: Apply the most efficient motor velocity and determine the output by modulating the gear ratio; >Energy flow control: Monitor the amount of energy distributed across the system at any point in time, and move it using the transmission ratio and the motor controller to optimize efficiency or power; >Impedance control: Because the transmission ratio is adjustable, the impedance of the system can be tuned for optimal environmental interactions (excellent for managing impacts and human safety factors).

>>8212 First post on this forum. Would it be viable to use a geared-down drone motor for a miniature robot with a limited range of motion?

>>8270 Not him, but I don't see why it wouldn't work. In the end, it might just be simpler to use specialized servo motors for most of the many motors probably needed. Welcome BTW. We have an Embassy Thread if you'd care to introduce yourself >>2823 (not required ofc). What kind of miniature robot were you thinking of Anon?

>>8221 I remember looking at that system before Anon. Very interesting actually. I wonder why it hasn't been pursued further by this stage?

>>8270 >Drone motor with gears I haven't tried myself, but I'm sure this would be working and I intend to go that way. >>8277 We might not know about it, or it's just because there aren't that many companies working on humanoid robots for dealing with people.

>>8271 I'll tell you in a spoiler. It may or may not be off topic to the forum. Test

>>8271 It's a my little pony robo-wAIfu. Many anons that I love seriously need one and would spend thousands of dollars to have/build one. Other aspects of the project (voice, behavior) are nearly complete by other anons, and in development as we speak. Someone or some people need to build it eventually, and I think I fit the bill.

>>8295 Ponys are perfectly fine here Anon. Good project goals, I'm sure we wish you well. Please share all your progress here with us. I would suggest keeping it all contained within a single thread so anyone offended by it can simply hide the thread. >=== -add thread suggestion

>>8296 Great, so back on topic: what kind of gear train would be best for a bldc that can run at many thousands of rpm?

>>8304 I'm not a mechanical engineer myself (we have at least one here), but I suspect there is more than one company providing solutions for this type need. Electrocraft seems to be one. www.electrocraft.com/

>>8304 They don't run that many rpm as soon as a load is attached. If it has a high rpm will need more reduction of course. My point is more about looking into these flat, and sometimes huge, outrunner motors with many poles. Those are often being used in drones or helicopters. There are also those which already are mini cheeta servos, but those might be very expensive and have the wrong size. I'm thinking, for experimenting I might get something cheaper and work myself up. One with Moteus controler: https://youtu.be/R2uuTfuyadk Blackbird Bipedal robot, by Gabreal Levine: https://youtu.be/Xf3DfE1KrwI Here is more on Strain Wave Gear from >>8221: https://youtu.be/xlnNj9F37MA

>>7951 >>8270 I finally stumbled over a video where all types are explained, including the difference between bldc and steppers: https://youtu.be/I2_-etus0KQ

>>1449 >Slow artificial muscle fibers >Technology doesn't scale Using one straight wire alone doesn't scale well due to the limiting heat dissipation. The heat transfer (surface area exposed to air) decreases at the same rate that the radius of the wire increases. Three ways to overcome this in order of least to greatest cost/complexity are: Use multiple thin wires in parallel, add coolant (water or phase change), use a NITI precipitation-hardening alloy (higher max stress = lower wire mass). Also, Nitinol wire has a limitation of only being extensible with 2-5% initial length recovery. Coils solve this problem by loading the wire in torsion, with the extended length dependent on the coil geometry. All this being said, this technology has the greatest potential but has a huge problem with repeatability. It's strengths and weaknesses alike come from being an active material which changes properties with cycles and loading conditions. TLDR: You are better off sticking with an electric motor for enough accuracy to use modeled movements for now, and nitinol for face/fun bits which don't have to be accurate.

>>8344 >The heat transfer (surface area exposed to air) decreases at the same rate that the radius of the wire increases. >not using monatomic wires for all our robowaifu needs <shiggity J/K. I had an interesting traversal through wiki articles learning about this. Have any education recommendations Anon? >NITI precipitation-hardening alloy Neat, I'm learning more still. >and nitinol for face/fun bits which don't have to be accurate. I would suggest that facial may in fact be literally the highest need for accuracy in the entire robowaifu system. At least the externally-visible motions. For example the eyes and eyelids are notoriously important to get just right in animation to bring the characters 'to life'. It's a remarkably subtle business tbh. Other areas on the face/head might need less precision, much of the cheeks, neck, and brows for instance.

>>8344 I'm going with dc motors, magnetic switches, and such, pneumatics and maybe some nylon. Maybe piezoelectric actuators. Nothing has changed. I also think that pneumatics are good keeping a body in a static position, with no use of energy, while being able to dampen some external force.

>>8347 Mini cheetah motors have support behind them and plenty of guides to get set up. The motor is 370$ on aliexpress and the boards that run it are probably not that expensive. These motors are the best viable option to get into reasonably strong, realistic movement at the present. The motors are bldc with a wide airgap and are quasi-direct driven (4 or 6 to one planetary gearbox) for minimal backlash. Do your own research to learn more about these motors.

>>8362 These mentions of mini cheetah are from me: >>8307 >>4890. There's also >>4898. However, I only see them being interesting for the hips and maybe for the shoulders. We'll need more motors, and also cheaper ones for learning, testing and prototyping.

>>8363 Holy shit I can't read. Good stuff, anon. Good luck on researching this stuff because my mech engineer brain can't understand it. >>8345 Attached is the general idea, and that is the subject of research which I am doing in person. It is from: https://ieeexplore.ieee.org/document/4412462

Quick reminder on the strength of printed gears: https://youtu.be/b6cDOyxub8w

>>8404 Forgot the link to the file: https://www.thingiverse.com/thing:2375916

>>8404 >>8405 Thanks!

>>8363 Added the Boolean OR logical operator to waifusearch for a convenient consolidated listing for various spellings (there are 3) now. Here you go Anon: >>8415

Actuator and muscle related: >bearings, ball-bearings, stainless steel bearing balls >>8364 and following >electically activated polymers >>8502 and following

Eccentric-Cycloidal gearing has been discovered a while ago, at least 10 years, but we have NO comment on it. Till now! Well, if you looked at the InMoov neck mechanism, or even printed it, then you already saw something like it. These can be worm-like 3d printed parts or disks. Here some videos: Basjc example: https://youtu.be/AMtyFwMDL7w Rack and pinion: https://youtu.be/g-CmpPBrFtE Reducer: https://youtu.be/XrSilJGHBCQ Comparison: https://youtu.be/kq-tHHwgGwU Channel: https://youtube.com/user/ECgearing Website has news on it, and more vids, though they're on Russian or Ukrainian: http://ec-gearing.com/ Here's a article from a company using it, though not the 3d printed version, but the kind which is made out of machined discs. I only read parts of the article, yet. https://www.sogears.com/products/manufacturer/31-eccentric-cycloidal-drive.html Then there are eccentric planetary gears, here done in wood, but would also work with laser cut or printed plastics: https://youtu.be/P_pNqpdLV-8

>>8718 Thank you for bringing it to our attentions Anon!

Related: >>8742 - It's about a silicone rubber printer for muscles and such.