>>22819 lol I just reread that eating and drinking wtf

>>22820 okay got a completely different answer from chatgpt Arranging these tasks from easiest to hardest: 1. Robot skin: Adding a protective or decorative covering to the robot's body is relatively straightforward and doesn't involve complex mechanics or functionality. 2. Robot mouth: Constructing a robot mouth to mimic basic movements, such as opening and closing, can be achieved through simple mechanisms like servos or pneumatic actuators. 3. Robot hands: Designing and building robot hands that can perform basic grasping and manipulation tasks requires more complexity and precision, but it's a well-explored area in robotics. 4. Robot arms: Developing robot arms involves creating articulated structures with joints and implementing motorized actuators for movement. It requires a higher level of mechanical design and control. 5. Robot legs: Designing and implementing robot legs for walking or locomotion involves complex mechanisms for stability, balance, and coordinated movement. It requires advanced control algorithms and sensor integration. 6. Robot head: Building a robot head involves integrating various components like cameras, sensors, actuators for facial expressions, and potentially speech synthesis capabilities. It requires mechanical design, electronics, and programming skills. 7. Robot balance: Achieving stable and dynamic balance for a robot is a significant challenge. It involves complex control algorithms, sensor fusion, and precise actuator control to maintain equilibrium and adjust to external forces. 8. Walking: Developing a walking robot is one of the most challenging tasks. It requires a combination of mechanical design, kinematics, motion planning, and control algorithms. Achieving stable and efficient locomotion across different terrains demands advanced techniques like dynamic balance and gait optimization. Please note that the difficulty of each task can vary based on the specific requirements and the level of complexity desired. This ranking provides a general idea of the relative complexity involved. chatgpt>>>>>>>>>>>>>>>>>bard

>>22821 one last thing we really should give that anon that made the robot head some props considering its so up there. Hope he didn't leave.

>>22808 Thanks, but we have a whole thread with making overviews, planning, diagrams, and subtthread does something quite similar what you're doing here: >>9554 - breaking down the development into smaller steps. Also, please consider the LLMs are likely not better than his board in knowing things in regards to robowaifus, since sadly we're somewhat the experts on this. On top of that, they're libtard censorship challenged, which means they might not know a robowaifu is supposed to be something like woman, what a good looking women is supposed to look like, and so on.

here's something I've thought up. It'd be rough and rigid but it'd get the job done I think.

>>23177 Please work a bit more on your ideas before you post them, and add a description.

>>23180 omg dude chill. What I just posted is fine. Clearly you never played with mechanix as a kid. Its not ideal but ideal will never see the light of day.

>>23180 Do you WANT this board to die? I get the drawing is completely childish, but your post is completely snobbish. Encourage the newcomers ffs.

>>23229 Adults often times draw on napkins to express their ideas too ehem

Related linkage mechanism, not only for walking >>24791

> HRC Model 3 Bipedal Robot IK Test by Letic Z https://youtu.be/i6j0hg5ZIPE

>>26403 Neat! Thanks NoidoDev.

Attached is an inside view of the "Femisapien" toy from Wowee in the mid 2000s. The leg mechanism uses only 3 motors, with the ankle synced to the hip via bowden cable and hip linked to knee via parallelogram. 1 motor to move hips side/side, and one motor per leg to move them forward/back. Demo Vid: https://www.youtube.com/watch?v=rNezxoSEEB0

>>26432 Mark Tilden is a genius tbh. (>>10257) There's a concept called Neuromorphics. It's basic tenets have in some cases been solved in large degree simply by clever designs using standard analog electronics. We can do with a bit more of that style of insight here on /robowaifu/ IMO. Cheers. :^)

>>26432 Femisapien is even simpler/stupider. That's just a floating steel rod,not a bowden tube. There's nothing that moves her hips. She is underactuated, relying on her bodies dynamics to move her hips. Those springs help to reduce needed energy to shift her upper bodies mass over to one leg. This is achieved by throwing mass (her arms) to alter her upper bodies center of gravity, which causes her to tip with the help of those springs. It's a great demonstration of Keep It Simple Stupid, reducing costs to an absolute minimum by using dynamics to simulate the functions of servos that aren't there. Her arms being able to turn her elbows, move the arm up and down, inside and out, all with one motor and potentiometer each, is also clever. Distribution of mass to help even out the torque needed to move her legs across their arc, and attention to her periods as a system of pendulums is worthy of note.My all time favorite biped. There's so much to love about her design.

>>26502 >She is underactuated Clever design work. After all, the simplest, most robust parts are the ones that aren't there! :^) Same is even moreso when it comes to software. As mentioned here (>>26447), if we can use simple analog electronics parts to fill in for what otherwise would need complex software to achieve, the parts attributes above can be expanded to include faster as well. Thanks for the nice breakdown analysis Kiwi. I wish the two of you had physical copies of Femisapien so you could really do proper teardowns/reverses. Cheers. :^)

Human Moves, Robotic Grooves: How AI Mimics Human Motion in Stunning Detail https://youtu.be/3LkWydJq6y0

>>26503 I do have a femisapien. Had to pop it open to fix the arms. No, not going to open 'er up again unless necessary. If I did the husbando Joebot would get lonely. Femisapien is the peak of the Wowee toy line when it comes to leg movement-of course RS media takes the cake for customization (which I also have). Everything after -like her husbando Joebot- is downhill. I think I got the order right idk)

I was looking at G-code for 3D printers and found this page "alternatives to g-code". https://reprap.org/wiki/Firmware/Alternative#alternatives_to_G-code In it they describe something I was groping towards here and in subsequent comments. >>21602 They say that, "... Much of the ""Firmware - experimental" RepRap discussion forum discusses protocols that are hopefully easier to interpret than G-code. splines and circles: By describing every move in the form of curved cubic splines, not only do we reduce the volume of data required to describe a curved part, but it may also make the microcontroller firmware simpler and faster. Faster: only integer arithmetic required. Simpler: the firmware only needs a single subroutine to handle cubic splines, rather than several subroutines to handle straight lines, circles, arcs, and the involute curve used for twice for each gear-tooth..." Curved cubic splines! And about them, "...Less volume: 8 cubic splines per circle are more than adequate for high precision CAD/CAM machines[5], while apparently (?) the current RepRapGCodes requires hundreds of tiny straight line segments to approximate the same circle. Elegant multispline motion controller and RepOlaRap and Repic5D mention such splines. Elegant multispline motion controller "will not use G-code. It will use a custom language based on cubic Bezier curves. This allows for much better description of arcs and will result in much higher quality prints with a much lower data throughput requirements."..." Some others. Any of you ever heard of Don Lancaster? He did articles on electronics and postscript. I think he came up with the "flutterwumpers", "... Flutterwumper Library generally focuses on generating shapes in PostScript format, and then converting directly from PostScript to "step, direction" pulses, without ever going through G-code. ".gcode versus .s3g" discusses the ".s3g" format file produced by replicatorG. It also mentions that USB seems to have high latency..." and etc. This is some good basic, foundational stuff. At some point all of this will have to be used in some form. That the high precision CAD/CAM machine use 8 cubic splines per circle tells you that it's likely it's close to optimal. They have thought deeply about this and all these machines started life severely compute constrained compared to even low power microcontrollers of today. So at the very least it should be close to being adequate for anything we could come up with. The tool heads of these use the same sort of 3D motion that I talked about for programming movement for limbs and body parts.

Search down on this page for a lot of references to "robotics". https://www.tinaja.com/flut01.shtml#hexapods Don Lancasters site. He has a ton of stuff on there that's related to robotics and control of.

I'd be willing to stick wheels under its feet. That's not a problem for me. What I'm not willing to compromise is in its ability to squat. I also do not want to spend thousands of dollars on actuators. I assume what matters in this case is the knees movements rather than the hip. Thoughts?

>>27887 So this is very crude but like the say. Perfection is the enemy of progress...

>>27892 Did a 3d print to illustrate the point. Should of made it bigger... Im going to have to make it bigger and do the whole thing in a small scale i think.

>>27903 Good ratchet design, Anon. Do you think you'll go ahead and finish it up with a proper housing too? Good luck with your project! Cheers. :^)

I changed my mind, i'm thinking this one. The can be spread apart on the sphere. The ms paint drawing didn't get the point across so here you go.

>>28255 I think id make the gears the same size. I don't know why i made the main one larger really.

>>28256 Okay so that leaves the problem of one leg being up at all times. However. I'm going to keep the solution a secret maybe. Unless you guys can find the solution.

>>28258 okay no i got it. Add that to this. The sphere is placeholder for how to make the legs stretch out.

This paper focuses on a quadrupedal robot but the findings are also applicable to bipedal robots. Just imagine what will be possible two papers down the line. Agile But Safe: Learning Collision-Free High-Speed Legged Locomotion >Legged robots navigating cluttered environments must be jointly agile for efficient task execution and safe to avoid collisions with obstacles or humans. Existing studies either develop conservative controllers (< 1.0 m/s) to ensure safety, or focus on agility without considering potentially fatal collisions. >This paper introduces Agile But Safe (ABS), a learning-based control framework that enables agile and collision-free locomotion for quadrupedal robots. ABS involves an agile policy to execute agile motor skills amidst obstacles and a recovery policy to prevent failures, collaboratively achieving high-speed and collision-free navigation. >The policy switch in ABS is governed by a learned control-theoretic reach-avoid value network, which also guides the recovery policy as an objective function, thereby safeguarding the robot in a closed loop. The training process involves the learning of the agile policy, the reach-avoid value network, the recovery policy, and an exteroception representation network, all in simulation. These trained modules can be directly deployed in the real world with onboard sensing and computation, leading to high-speed and collision-free navigation in confined indoor and outdoor spaces with both static and dynamic obstacles. Project page: https://agile-but-safe.github.io/ Paper: https://arxiv.org/abs/2401.17583 It's not 100% collision-free but their system can work on icy snow, bear a 12-kg payload (equal to its own weight) and withstand perturbations. It can reach a peak speed of 3.1m/s and an average speed of 1.5m/s in cluttered environments while switching back and forth between the agile and recovery policy. Some key takeaways: >For agility in collision avoidance, the best practice is to integrate locomotion and navigation rather than decoupling them as separate subtasks. >A low-dimensional exteroception representation can facilitate policy learning and generalization. >Their system can adjust reward weights to trade off agility and safety. And some other interesting bipedal locomotion papers cited in the paper: >Learning Vision-Based Bipedal Locomotion for Challenging Terrain https://arxiv.org/abs/2309.14594 Which inspired them to use a similar low-dimensional exteroception representation. >Robust and Versatile Bipedal Jumping Control through Reinforcement Learning https://arxiv.org/abs/2302.09450 Which was mentioned as one of many examples for robust locomotion using an RL-based controller.

>>29059 Remarkable. Two things I would notice about the video and the graph. * 1a. Clearly, the ABS achieves higher-performance metrics. * 1b. The tighter 'spread' of the data points on the graph, possibly (probably) indicates a sounder, more-efficient solution to the problemspace itself, when compared to the competition. * 2. The (common) design of the actuated legs is an exemplar of one of my key design opuses (>>28664), namely: > "The designer must keep the very dense elements (such as the biggest of the actuation motors & batteries) inboard within the torso framework." These doggobots do exactly that, and is a fundamental reason this general design approach is so popular now: because it works, and it's energy-efficient. --- >exteroception Lol, that's a new one on me even though I've had the exact idea in my head for probably two decades now. :^) Top-quality post Lad. Cheers. :^) POTD

>>29766 Thread related, an idea that didn't work very well.

>>9053 >Unfixed center of mass in design: as in "megaman" robots (and the Roll example), center of mass is drawn to the Boots, which themselves could be weighted further. (for "intimate" times the boots could be taken off to reveal more human-like feet, when sitting comfortably or lying down, etc). Somehow came up with this exact same idea yesterday or so, so I might as well voice my support for it here. Seems like it would make it easier for the robot to balance if the center of gravity was made closer to the feet, even if it makes the bot 20+ lbs heavier.

Dont really have robotics or simulation experience but cheat mode Black boots with one rubber wheel per foot to walk in a manner giving illusion of walking with feet very close to ground. This can be done with wheel mid foot for heel strike mimicking or forefoot for using the heels as a stable ground and do a forefoot walk with front wheels dragging along and heel raised. There is a few options for more on monster girl side of design such as a dragging dragon tail for extra backward support with or without a wheel in it or a monkey tail tripod leg that walks as a third smaller supportive limb There is disabled approach too such as not walking at all using a wheelchair or vehicle or walks with cane, walker or crutches. Embrace tradition reject bipedal and crawls on hands and kneeds going "nyan nyan nyan" or "wan wan wan wan" :3 bipedal but not quite human mimetic Gyroscopic balancing is an obvious option but it also will add weight and likely cant be entirely relied on but can at least help in some situations if there is room for it. Similar to OPs mention of a deer girl what about a harpy with backward bending bird legs? A bit creepy but kinda cute maybe? "bladeless" fans using air propulsion to aid in balancing. By bladeless I mean hiding the blades for safety reasons. This has obvious down side of increased noise and likely is not that energy efficient. human bipedal Four legged animals like cats, bears, dogs, raccoons can walk on two briefly and humans can hop on one briefly so any bipedal needs to be able to handle that so maybe look at monopods and just double the legs. Problem with this design is it will be a very hoppy walk which might look cute till your foot gets crushed. But even if not monopod based just keep in mind maintaining momentary balance on a single leg will help with walking smoothly. I watched a few vids of bipedal robots closely and i see a common flaw due to my exposure to the barefoot footwear crowd. Bipedal robots often have stiff boards for feet mimicking shoes not the feet humans evolved with and either walk with a heel strike or a mid foot strike in short steps. Heel strikes are often hard impact to the ground and midfoot strikes while you technically can do them the increased landing surface on smooth hard ground will be noisy too. I suggest although most people dont walk this way and all primates walk heel strike i say why not try a forefoot strike walk? Or at least do a forefoot strike for jogging or running speeds. It is second most popular method for runners and many cultures this is their main running style barefoot or thin sandals and the like. What this does is provide similar movement as backward bending knees just different proportions of joints and the fastest bipedal bots have that ostrich design. This also allows for sensors in the feet to adjust step length for say approaching stairs and to feel the ground for stability before stepping forward. This walking method is already required for walking down stairs anyway and logically it will improve downward incline walking. It may increase tripping or slipping risk for upward incline walking though so maybe adjust walking style for that situation. Feet soles could be fit with thin flexible grippy rubber like that used in bouldering shoes with small treads like boating shoes to avoid wet floor slipping. In theory this should make for a more dainty walk and less noise on impact because the ankle joint will be like a shock absorber. The design of the foot should have movement more like HyperLeg by IRIM Lab Koreatech not a flat plank. https://www.youtube.com/watch?v=wLFCMwRvhVI

>>30137 I probably should have started with this in reverse order I wrote but i wanted easiest to harder methods as order. Looks like I didnt realize the Hyperleg was already posted in here so I didnt need to link to YouTube when referencing it.

Crosslink >>31550 >Small robot project http://zlethic.com/hrc-model-0/

Interesting project. Discusses the general topic of balance. In brief. https://www.youtube.com/watch?v=z2qh1hENfq8 > ( via : >>33233 , thanks Anon! )

>>27069 Have to post in this thread, because I discovered a video of a Japanese bipedal robot that has toes :D At the tail-end of the vid, of course: https://www.youtube.com/watch?v=QHjd-O8S0nI

Potential method for developing legs https://www.youtube.com/watch?v=i1xM61WBOBQ

>>33868 Thanks, Mechnomancer! Hope you're having a relaxing break rn. >toes I think many robowaifu designers may overlook how important the the ankle/heel/ball/toe flexion-complex is to good ambulation for bipeds. Essential. Cheers, Anon. :^) >>33876 Antie FTW! I loved that part of the story too. :^) >line-loaded, camshaft-driven That's actually a really interesting concept, Anon. Thanks for pointing it out. I especially like the fact the guy realizes how important it is to keep the 'outboard' weight down, and focus the mass inside the bot's torso. Cheers, Anon! :^)

>>33898 >relaxing break rn Not really, burned out a power converter in the non-waifu/mek project and faffing around with chonky atv tanktreads. > many robowaifu designers may overlook how important the the ankle/heel/ball/toe flexion-complex is to good ambulation for bipeds The flexing of the toes helps form the equivalent of the femisampiens leg parallelogram than allows the robot to push forward without being flatfooted.

>>33902 >Not really, burned out a power converter in the non-waifu/mek project Oh no! :/ >and faffing around with chonky atv tanktreads. Sounds pretty cool tbh. >The flexing of the toes helps form the equivalent of the femisampiens leg parallelogram than allows the robot to push forward without being flatfooted. That's a cool way to think about it, Anon. I'd love to eventually offer a Steampunk Sadie robowaifu model/kit, that looked full-on authentic OG 1800's retro. :D

>>33915 >Sounds pretty cool tbh. Yes and no. Yes in that I now have what looks like a miniature nasa shuttle mover. No in that I need to slap a buncha 3d printed parts on 'em to make the treads stay on their guide wheels (the yellow and orange bits). Without my mods probably would've been crappy to have on an ATV. Also need proper wiring techniques otherwise the splices heat up after a minute or so of driving lol. But figuring out how to digitally control a 2000lb motor is nice (never mind controlling 4 of them), because I can probably look into reciprocating walking mechanisms for lighter robots, like an owo version of zaphod daederik's steam powered man, or a spider-like platform for SPUD to ride around on.

>>33962 Dang, that looks too cool, Mechnomancer. Be careful on that thing, lol. >muh le 2000lb motor Lolwut? >or a spider-like platform for SPUD to ride around on. I'm going with the pony motif for dear robowaifu Sumomo-chan rn. It'll be her 'hoers', and she'll be it's rider. But even the pone will be able to climb up with her onto a wheeled zipper platform for scootin' around fast (for example, going with Master to the park for picnics).

>>33969 >muh le 2000lb motor >Lolwut? A 2000lb winch motor, each motor pulls 15A for a total of 180 watts.

>>33972 Oh, OK got it. Heh, I thought you were saying the motor weighed a ton. :^)

Bipedal character design. https://www.youtube.com/watch?v=7_LW7u-nk6Q

>>33988 related: https://www.youtube.com/watch?v=_0F3l-skLos

Thanks for all the YouTube links, but please add some some description when posting. Ideally more than one line. These videos have a text description on YouTube which can be copied. With ">" you can quote it, though make sure to put this in front of each paragraph when doing so. Cheers.