Why Walking Robots are Easy
HTML-код
- Опубликовано: 30 окт 2022
- Go to establishedtitles.com/JAMESBR... and help support the channel. They are now running a Black Friday sale, plus 10% off on any purchase with code JAMESBRUTON. Thanks to Established Titles for sponsoring this video!
I’ve built quite a few walking robots in the past. If you head right back to the start of my RUclips channel there are quite a few large robots that could just about hobble along made out of wood and driven by windscreen wiper motors and other parts I could get hold of. I built these early in this century in the room of a house share I use to live in, which is why there’s a bed in the corner of the shot.
I’ve moved on quite a bit since then to some robot dogs that I’ve built recently, but a few years ago I built some bipedal robots like RobotX and my Walking Gonk Droid. These actually had a sense of balance, using an inertial measurement unit to read how far they were leaning over and make the motors respond accordingly to keep them upright. That can be quite tricky though and most of what’s on RUclips is the good cuts of footage where it worked just about ok.
I’ve taken some of these projects to events like Maker Faires, and you’ll always find one person…
So in this video I’m going to build that simple sounding robot which just shifts is mass to one side and moves the other foot to take a step.
CAD and Code: github.com/XRobots/SimpleBiped
You can support me on Patreon or buy my Merchandise:
***************************
Patreon: / xrobots
Merchandise: teespring.com/stores/james-br...
***************************
Affiliate links - I will get some money of you use them to sign up or buy something:
***************************
10% off at www.3dfuel.com/ - use code XROBOTS at the cart screen.
Music for your RUclips videos: share.epidemicsound.com/xrobots
***************************
Other socials:
***************************
Instagram: / xrobotsuk
Facebook: / xrobotsuk
Twitter: / xrobotsuk
***************************
CAD and Code for my projects: github.com/XRobots
Huge thanks to my Patrons, without whom my standard of living would drastically decline. Like, inside out-Farm Foods bag decline. Plus a very special shoutout to Lulzbot, Inc who keep me in LulzBot 3D printers and support me via Patreon.
HARDWARE/SOFTWARE
Below you can also find a lot of the typical tools, equipment and supplies used in my projects:
Filament from: www.3dfuel.com/
Lulzbot 3D Printers: bit.ly/2Sj6nil
Bearings from: simplybearings.co.uk/
Lincoln Electric Welder: bit.ly/2Rqhqos
CNC Router: bit.ly/2QdsNjt
Ryobi Tools: bit.ly/2RhArcD
Axminster Micro Lathe: bit.ly/2Sj6eeN
3D Printer Filament: bit.ly/2PdcdUu
Soldering Iron: bit.ly/2DrNWDR
Vectric CNC Software: bit.ly/2zxpZqv
Why not join my community, who are mostly made up of actual geniuses. There’s a Facebook group and everything: / 287089964833488
XROBOTS
Former toy designer, current RUclips maker and general robotics, electrical and mechanical engineer, I’m a fan of doing it yourself and innovation by trial and error. My channel is where I share some of my useful and not-so-useful inventions, designs and maker advice. Iron Man is my go-to cosplay, and 3D printing can solve most issues - broken bolts, missing parts, world hunger, you name it.
XRobots is the community around my content where you can get in touch, share tips and advice, and more build FAQs, schematics and designs are also available. - Наука
I started a Discord! discord.com/invite/fc6MedG7eW
Wow, you got a shoutout from thundf00t on you tubers who can make better robots than Tesla Motors👍🏻
Established titles is fake and a scam you can't actually change you title to Lord or Lady.
@@shaider1982 dude, that is probably an insult. Phil has musk derangement syndrome.
TESLA BOUGHT THAT ROBOT AND THAT ROBOT IS A COPY OF THE COPY OF ASIMO TESLA I DON'T BELIEVE THIS ROBOT THAT'S WHY IT'S NOTHING NEW AND IT HAS THE SAME DEFECTS UP TO THE SAME JOINTS AND THE SAME DESIGN ONLY THE PLASTIC EXTERIOR IS DIFFERENT AND THE COLOR
As a non engineer that has worked in the animation industry, I am very curious about why usually there is no rotation or torsion in the torso of walking robots, and why they don't use the arms to modify the center of mass even if they are not fully functional. Those two are insanely important for equilibrium and they seem to be completely underused
I would assume that it would have something to do with the flexibility of the hardware and the cost of manufacturing, could also have something to do with the extra weight making the robot less stable?
I need a robotic engineer to confirm this
not an engineer (yet) but i do have a good amount of knowledge about robots--my guess is that it's too challenging to design arms with enough weight to actually change the center of mass. powerful motors are heavy, so to add arms you need to use motors powerful enough to support the weight of said arms, but those motors are now super heavy themselves, which means the arms won't be as useful. as for torso rotation, some robots like atlas do use it, but it's generally easier to design a robot which doesn't need torso rotation rather than try to implement it
Typically they do try to use the arms and spine of the humanoid...but it's just another layer of inverse kinematics with multiple possible correct solutions and finding one before it falls over. (look at the boston dynamic clip at 14:55...that is very clearly rotating at the spine above the hips. The arms are pumping with the steps, but only enough to keep it straight. They would do more motion...but that would require the actuators to be even more powerful to justify needing more moment/cancellation of moment.
Because using arms to modify center of balance is hard. Like it takes humans a long time to figure it out as babies hard.
@@rudrecciah Because everyone ridiculously puts heavy motors all over the place rather than just putting them where they want the mass and using remote drive (spinning flexible shaft, cable drives through bowden tubes, etc) with lightweight stepup and reduction gearing at opposite ends...
I think people tend to underestimate or simply forget how critical and complicated feet are for walking
and that's just what they'll do
Yeah, even baby need to train a lot to walk
@@simpasalsunda7244 Only human babies, due to the fact that they still have a lot of brain to develop after birth.
I don't blame them, it comes naturally to us all and is something we do on a day to day basis on autopilot
@@DrWhom one of these days robot feet are gonna walk all over you
I really like this robot. I know it's slow and awkward but the linear rails and shifting mass make it's movement very unique. It looks like a 3D printer learned how to walk.
As long as they don't figure out how to open doors, we'll be OK.
Describing 2007 as "earlier this century" knocked me out cold. At 32 I'm freaking ancient. I was born in the late 1900s. 🤯
You're ancient?? Imagine how I feel being born 6 months before the 70's started 😂😂
You make me feel old.
@@Galerak1 Kid... seriously...yeah I'm talking to you Mr "I was born in 1969".
Lol.!! I got nearly a decade on you...guess how I feel...well great actually, the arthritis drugs work well and I can't remember if I was in pain half an hour ago, so life's good. 😝
@@NeilABliss You win sir 😁
I feel your pain, I'm 35 and it still fucks me up thinking that the year 2000 was 22 years ago 😂😂
It's great to see that your dad takes an interest in your work and even makes recommendations as to how you should accomplish certain tasks.
It's strange how you never picked up his accent though 🤣🤣
Should I attempt another Dynamically Stable Bipedal Robot project? This time I'd use the Quasi Direct Drives/Cycloidal Drives and Inverse Kinematics like openDog3.
Why not? I've seen you nail down the mechanical aspects of balancing and walking machines. As one of those commenters you aptly jest about, why not leverage some of that AI hardware you have laying around to improve the software?
Your dog designs are great, and yet your commercial and university rivals let AI/machine learning refine the IK models, PID parameters, and gait patterns for more effective movement.
You have the computer science chops to do it.
Imagine humans with broken legs, joint and muscle pathologies, etc. What makes them ambulatory is their software IN Spite of their physical imperfections. Keep it up, you're a great engineer.
Some thoughts on feedback were the machine learning idea interesting, in addition to simulations based on IK, mass, etc., you can also play with computer vision, kinect sensors, heck, even VR trackers with lighthouses to allow for a physical machine to be part of the iterative learning process.
You could do the crazy evolution with millions of generations, or simply chunk the problems and refine smaller parts of it over fewer cycles with training.
Absolutely! I find your approach of iterating your robot designs, applying what you've learned from previous projects fascinating. You have come such a long way with these desings, why stop now?
Maybe use your gyroscopic flywheel on rotator to handle the counterbalance and see how fast you can make it move.
Absolutely. Dynamically stable bipedal robots are the coolest thing. They can be quite high and I think they will be most impressive on shows.
Another fantastic challenge would be any robot that can elegantly walk up and down stairs and maybe open and close doors.
If you want to make something useful for society, you could consider a robot or machine that would be useful in elderly homes or hospitals. I think many people and schools/universities would be happy to see you endeavor in this direction.
The way you built this new bot is how I imagined the Gonk droid worked (in universe) when I was little. I imagined a chunk moving back and forth, or some kind of hinged hammer-like thing swinging side to side to make it balance.
This is such an amazing gonk droid mechanism. It really embodies its slow lumbering movement.
I prefer to imagine that (also in universe) the droids actually all have slave midgets inside them (except the big ones, those are full-size slaves) and the entire universe just pretends they're not even organic because it feels better.
This might sound weird but i truly hope the joy and wonder of making these robots, even one so "simple" hasn't been lost on you at all.
I can only imagine the ammount of joy gained from seeing a creations first steps
The behaviour at 14:30 actually looked kinda interesting - makes me wonder what would happen if your robot had a greater focus on throwing mass around up top in two axes to create rapid dynamic movements a bit like some of your balacing robots, paired with lighter fast-moving legs to "catch" and redirect the movement to run (or stagger!)
Look up advances in underactuated robotics, sort of similar
I love how passionate you are about your work, always a gem to see the level of ingenuity you put into your videos, CAD Designs, and Robots. Cheers!
Perhaps a good idea would be to mechanically link the leg lift with the weight shift - as the leg lifts it swings or rotates the counterweight to the opposite side of the machine.
You might be able to damp the wobble by using a simple Input Shaping algorithm. This shaping process splits the acceleration into two phases, where the second phase cancels out the vibration created by the first. As always, an insane amount of design work done in an insanely short time. Another great piece of work.
Can you imagine we live in an age where a 2005 tech prototype robot is used to pump the stock price of a car company. Not to mention its more improved and advanced successor had to be cartwheeled on stage by three people with a stick up its robotic butt. Amazing times we live in.
No, too far out there.
Didn't we have Toyota Asimo?
@@ingni123456 You're technically right there. So basically, Tesla went for stealing the dumb's kid homework thinking it would get all A's :p
Yep, the whole upper echelon of our civilization is a stinking cesspool of incompetence venality and corruption.
Go watch the AI day 2, and say that again. What can you do.
Some STM32 chips can use their timers to measure quadrature encoders, it has the advantage that it keeps track of them in hardware so you don’t need interrupts or to poll the pins, you just ask the timer what the counter is at whenever you want.
I am soooo glad you did this one. I put together a small tabletop version similar to this many years ago. Incredible to see one at a scale that would actually be useful for something. :)
Love your work James. You have no idea how much you inspire me on a daily basis
A great vid James. I found the detailed construction bit really interesting, showing how to integrate off the shelf components. Many thanks.
No compliment to your creativity and skills could ever be an exaggeration ! Again a very inspirational video, presenting the construction of a walking mechanisme as an easy thing. Fabulous!
This robot is fantastic! The movement is very funky. Just needs some googly eyes on the moving mass.
if you can drive the servos partially while one leg is lifting and a bit as it is coming back down, you could spread out the servo motion over a longer period and possibly make the turning less jerky (since the mass on top would rotate at a smaller speed). Great stuff!
there used to be a simple 2 servo bipedal robot, I would love to see your take on simple walking robots that could be made with toys like lego.
I made a very similar robot when I was 10 out of Legos - I wish I had photos. I love your channel and it inspired me to make more cool stuff of my own (I'm a senior mechanical engineer with a bunch of robotics experience). If you ever need help designing any particularly complex mechanisms, I would love to help!
Brilliant, love your enthusiasm.
It look surprisingly elegant, I really like the motion. Thank you again, for sharing the cads. :)
Love your mechanical designs! So much of these robots is about your creative solutions.
I'd love to see you ride a larger model around. Great work and fascinating as always!
the issue with the reaction force from high accelerations you mention at 14:27 could be reduced by using a trapezoidal or triangular motion profile where the acceleration and deceleration is limited rather than exponential smoothing which has high initial acceleration.
James, I'm fully convinced you could build a hollow Easter Island Moai Statue that walks by itself by wobbling side to side and rotating between steps. You could use reaction wheels to cause the statue to wobble.
That does it. The ATATs don't shift their mass laterally, and having seen this, I now call foul! Unless the Empire just stuffed a bunch of Storm Troopers in the main body and they fly around from side to side as counterbalances. That would be hilarious, and mechanically acceptable.
i really enjoyed that little comedy bit at the start james, keep it up!
Even easier, have two "meshed" forked legs that walk over each other, like many wind up toys. Personally I was unimpressed with tesla's reveal, and I have serious doubts about everything elon claimed. The robot they showcased walking was doing a static walk, just like this robot.
I'd be pretty keen to see you tackle a walking robot that's uses a control moment gyro, just like your one wheel balancing robots. I (and I'm sure everyone else) really appreciate your honesty and frankness with how things are going. You don't exaggerate what your robots can do and you show a lot of the R&D, and how you work step by step to fix problems and improve whats there. Elon could learn a thing or two from you.
Love the sped up edit. Between how it moves and the music reminds me of older cartoons.
Oh the Tesla platform was impressive but not for the reason people hype it for.
If they actually managed to build this in a years worth of time it is a considerable feat.
It looks suitable for mass production and should be a good platform foundation to be expanded on.
The static gait seen in the demo is mainly a software problem, and I'm pretty sure they toned it down a lot so it doesn't fall over for the demo.
Tesla's bot was boring. James' bots mop the floor with Elon's lame effort.
I agree with ya, so many youtubers and stuff are not up front with the limitations of their builds, so when someone is like james here it is quite refreshing.
@@DerSolinski right now robotics are mainly software limited. That is the really hard part.
@@DerSolinski they simply got people wich know what they are doing or did build a robot bevore.
This will be a case in how much money can be thrown into it. With enough funding you simply buy what you need. Not necessarily stuff but people.
That was such a nice smile when you were throwing that weight around 😁
Thanks! Nice that a simple design such as that functions. Yesterday the stepper motor on one of My robots "passed away". Those need to be lower cost and lighter weight, so I can buy a replacement without worrying about if it will get on the plane easily.
If it's just for left and right balancing, maybe a reaction wheel could do the trick ?
I've considered it, reaction wheel also cause a gyroscopic effect when they are rotating which causes some other issues.
That's a big bruise on your forearm, hope it doesn't smart too much!
I'm always blown away with what you can accomplish in a week. I made a few servo robots in college and it would take an entire semester of work to get one working XD
Great to see you back on walking bots. Applying your knowledge to a biped would be great, but please a small one would be far more appealing to other makers 👍
Your statically stable leg system would make an awesome star wars droid
Heck my Annie is 6 '8" the chassis is built out of dollar tree items and the project can stand but has no gyroscope and can also roll .which I think is more efficient in my opinion. And heck yes I want to see another Bruton walking robot .I love those builds .they have good ideas😁👍
Love the bright primary colours on that robot :D
I don't know what it is about this particular robot but I absolutely love it. More than any of the others you've built which doesn't make a lot of sense since this robot is a lot more crude than the others.
You are a bloody genius. I have so much respect for you. Keep up the great work!! 💜💜😍😍🥰🥰😊😊🙏🙏🙏🙏
A working AT-AT would be cool as well.
Very impressive !
Maybe ten years ago I built a prototype of a tap dancing robot called Paddy O'Furniture. I used Basic Stamp and pneumatics. It was pre-3D printing so the project was pretty primitive but I made a few observations along the way. One is that a tapping toe needs two axes of movement. The other discovery was that knees are also essential: tap dancing without them is pretty awful. :-)
If I ever see you at a maker fair I'm totally saying that to you!
ok you definitely need to do more impressions of people. "why don't you have it shift its weight while it lifts up the opposite leg? seems easy to me" rofl
Cool idea!
This is so cool!
As a technical productdesigner, I understood almost everything, and could really understand the reasoning behind the decisions.
AFAIK, human walk is anything but stable, therefore trying to keep the robot statically stable at every moment (i.e. you power off, it's still in equilibrium at any stage), while human walk is a very different thing: if you froze when you start running, you'd fall. Thus, here comes the MPU and a lot more maths so you combine movements together (such as up+forward, or spin+up), including movements of multiple limbs (arm opposite to the leg, etc.)
Also, a flat foot will never bring you too far. Try spinning half-hexagons instead of static feet, you may figure that you can use torque and forward motion to spin, and that springing your movements is just about controlling the instability mid-air...
Awsame perfect job, with the simplest mooving walking robot. A improovement is rotation, when you rotate pleae make it rotate faster with full rotary motchen, baybe with a pinion and sun-gear.
This is fantastic
Would be cool to see a setup like a delta 3D printer used to balance. The mass could be lifted and lowered to reduce sideways inertia. And only one mass would be required.
Would be funny to see this robot laying tiles on its path.
AWESOME!!!!!!!!!!!!!!!!
KEEP IT UP PAL!
Your "That _One_ Person" impression had me in fits!
Cool. Your into some funky stuff.
additionally id LOVE to see how you'd tackle doing robots that are a bit more abstract (or Geometric?) where everything is function for its form. like the TARS robot from interstellar, or some other sci-fi stuff where the robot is basically a cube or shape that has sections that fold/unfold
i love you, you're a great person!
I would be very interested in seeing your take on pneumatic muscles!
My mind is blown by these projects, I cant wait for robots
You are surrounded by robots already. Soon....you will join us
Excellent robot James, have you considered using a pendulum to shift wait from the center...that would reduce the width of the robot and may work faster.
What a legend, you grew a beard just for one scene in the video.
I wish you would still make more stuff with bipedal robots.
I'm considering it
@@jamesbruton Glad to hear I'm really looking forward to it. Your videos have really inspired me for my own project.
Tesla Bot will be available alongside the Tesla Semi, the Solar roof, the Hyperloop, the manned Mars missions, rocketships used as airplanes, full self driving, and robotaxis.
I can't wait!
Go with the mass reaction and use a heavy pendulum to throw the mass over and then make it run :-) Love your videos.
Always admire your approach of research, investigation and experimentation. Just curious how often you wait ages to 3D print something and then go "dash, that isn't right" (being polite). Kudos for sharing all your research and development with the audience. Si.
could you tell us how much time you take to make like each section? Because i know for you, this litterally no dig deal or its your "Job". But whenever i try to do anything it takes ages, so seeing your stuff is mine blowing. dedication👍
Really cool! I'm sure the bearded guy is so proud right now. Maybe a tiping gyro to lean sideways?
epic love your videos james
an idea i had to make this way lower tech is to have it so that the mass has an electrode on it and so does each side of the top which completes a circuit to activate the motors for the leg (which could probably be done using the same concept). which eventually activates a circuit to bring the mass back to the otherside
thats so cool!!!
Your Impression of a channel viewer was absolutely incredible 😂😂😂
Amazing 👏
Glad to finally see a guest appearance by Professor Noel Sharkey at 1:10.
I was wondering if there is a store where you get all the hardware (bearings, lead screws, motors, nuts) or do you source them from various different sellers?
Hm, that clearly explains why the full lean over to one leg is slow and impractical. (Especially thinking they wont just walk, but may need to move payloads.)
There could be some speed up by using pressure sensors on the feet to move the weight only far enough to notice either the supporting leg to not increase in load (took all the weight already) or get the leg that is being unloaded detect its not bearing load anymore.
Could even start the raise of the leg a few % earlier and use the weight carriage slow down to add extra force 'off the leg'.
Sort of using the inertia to use the force that made the walker fall to the side as extra speed.
Maybe try a big tilting lever with weights on top next time. That could move faster than the carriage and balance the mass better by going fast and farther than needed and then jerking back to cancel the innertia.
I feel like for a robot biped
You should actually make it's hips wider
And have mechanisms that smooth out the walk cycle to eliminate jitter. Taking a more feminine one foot directly in front of the other walk cycle with the sway necessary to ease into each stride and smoothly transfer center of mass in each step. Leaning into the next by rotating the hip. Just allow the knee to just slightly hyper extend and lock up when necessary.
Feet articulation and things that help the feet change shape to distribute pressure, or work together to form a sense of springiness, is a big deal
Listen it's not a kink
It will just help it walk and balance better
....robot feet 😳
This is how I built two-legged walkers out of LEGO when I was a child.
you just gave me an idea.
I like that you went with the "hold my beer" approach for lodging a complaint.
You can add a rotating weight, that can counter the inertia of moving weight.
14:38 Is that a gonk droid? lol kinda awesome
what if you reduced the mass but accelerated the speed of the action cycle, to make it go much faster
this way the weight shift would be caused by the inertia rather than the rest mass, you could probably even switch the rail at the top by some sort of pendulum for natural mechanical stability, though that would of course get in the way of any ability to turn, which the robot doesn't have anyways
anyways that's a fairly small modification, basically just redoing a third of the robot, which could make it walk much faster, just a thought
Looks like the beginnings of bad box art mega man. Are you the future Dr Light? Great vid!
Can you plz send me a link tell me where I can buy the DC motor and encoders you always use for your projects?
It looks so happy when walking
Goes to show how simple things like walking are so hard to recreate, eventhough you are doing a good job with your inventions.
Life is not created by chance.
Bearded critic James is a great addition to the channel hahaha!
he should get a 1-minute look-in at the start of every episode
I'm curious why you didn't interpolate the upward and forward movement of the leg and have it sinusoidally starting and stopping so as to minimise the jerk?
what about the belts and pulleys; do you have a link for those or are they from the same bearing site???
My question is and always will be: "But why?"
The human shape is so inefficient in doing literally anything. What is this robot actually made to do and what makes the human shape better at achieving that than literally anything else.
So far the only real argument I've heard is "Looky like human, coooool. Walky like human wow".
exactly.
I'd be interested if you could use this robot as a platform and write a more sophisticated control loop that takes inertia into account. In theory, you wouldn't need to wait for the mass to reach the other side if you could use the inertia of the mass decelerating to balance out lifting and moving the other leg in just the right moment, right?
You could decrease the mass by increasing the lever. If you use a rod with weights on both sides you would only have to move it a small distance.
With the turning, would it be possible to take advantage of the wobble to maximise the turn with timing, (or at least some measurement predictions?)
I have no equipment. However, if I were to build a bipedal robot, I’d certainly have a snake like spine in my robot. Batteries would be housed in the rib cage area, and quads.
Lots of hoses filled with oil, and a oil pump/air pump..
I’d also have a laser level/spirit level, with potentially an x/y/z axis cameras monitoring if the bot is level.
I’d try my hardest not to put any extra weight in the back area, like Boston have done. And keep the weight as central to all parts of the body.
The skeletal body itself id make out of titanium tubing. Where id be able to pump oil inside of the physical structure reducing the weight overall.
Also less weight means less strength required.
I don’t know, just saying what I’d like to do and experiment with.
Wow, looks and works amazingly but just wondering what it would take to add knees!
11:26 JUMP AROUND! JUMP AROUND! 🎶🎵 so that's what they were sampling for that sound effect in that song
can i ask what CAD software you are using? an can you recommend any other softwares that help in making robots