Building Humanoid Robot Arm (part 2)
Вставка
- Опубліковано 15 вер 2024
- I have finished the build of the affordable 3D printed robot arm. It has 7DoF, brushless motors and quasi direct drive actuation on all axis. It is highly back-drivable. The size of this arm is similar to the human size, thus it should be great arm for the humanoid robot. So should we build a humanoid with it?
The CAD and 3D printer files will be available through my Patreon page: / skyentific
Special thanks to my special supporters Cam Smith and Jack Wright!
Please subscribe. This will help me to develop other projects like this, to bring the bright future closer!
One time donation:
www.paypal.me/...
If you want to help this channel, please support me on Patreon:
/ skyentific
Instagram: / skyentificinsta
Facebook: / skyentificface
Twitter: / skyentifictweet
#DIY
Open source humanoid robot?
Yes of course we should do this!
Awesome seeing you again and... If Terminator teaches us anything is that the distance between the shoulders is never to wide.
Ahahaha... Is that what will be built, a Terminator?
I missed your posts! Keep going and good luck with your projects !
Thank you!
Thank you. Don't stop.
Glad you're back! Would be super cool to see this integrated with something like huggingface/lerobot. So far they're only doing pick and place on small objects on a table, but maybe you could pick and place the table instead with that beefy arm 😂
Did you ever get LeRobot to work? Its on my bucket list! 😊
Thanks for bringing so many talented people together! The comments are a knowledge base on its own already!
Congrats and keep up the good work! 😊
Hello, you are an inspiration! What you propose is the holy grail of the open-source technology world! I have watched your videos for a couple of years now, and I have often wondered if your current path would lead you to this precipice. I say yes, lets dive in!!! I will gladly follow you into this great endeavor!!! Love the content, love your approach!!!
I am a CAD designer who would love to help design the robot. 6+ years of design experience in Fusion 360. Also, have some knowledge of coding robots using ROS.
How's you get ros and cad training? Ros tutorials suck
Ros tutorials do suck, you eventually develop programming and peripheral skills sufficiently to be able to navigate past that hurdle
If you are interested, I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). Please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
@@Hoenhighn1 Self-taught on both. A few guys on UA-cam teach ROS, but you will need some understanding of Python, coding, and Linux. Check out Articulated Robotics. I started with reading books on ROS1 years ago and it took me years of fiddling with it on and off to understand any of it. I suggest The Constructs ROS tutorials. You do have to pay for it monthly. But, the classes introducing you to Python for robotics were beneficial for me.
For CAD there are so many tutorials out there. Just pick something you want to design and start. Once you run into an issue, google it and there will 100% be a tutorial showing you how to do it. Keep just plowing your way through it. That's the best way to learn.
@@Skyentific Email sent. Thank you!
Fantastic new designs... Just a little bulky but, I understand strength of material as well. Great Job and thank you for sharing. 😲🙏🤔
yes , I can build and program the head
Glad to see you are still active!
Yes of course we should do this!
really happy to see a video from you! congratulations on 100,000 subs!
Thank you a lot! And thanks for following me!
good to see you back!
Totally agree with you. I can help :)
Thank you. I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). Please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
Congrats on the 100k subs. You earned it, great job.
Thank you a lot!
ayy congratulations on the play button!
Thank you!! 😁
This is a masterclass of advanced machine building where we all can learn from
Поздравляю с наградой!🎉
Thank you!
I'm very interested in helping to make an open source humanoid robot. Actually it's very exciting to find someone else interested in the same goal. I would be interested in working on the software for the robot's motion.
Do it! Power to the people!
I like the colour choice , would be perfect at Ikea 😉
Exactly!
yes of course, we wish to see all milestones of this project.
It is yet to be decided. I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). If you want to participate, please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
I finde your idea to make open source humanoid robot very interesting but also challenging. And I'd like to participate in the project as mechanical designe/production & embedded systems engineer. And I love the colours, they are my favourite! 👍😉
Thank you! If you are interested, I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). Please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
Whats your field? I am also working in automation as a software developer for PLCs 😊
So cool, I’d love to try this one day
I am on board with your idea of open source humanoid robot!
I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). If you are interested, please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
@@Skyentific Email sent!
о май гадбл, ты жив ))
Open-source = NOT locked behind a paywall. K-Scale Labs, Anthrobotics, and PointBlankLLC are all truly open-source humanoid projects.
My robot arm is not yet open source. I spent more than 3 month to build it. With zero investors. Spending my own money on this. Now I have exactly zero savings left. And still, I am ready to make it open source if enough people would find the idea interesting and will help.
At least right now everyone can get my design for only 10USD. I looked at the project that you mentioned and I have only found CAD file for K-Scale, and only for there small robot. If I am wrong, please send me the link, I would be glad to check them (you can find my email in the channel description).
Hey, could you give more info on this? Because I did not know any of these yet! I found K-Scale Labs but the other ones do not seem to have a website and or are registered with multiple domains! 😊
yes we should do it
oh it's interesting, I'm not sure how I could help, but if I see a way I'll reach out, otherwise count me in for the ride!
Thank you for the video, I'm interested and pumped for this.
Thank you. I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). Please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
It should be done.
Yes!!!
For sure we should be doing this . I'm in :)
Cool! And thank you for your long time support. I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). Please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
Another great robot design video.🎉🎉🎉🎉.
Open-source!!! Obviously is a good idea among makers and tinkerer. I am working with industrial automation system designing & Embedded IOT design field i vote for opensource. And obviously wish to support.
Thank you a lot! I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). If you are interested, please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
Cool idea
Thank you.
Great idea! I'm in for the project! Actually I already started to do something, I am already tinkering with 3D printed or CNC prototypes, I am trying to do something very easy and cheap, using ESP32, STM32, Rpi etc, with ROS2 and MicroROS. I'm also experimenting with Liquid Neural Networks for the AI part, I have a beefy cluster for training at my disposal. If you are interested let me know! I am 100% willing to release in a free open source, will make some video tutorial too!
This is great! I've heard a lot interesting things about Liquid Neural Networks. It would be interesting to see how it works with robot arm manipulation. I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). Please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
Any website on your project? This sounds super interesting! 😊
YES!!! We should absolutely strive to develop an open source general purpose humanoid. With the number of companies that have almost fully realized that goal, we have plenty of inspiration and reverse engineering to do. I'm confident an open source humanoid is possible, but it would greatly benefit us as makers to have a standard to work off of. If we can all collectively agree on a standardized set of parameters for devloping a humanoid, I'm sure it's a feasible task
This is a great point. I think it would be hard to agree on all parameters, but we can start with something. And if necessary we can always change these parameters during development.
@@Skyentific Yes, the hardest problem with open source projects is that people isn't properly educated to collaborate. For example there are many other robotic arm projects out there but you decide to make yours instead of collaborate with them. Your arm design is too heavy, have too much torque, drains too much energy to work, is too bulky, so in resume is completely inadequate for a humanoid robot. Congrats, you have your OWN project.
OF COURSE WE SHOULD DO IT. I would like to work on the hand part! I think that you/we should start thinking about topology optimization in order to reduce weight and spend less money.
Perfect! I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). Please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
Hello, great ideea. I will be interested in this project. I’ve followed along since you started making gears.
I would be able to help to project on the software side. 5+ years of experience in embedded software.
Good job, and keep up the good work.
Perfect! Software side is something where I need a lot of help. I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). If you are interested please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
Whats your field? And how would you drive the arm? Some embedded real time Linux system might be good, although I guess any up to date laptop will be able to keep up with sending stuff thru the serial or can bus?
Let's do it!
great idea, I want to participate but I am a junior engineer
i want a robot arm in my kitchen that cooks for me
yes. you should proceed
top, complimenti! 🦾
I’ll build too! Let’s open source it!
If you are interested, I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). Please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
I will follow with interest. I don't think I can add anything to the project though. I'm wondering though about scaling it. Like how much can it lift with approx 10 motors like that in it. But interestingly it would still be interesting at a large scale. Like you never know. 🤔
Я думаю, надо добавить вторую руку и научить их завтрак готовить.
This would be perfect.
welcome back brother :)
Thank you!
Designing an advanced robot arm from scratch and then realizing there are issues with the way it would be held in place feels familiar to me. It's like designing a circuit and then realizing that how it gets put inside a box is a whole huge pain on its own.
Mannnnn good job tho i WISH you'd show your process for Making these in fusion, i've been trying to do something similar but am Terrible at fusion
It is complicated to show it. As the design takes a lot of time. And most of the designs I start, and at some point I realise, that I should use another approach. So I discard, and start new design. So for each published design I have 3 or 5 attempts, which I discarded for some reason.
great, im designing one too, except it goes on your back and straps to your arm
That's cool! What is application? Or just for fun?
I always wanted to do something like that! Kinda like a cybernetic arm!! 🦾
Awesome project. Especially thanks for blue-yellow robot 🇺🇦!
one practical application for cheap robot arm, yes cheap, easy to make, is to make a two-angle solar tracker, on a arm pole, up high. and needs to be fraction of the cost of the panel itself, say 100bucks for 400W panel. so at most 25bucks to the mounting assembly. or even less say 10bucks. or even less like 5bucks.
Wonderful initiative. I am the author of MachinaScript For Robots, a framework to control robots using Multimodal Large Language Models, so anyone can built Ai-powered robots in their garages. It is basically giving the control of a robotic mount to an artificial intelligence. The project is also openly available on github for anyone to try out, under the apache license.
A question: wouldn't it be more effective to use those MIT cheetah-like actuators for these parts, something like the Atlas 2 is using? Or is it too expensive? I believe for a full body robot these are the best ones, no? 3D printed parts would still be super important, as well as reinforcement learning models for walking, picking things, equilibrium as well as some cognitive reasoning capabilities using LLMs.
I would love to help on the development of the project!
How do you know that Atlas 2 use it? I would be interested to know this for sure. If you have reference please send it to me (you can find my email in the UA-cam channel description). And if you are interested in this project I can send you discord link.
MIT cheetah-like actuators unfortunately not cheap. They are becoming less and less expensive, but not cheap yet. Plus, most of them have quite bad drivers. Probably this is why most robotics companies build there own actuators.
I would like to check your project! Any website or github page where I can find it? 😊
Found it!! Thanks a lot!! This is awesome!! 😊
@@swannschilling474 hey swan, a quick google search ober Machinascript For Robots will show everything you need! Github and medium/hackernoom posts all there. 😉 Free the robots!!!
I love the idea of the open source project, any telegram, discord, slack or any other channel to connect for this?
I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). Please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
Im a engineer spezialized on power electronics and software for drives. Im able to design our own open source power electronics to drive each axis of the arm to replace the odrive. We implement open source software for field orientated control, the kinematic model, axis interpolation, path planning,...
If you are interested im available for this Project
Hey, I am still working on figuring out the kinematics of my SPM because I would like to switch to an embedded solution...would you mind taking a look at the project? 😊
Why not use large reduction like harmonic drive?
I'm concerned on the arrangement of the Noctua fans. They are "facing" each other and they're very close. They're basically fighting themselves. If you flip one around they will work together pulling from one inlet and blowing out the other.
Year, but in this case they would not cool drivers. Right now they blow air through the motors to the drivers. I hope like this they will cool both of them.
Humanoid time!
Yes!!!
I'm building a hand/wrist/forearm. Happy to throw that into contention for this open source project once I'm done.
If you are interested, I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). Please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
@@Skyentific I have sent you a message on discord!
3D printed gears seem kind of silly to me. Isn't it possible to find cheap commercial-off-the-shelf gears with less play and better longevity? Even injection molded plastic gears should be a big step up, and maybe even cheaper than 3D-printed? I also like the idea of using simple sheets of carbon fiber or aluminium in some places (you have used that in some other designs if I remember correctly), they shouldn't be expensive and can greatly increase stiffness I think.
Out of curiosity, how much time it took you to design and finish the affordable 3D robot?
For this arm it took around two month for design and around 1 month to assemble.
Hello, I am a college student who is making a robot dog by watching your video. I am currently trying to use odrive v3.6, tle5012b encoder sensor, and steadywin giim8108-8 motor, but I am having trouble with motor calibration. Could you give me some advice on calibration?
My advice would be not to use tle5012b. I had so much trouble with it. Unless, you know exactly what you are doing. I spend too much time trying to make it work. It finally works. But as5048 or as5047p is way simpler.
@Skyentific thank you! I will try it asap
I could not even setup the the as encoder! Is there any tutorial on how to do this? The ODrive 3.6 is pretty outdated but I am also still using it because it us reliable and there are good quality clones out there! 😊
I think we can keep this crude style for the first prototype of opensource and later optimize more of the stuff, because we want something that we actually have and can call open source humanoid robots instead of just a highly optimized arm. also how do i contribute to it?
I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). Please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
I just reached out via email, I'm very interested in this project of a humanoid open source robot.
I would love to participate
Your watermark doesn't match your channel name Skyentific!=Skynetific. But maybe that was on purpose? Funny how you check the spelling on the plaque... Anyway, go robot arms - I hope this sparks a robot arms race!
I wish I can say that this was on purpose :) . Thank you for pointing this
Hello , very good job for your 100k 👍👍🍾
How many kg 1 arm ?
What I'm super interested in is not only an open source arm but perhaps USB-C standard wired controllers and Machine learning without weird proprietary garbage. USB-C has PowerDelivery if you need 5,12,20,etc volta whats the discord or whatever?
Good point. Unfortunately, each motor of this arm requires at least 200W of power. The USB-C can only provide 60W. So it is not enough even for a single motor. And it is definitely not enough for the entire arm.
I would like to participate, I'm a mechatronics engineer.
I'm making a mobile robotics platform at the moment. Kind of like mobile aloha but much cheaper.
This is really great! One of the first things, which I want to try with such arms is mobile Aloha. This is really promising technology.
@@Skyentific I think your robot arm will lift more weight than the viperX robot arm. This is one reason why I was interested in your design.
Any github project or website I can check? 😊
Any github page or website on your project that I can check? 😊
@@swannschilling474 hi. I will put instructions, code and bill of materials on my GitHub once I’ve finished filming. Should be around two or three weeks. The frame and motors are from gobilda, the arms I’m using at the moment are hiwonder. Then I have an rpi 5 with a USB stereo camera, an ESP32, four cytron motor drivers and some lithium lipo batteries
How are you going to program the arms ?
Pleas let me know if you have any git library..... As I am also working on arms like this but at a small scale.
First, I would like to do the telepresence demo. This would alllow to check the capability of the arm, without complicated software.
Where were you, I hope you are okay, I thought has something happened to you😅
Wonderfull idea. why not joim forces with similar youtube channels like: jeremey fielding and James Bruton
Большое спасибо за видео по которому я учу английский
You are welcome! Thank you for following me.
What are the models of these main base engines?
If you ask about brushless motors, these are Multistar 9235 from HobbyKing.
@@Skyentificit’s discontinued fyi
It is the same as Eaglepower 8318
@@Skyentific thanks
That what she said 6:52 😉
How can I help ;)
I have made a discord (I am not sure that we will stay there, but I think it is a good place to start). If you are interested, please send me an email (you can find it in the UA-cam channel description), I will reply to you with a discord link.
@@Skyentific ok c’est fait
Можно было просто сразу на Русском говорить)))
UA-cam statistics shows that I have only 0.2% of Russian speaking viewers.
For some reason, the translator in Yandex browser thinks that the source language on the video is German 😄
Нормальный четки руинглиш
i was wondering why so many motors ,?? could this arm be done with almost half the motors??
also a robot head is a bad idea... you can mount a 3d camera in the chest and back also in the arms or hands
robots do not need heads
This arm has 7degrees of freedom. For 7 degrees of freedom you would need 7 motors. This is exactly the same quantity of DoF as in a human arm.
С таким английским можно озвучивать русских бандитов в американских фильмах :)
Hes bach
subebd
Мужик ты ведь Русский, хотелось бы и на русском твои видео тоже 🙏
этоже руинглиш. все понятно
Exactly! :)
You're soooo russian! 😁
Are you in the Ukraine 🇺🇦
No, but I was born really close to Ukraine. And I have a lot of friends from there. With the colour of my builds, you can understand who I support.
A YT video about the rise of HUMANOID ROBOTICS was released just yesterday by David Shapiro @DaveShap
, who is a great AI and systems thinker.
Maybe you could get in contact or interview with people like him, to get some valuable ideas about the perspective, framework and characteristics of an open source project like this.
Twotyer user lethic1 is your guy. He isbquite far in his process