Perfectionism can be the enemy of progress -- a lesson I feel I have to re-learn a bit with every complex project! There is absolutely some magic sauce in maintaining momentum even if you have to pivot to keep it rolling. Awesome work!
100% this. Setting small achievable goals is so important otherwise you can get burnt out when you're judging yourself against unrealistic complex goals that should only be for the longer term. What I've found is defining goals in the first place can be a minefield as we tend to naturally set optimistic goals to try to 'encourage' us to work harder or because we think we 'should' be able to do that. Be kind to yourself and set very small incremental ones, then you'll have more fun and your accomplishments will always outweigh your setbacks. I've only just come across this project and its amazing how much progress Will has made!
Glad I don't have to worry about that. I'm so lazy that it impacts my grades, and even though I know it's important I don't care too much. Adhd is fun 👍
@@Nerd1704I don't know where you are in your schooling, but the elementary education system is designed to make good factory workers. Don't feel bad for not excelling.
@@Nerd1704that just means you’re above school. You gotta realize you have to make your own success because the boxes that other people get put in, doesn’t fit people like us.
As an orthopedics technician - I damn well love your enthusiasm and passion for this. Can't wait to see what our patients will be able to be fitted with in 10 years' time.
Consider using miniature stepper or BLDC motors and field-oriented control- this will give you “free” compliance via software-defined “virtual springs”. It will also eliminate the fiddly servo modification step.
From the moment I understood the weakness of my flesh, it disgusted me. I craved the strength and certainty of steel. I aspired to the purity of the blessed machine. Your kind cling to your flesh as if it will not decay and fail you. One day the crude biomass that you call a temple will wither and you'll beg my kind to save you. But I am already saved. For the machine is immortal. Even in death i serve the Omnissiah
He failed to complete the marble X and has cancelled the project per say, he really handed it off to others that may be able to complete it, he is now murmuring around with the idea of try to build another marble machine but its more a stage then a single device
Stayed subscribed but it's just to read the titles and see whatever nonsense is in his thumbnails, like keeping tabs on the world's most volatile mad scientist
That crazy number of servos perfectly demonstrates why it's so incredibly difficult to build a compact humanoid hand with all the same degrees of freedom as a real one. If only we could create tiny self-repairing muscles the way humans do!
The real trick will be integrating them into the human nervous system (and vice versa with the mechanical electrical systems) AND adding additional functionality beyond what a normal human hand can do, like making some sort of telescoping endo-strucure so people can go-go gadget that pickle jar that's just a little too high up to reach, or strangling your friend from across the room without having to get up when they blueshell you in mario kart.
Just a few years away imo with the advances in lab grown meats and flesh. Im not saying it doeant diagust me to say that, but it could be a thing soon in robotics too.
I’ve been watching this project unfold for nearly 5 years now, it’s so great to see it getting to this point, I’ve made my own models here and there but I really want to get in and make one when you release this, I want to make a rubbery “skin like” shell for it
Hey Will, I’ve been following your project for years now and have been making my own hand as inspiration. I ran into many of the problems you stated in the video and made sure to make those system as resilient as possible. For the processing power get yourself a Teensy 4.0 can be compiled using arduino ide and has an arm processor, second I got a 20amp power supply and have a voltage regulator set to 7.2 volts for the servos, a seperate regulator set to 5v to supply power to teensy and a third set to 3.3v for the multiplexing analog inputs. Never had a problem with power since. I’ve spiked up to 12 amps while using my servos, could lead to your arduino board temporarily resetting if there’s not enough power. Hope this helps in anyway, keep up the good work!
I'll take at least 5% credit for the extension spring idea haha. I'm confident several people have mentioned it besides myself, but it was my first comment on one of your videos and I've been following since. Absolutely astounding progress! Congratulations!
There are some (like myself) that from an early age recognize that it’s nice to be perfect, but it’s also nice to do the task in an acceptable timeframe. I call the balancing between these two concepts a fully-optimized approach, but given that human bodies have had far more time to ponder such trivial tasks with themselves, it’s worth recognizing those that have spent time to understanding exactly how these tasks are mechanically achieved. Bravo!
I have been following this project passively for I think a year or so and every time I am amazed. As a software developer who loves embedded engineering and Biology this just tickles me in all the right places
- Metal linkage is probably a *very correct* decision you made. - UHMWPE sleeves would probably improve wear survival and decrease friction of the sleeves, and you'd want smooth cables, not braided. - Try feeding the sleeves through a wrist channel.
Top marks Mann! Impressive sir. I admire your dedication and altruism. By opening up your shop and sharing your experience and knowledge you prove we have a reason to be here in the first place.
This is literally sci-fi type stuff from my childhood, and now it's just like a group of FOSS people printing out limbs. Lol. I know it's not that simple, but wow. Just amazing stuff you're doing!
ive been keepin up with this project since i was in like 8 h grade and im so glad you sstuck with it and finished it. cause it looks insane cant to see it better
so cool that you're trying to replicate the intricacies of hands always annoys me a bit when people say they made a robot hand but don't even try to replicate the movement actual hands
It's awesome to see how far you've come with this project, and to see how much the discord community comes together to help each other when they have questions. I wonder if there is a way to redesign the servo board so that it could handle more current and not have the magic smoke come out.
I've seen this design through my friends showing me your designs. I have similar ideas and it's now converging to your design out of necessity. Really, I've been thinking about how to reduce motors count and what's the best placement or motor shapes. But this design actually balanced everything out so well that it is the converge point of all designs.
I went through making a similar project, and soldering the microservos to make them the same way you did turned me off to it that I took a while to pick the project back up. The smaller motors kept overheating without a closed loop system in place to control the output current. It takes a PID controller for each individual motor to regulate output voltage and thus current for its position.
re you pca9685 power issue, break out the power to the servo's separately (keeping a common ground) rather than supplying via the PCA9685, only use that for the signal pin. There are boards out like the Gobilda servo power distro board for something prebuilt...or *shrug* make your own?
Don’t know if you will read this but the length of the 1, 3, and 4th digits on hands are smaller and if you shorten the digits and found smaller motors the power would balance out, also the distal portion of the palm moves but the fat pad there doesn’t so if you two pared the handle where the inside palm is the stable part you can use a joint system on the outside for let and right movement. Also gyro for wrist to a fixed plane horizontal may give you a wrist as well. Just some ideas from a medical minded person cheers
a significant portion of the lateral movement for the fingers is specifically the index finger and pinky, while the other two do move, most of what they do is merely conform to the bounds that the end fingers have set up so unless you want to specifically commit to making things such as the "live long and prosper" thing from star trek, you wont find the middle two fingers specifically forcing their movement that way seriously think about it and try to move your fingers for a second, aside from keeping a near equal distance between all fingers, even when you are typing, most of it is just your index, pinky, and thumb moving laterally, and when trying to grab things, you find yet again, there isnt much intentional lateral movement and even then, you can pretty easily do away with that extra movement with very small adjustments when we are talking about early robotics, its pretty cool to see fully articulated appendages, but we have yet to master the art of the two-pronged claw then its a matter of cost each individual servo is somewhere between $5 to $30 a non-insignificant price, especially once you start adding all of the control circuitry and the materials required to make the joints robust enough to be worth using in day-to-day applications, or industrial/rescue situations and even then, you are using custom servo control systems, and rather specific positional encoders but thats ultimately the beauty of prototyping, it doesnt have to be cheap and accessible immediately, it just has to work well enough to prove its a viable idea, and then you start changing it based on the application but thats also the achilles heel of prototyping, if its too expensive to begin with, it reduces the amount of people that can attempt it and give you feedback on how to improve it
I am so freaking stoked! I think the new metacarpal solution is good for now. I was struggling trying to get the coronas to fit. I'm using this hand to replace the Inmoov robot hands in my nearly 7ft tall android project. I'm making a silicone cast skin, and this hand fits miraculously well within that and the scale for the rest of the bot.
If you add more meat or a palm or some type of bracket for the fingers to have more surface to catch things on it’ll help it a lot, after all it’s basically a skeleton compared to our meaty hands that have better grip of objects
I'm so glad to finally see this reach its conclusion all these years later. It's always bothered me so much that even the most advanced robotic arms now aren't even remotely as dexterous as this. I so look forward to seeing this developed further and further until we finally get something actually worthy of being a replacement for the human hand!
Stumbled on this in my recommendations and it’s very cool. I’ve thought about building a hand of this nature and make ml based control system to operate it. Since I’m more interested in the control system, having a well designed hardware hand will be very helpful for me. Looking forward to the designs!
Glad you switched to FDM. Currently working on a hand of my own with a much different cable routing solution. I love watching your videos for inspiration. I'm keeping compression springs in mind for when it doesn't work out :p
For applications that require more power, I've had very good luck bypassing the voltage on servo driver boards- splicing the red leads directly to the power supply. I've also found controlling the servo pwm directly off the Arduino using the servo.h library, tends to be more reliable for higher load conditions for some reason. Arduino Due should have sufficient usable pins and clockspeed to handle that number of motors. Although most hobby style servos are intended to operate on 5v logic, I've found many models tend to work just fine with the pwm direct from the Due and other 3.3v boards without the need for logic level shifting. The downside of using this method is you only tend to get 1000-1500 "steps" per full rotation as opposed to the around 4000 on a 12bit controller like PCA9685.
There is a DLR robotic hand project and it is described in details in the book "Humanoid robotics A reference book" by Ambarish Goswami and Prahlad Vadakkepat. The big nuance that was found by DLR folks is that you need to limit freedom of movements of fingers in order to save power and weight. You see, when you bend your finger, because of shape of the bones and ligaments your finger can't move in any direction, except of un-bending. And if you have strong muscles for bending fingers you can oppose all of that unbending force. So, humans store pretty big muscles for bending fingers in their forearms and thus keeping hand itself lightweight.
I am so excited to see this progress! It's so easy for me to lose interest in a project but I find it really inspiring to see someone else overcome that entropy and create the thing they envision. I know things aren't just free but I'd recommend looking at a newer 3d printer. I recently upgraded from a decade old Taz 4 to a Bambu P1S and it has just so much more potential. My only complaint is that I have to revise tolerances on any old parts I've designed because it is so much more dimensionally accurate.
I ran into the same volt/amp issues with PCA9685 and found some room using the Pololu mini maestro 28 channel board. Just a though if you wanted a higher voltage/amp range across your servos.
I wonder if they even exist yet, but I think using micro hub motors would be perfect for using as joints in fingers. I think I will look into this more, maybe even a 3D printed version could work. I know torque will not be great, but each finger joint wouldn't need to have a ton of strength.
this reminds me of the hand in T2; which was indeed fully functional and puppeted. i wonder why prosthetics haven't reached that level, when a movie prop pretty much proved it could work? they got new prosthetics that have successfully fused human nerves into them recently, so i bet we are close!
for resin printing you really should try water washable resin it is almost just better in every way, about the same price barely any smell at least not a harsh smell and the obvious water washable instead of ipa. its still brittle so that wont change but I am not sure of the change in strength but it is just as detailed
I'd also try out the formfutura tough resin. Depending on exposure settings, it'll be realy hard (and slightly brittle) or permanent flexible and rubbery even after post hardening.
You can power the servo's directly, and only use the signal line from the driver board. Just make sure the ground for the board is the same as the servo ground. You could even use servo's with different supply voltages this way. Since the data input is PWM, you can get away with a lower voltage on the signal line than the power for the servo, but don't make the difference too big. Is it possible to use smaller servo's, or are these the smallest you can get that still are strong enough to be functional? I mean, I'm just interested for animatronics purposes, so for my needs they don't need to be that strong. But I'm still very interested in actual functionality, and how much force is needed ;)
Was looking to see if this comment had already been made. The way servo driver boards are marketed is a bit misleading and acts like its as complicated as controlling a DC motor or stepper even though all you need is the enough PWM signal lines for the actual control and the power delivery can be handled by something external and more chunky :) Common ground is important!
@@nestorovski1993 You can use an arduino to connect to the driver board, but there's no arduino that I know of that has even nearly enough IO pins for all the servo's needed. And even if it had, the PWM timing has to be spot on. You'd need an internal timer for each servo, and arduino's have AT MOST 4 reliable internal timers.
I think an important milestone to this project is the strategic requirement to fullfilling the hairy palm process. A community of testers in every adult entertainment video provider will surely accept this challenge and support further development. Greetings from an Engineer.
In a previous video you mentioned you used ultra high molecular weight polyethylene fishing line for the cabling. Has that changed / do you have a link or specific spec you use? And how do you manage to terminate each end while keeping tension? Cheers
The "springs on linear guide" cable tensioner thingy sounds like you way want to consider a 3D printed flexure, as you can do the spring and the constraint as a single piece
Ayyyy, I think the spring was my comment. You can run the PTFE through the spring, too. We get custom spring wound to fit the PTFE liner we use. Always a pleasure to see another video from you!
Hope this one gets a full tutorial on how to build it, I wanna set em up at my work bench too help since my current 3 fingered grabbers suck at holding stuff for me
I think it's really crazy just how much effort it takes to replicate the behavior of just one of our hands? It really goes to show how amazing nature is..
Cool stuff. I wonder about the purpose of the transistor on the servo board. Seems too small to be used as a switch for servo power, why did it blow? Hmm... Could just buss all the power connections together and just used the board for signal distribution.
why is it that you use a motor for each joint of the finger? that's not how our hands work, although i'm not an expert on this, but i think each finger has two bending type, from the base (which is connected to the rest of the hand) and one for the whole finger to bend inward on itself, which can be achieved with a cable mechanism but for the whole finger, i think this would result in a sufficiently good and accurate movement while reducing the servo numbers (and so you can use more powerful/bigger/ servos (sorry if my explanation was not fluent english is my second language)
Can the camera read the position of the artificial hand the same way it reads the position of your hand? If yes, you can calibrate the artificial hand in real time until its actual position matches the actual position of your hand. Let Unity try to adjust the coefficients to make the hands match.
Very cool video. I also Finished my first hand yesterday. I got some inspiration from one of your old videos. :) But mine cannot move in with the precision that yours is able to 😂
Maybe a recommendation for a decade from now, but look into electrorheological fluid as a replacement for the actuation system, should be the ultimate way to shrink things down I think. TiO2 nanofluid can be made with pretty low tech and a microwave.
The hand looks great. That said I think it's overengineered and too complex. Humans do non have explicit control of each and every joint via separate means, there is basically one tendon/sinew per entire finger motion. Removing excess complexity would allow for more robust and easier to control build.
Проделана невероятная работа, я впечатлён! Очень, очень сложный механизм. Я думаю, все-таки, в быту обычная двухпалая клешня намного функциональнее и проще в изготовлении. К тому же, если использовать этот механизм в качестве протеза, то к нему нужен какой-то интерфейс управления.
are you planning to add touch sensors on the fingertips ? ,,, would you kindly share with me information about how effective the extension spring cables by numbers if you have, and can you please share the supplier for this type of cables ?,, thank you very much
This video just popped up in y recommendations, so I don't really know all the history behind it, but congrats on getting the thing to work! P.S.: also some valuable general advices about keeping moving.
this plus a batter and Neural interface you Might be able to fit it to someone missing their arm. Some universe where a bunch of modifications were done.
maybe the compliece could be built ilnto the into the rotation of the pulley on the servo with a simple rubber bushing between the shaft and the pulley. it would tie the two sides togther. give some complience but hopefully be quite durable?
Perfectionism can be the enemy of progress -- a lesson I feel I have to re-learn a bit with every complex project! There is absolutely some magic sauce in maintaining momentum even if you have to pivot to keep it rolling. Awesome work!
100% this.
Setting small achievable goals is so important otherwise you can get burnt out when you're judging yourself against unrealistic complex goals that should only be for the longer term.
What I've found is defining goals in the first place can be a minefield as we tend to naturally set optimistic goals to try to 'encourage' us to work harder or because we think we 'should' be able to do that.
Be kind to yourself and set very small incremental ones, then you'll have more fun and your accomplishments will always outweigh your setbacks.
I've only just come across this project and its amazing how much progress Will has made!
Glad I don't have to worry about that. I'm so lazy that it impacts my grades, and even though I know it's important I don't care too much. Adhd is fun 👍
@@Nerd1704I don't know where you are in your schooling, but the elementary education system is designed to make good factory workers. Don't feel bad for not excelling.
It’s curse, what else can I say? In art class I always had to haggle the teacher to slow down just so I can get a thing right.
@@Nerd1704that just means you’re above school. You gotta realize you have to make your own success because the boxes that other people get put in, doesn’t fit people like us.
As an orthopedics technician - I damn well love your enthusiasm and passion for this. Can't wait to see what our patients will be able to be fitted with in 10 years' time.
Hell in 10/15 years you could have people changing natural parts for prostetics ones that works better than natural ones
With how slow and regulated progress in medicine is, I'd say in 10 years it'll be more of the same. More like 50 years.
Consider using miniature stepper or BLDC motors and field-oriented control- this will give you “free” compliance via software-defined “virtual springs”. It will also eliminate the fiddly servo modification step.
Modding off-the-shelf parts is just part of prototyping. should he or anybody else move it to production, custom parts can be made.
It'll also completely mutilate the strength of each joint
From the moment I understood the weakness of my flesh, it disgusted me. I craved the strength and certainty of steel. I aspired to the purity of the blessed machine. Your kind cling to your flesh as if it will not decay and fail you. One day the crude biomass that you call a temple will wither and you'll beg my kind to save you. But I am already saved. For the machine is immortal. Even in death i serve the Omnissiah
Jesus Christ… I just wanted to let you know that I’m on your side. Whatever side that may be.
1:52 This sentence pretty much sums up the guy behind the Marble Machine X.
Martin is going on side quests every other month
He failed to complete the marble X and has cancelled the project per say, he really handed it off to others that may be able to complete it, he is now murmuring around with the idea of try to build another marble machine but its more a stage then a single device
Stayed subscribed but it's just to read the titles and see whatever nonsense is in his thumbnails, like keeping tabs on the world's most volatile mad scientist
@@quantumblur_3145 I'm also subscribed in case he get his priorities straight again.
@unogazzy84 he's milking his patreon donaters. He knew if he completed the project, the content would dry up.
That crazy number of servos perfectly demonstrates why it's so incredibly difficult to build a compact humanoid hand with all the same degrees of freedom as a real one. If only we could create tiny self-repairing muscles the way humans do!
The real trick will be integrating them into the human nervous system (and vice versa with the mechanical electrical systems) AND adding additional functionality beyond what a normal human hand can do, like making some sort of telescoping endo-strucure so people can go-go gadget that pickle jar that's just a little too high up to reach, or strangling your friend from across the room without having to get up when they blueshell you in mario kart.
Just a few years away imo with the advances in lab grown meats and flesh. Im not saying it doeant diagust me to say that, but it could be a thing soon in robotics too.
@@cexeodus robotic flesh hand prosthetic, neat
I’ve been watching this project unfold for nearly 5 years now, it’s so great to see it getting to this point, I’ve made my own models here and there but I really want to get in and make one when you release this, I want to make a rubbery “skin like” shell for it
Hey Will, I’ve been following your project for years now and have been making my own hand as inspiration. I ran into many of the problems you stated in the video and made sure to make those system as resilient as possible. For the processing power get yourself a Teensy 4.0 can be compiled using arduino ide and has an arm processor, second I got a 20amp power supply and have a voltage regulator set to 7.2 volts for the servos, a seperate regulator set to 5v to supply power to teensy and a third set to 3.3v for the multiplexing analog inputs. Never had a problem with power since. I’ve spiked up to 12 amps while using my servos, could lead to your arduino board temporarily resetting if there’s not enough power. Hope this helps in anyway, keep up the good work!
A step closer to Venom Snake
Jokes aside, I can't imagine how people make this stuff. Incredible work.
Modeled in CAD. Then each part is 3D printed and assembled.
Just give it a good looking red shell cover and wear a eyepatch and BOOM!
You’re all diamonds
I'll take at least 5% credit for the extension spring idea haha. I'm confident several people have mentioned it besides myself, but it was my first comment on one of your videos and I've been following since.
Absolutely astounding progress! Congratulations!
Yah my ideas are better
Now -kith- *[fight]*
There are some (like myself) that from an early age recognize that it’s nice to be perfect, but it’s also nice to do the task in an acceptable timeframe. I call the balancing between these two concepts a fully-optimized approach, but given that human bodies have had far more time to ponder such trivial tasks with themselves, it’s worth recognizing those that have spent time to understanding exactly how these tasks are mechanically achieved. Bravo!
I have been following this project passively for I think a year or so and every time I am amazed. As a software developer who loves embedded engineering and Biology this just tickles me in all the right places
- Metal linkage is probably a *very correct* decision you made.
- UHMWPE sleeves would probably improve wear survival and decrease friction of the sleeves, and you'd want smooth cables, not braided.
- Try feeding the sleeves through a wrist channel.
My jaw dropped when I saw the finger splay 👏🏽
Top marks Mann! Impressive sir. I admire your dedication and altruism. By opening up your shop and sharing your experience and knowledge you prove we have a reason to be here in the first place.
This is literally sci-fi type stuff from my childhood, and now it's just like a group of FOSS people printing out limbs. Lol. I know it's not that simple, but wow. Just amazing stuff you're doing!
The thing is, it's overcomplicated still.
I've been following your project for years at this point and love to see your updates!
great to see that you're back onto this project!
ive been keepin up with this project since i was in like 8 h grade and im so glad you sstuck with it and finished it. cause it looks insane cant to see it better
I'm so happy how this project turned out so nice and did not stop at some point in the previous years. Great work!
so cool that you're trying to replicate the intricacies of hands
always annoys me a bit when people say they made a robot hand but don't even try to replicate the movement actual hands
It's awesome to see how far you've come with this project, and to see how much the discord community comes together to help each other when they have questions. I wonder if there is a way to redesign the servo board so that it could handle more current and not have the magic smoke come out.
I've seen this design through my friends showing me your designs. I have similar ideas and it's now converging to your design out of necessity. Really, I've been thinking about how to reduce motors count and what's the best placement or motor shapes. But this design actually balanced everything out so well that it is the converge point of all designs.
Was a follower since the first prototype and it’s incredible seeing how far this project has gone !!!
Looking good. Glad to hear you are still working on this!
Hey man. Great work! I was born with a limb abnormality and this is really inspiring.
I went through making a similar project, and soldering the microservos to make them the same way you did turned me off to it that I took a while to pick the project back up. The smaller motors kept overheating without a closed loop system in place to control the output current. It takes a PID controller for each individual motor to regulate output voltage and thus current for its position.
re you pca9685 power issue, break out the power to the servo's separately (keeping a common ground) rather than supplying via the PCA9685, only use that for the signal pin. There are boards out like the Gobilda servo power distro board for something prebuilt...or *shrug* make your own?
Don’t know if you will read this but the length of the 1, 3, and 4th digits on hands are smaller and if you shorten the digits and found smaller motors the power would balance out, also the distal portion of the palm moves but the fat pad there doesn’t so if you two pared the handle where the inside palm is the stable part you can use a joint system on the outside for let and right movement. Also gyro for wrist to a fixed plane horizontal may give you a wrist as well. Just some ideas from a medical minded person cheers
Thank God ! You're alive !
The mechanical modeling is just top notch man!
a significant portion of the lateral movement for the fingers is specifically the index finger and pinky, while the other two do move, most of what they do is merely conform to the bounds that the end fingers have set up
so unless you want to specifically commit to making things such as the "live long and prosper" thing from star trek, you wont find the middle two fingers specifically forcing their movement that way
seriously think about it and try to move your fingers for a second, aside from keeping a near equal distance between all fingers, even when you are typing, most of it is just your index, pinky, and thumb moving laterally, and when trying to grab things, you find yet again, there isnt much intentional lateral movement
and even then, you can pretty easily do away with that extra movement with very small adjustments
when we are talking about early robotics, its pretty cool to see fully articulated appendages, but we have yet to master the art of the two-pronged claw
then its a matter of cost
each individual servo is somewhere between $5 to $30 a non-insignificant price, especially once you start adding all of the control circuitry and the materials required to make the joints robust enough to be worth using in day-to-day applications, or industrial/rescue situations
and even then, you are using custom servo control systems, and rather specific positional encoders
but thats ultimately the beauty of prototyping, it doesnt have to be cheap and accessible immediately, it just has to work well enough to prove its a viable idea, and then you start changing it based on the application
but thats also the achilles heel of prototyping, if its too expensive to begin with, it reduces the amount of people that can attempt it and give you feedback on how to improve it
I am so freaking stoked! I think the new metacarpal solution is good for now. I was struggling trying to get the coronas to fit. I'm using this hand to replace the Inmoov robot hands in my nearly 7ft tall android project. I'm making a silicone cast skin, and this hand fits miraculously well within that and the scale for the rest of the bot.
First bionic hand I've seen trying to handle adduction and abduction. I'll be following this project.
If you add more meat or a palm or some type of bracket for the fingers to have more surface to catch things on it’ll help it a lot, after all it’s basically a skeleton compared to our meaty hands that have better grip of objects
I'm so glad to finally see this reach its conclusion all these years later.
It's always bothered me so much that even the most advanced robotic arms now aren't even remotely as dexterous as this. I so look forward to seeing this developed further and further until we finally get something actually worthy of being a replacement for the human hand!
There is an open source hand tracking glove, open gloves made by lucid VR that may be used as a better method for the tracking in this application.
This is awesome!! Staying put on a project for that long is also super impressive!! 🎉🎉🎉
With the way neurolink is coming along its great to see people developing this end of the tech that will help many people..
Stumbled on this in my recommendations and it’s very cool. I’ve thought about building a hand of this nature and make ml based control system to operate it. Since I’m more interested in the control system, having a well designed hardware hand will be very helpful for me.
Looking forward to the designs!
Glad you switched to FDM. Currently working on a hand of my own with a much different cable routing solution. I love watching your videos for inspiration. I'm keeping compression springs in mind for when it doesn't work out :p
For applications that require more power, I've had very good luck bypassing the voltage on servo driver boards- splicing the red leads directly to the power supply. I've also found controlling the servo pwm directly off the Arduino using the servo.h library, tends to be more reliable for higher load conditions for some reason. Arduino Due should have sufficient usable pins and clockspeed to handle that number of motors. Although most hobby style servos are intended to operate on 5v logic, I've found many models tend to work just fine with the pwm direct from the Due and other 3.3v boards without the need for logic level shifting. The downside of using this method is you only tend to get 1000-1500 "steps" per full rotation as opposed to the around 4000 on a 12bit controller like PCA9685.
Man, you're legend. This is soooo amazing.
Human anatomy, Embedded Systems Programming, CAD Mechanical design and Electrical engineering - all in one. Amazing feat!
There is a DLR robotic hand project and it is described in details in the book "Humanoid robotics A reference book" by Ambarish Goswami and Prahlad Vadakkepat.
The big nuance that was found by DLR folks is that you need to limit freedom of movements of fingers in order to save power and weight.
You see, when you bend your finger, because of shape of the bones and ligaments your finger can't move in any direction, except of un-bending. And if you have strong muscles for bending fingers you can oppose all of that unbending force. So, humans store pretty big muscles for bending fingers in their forearms and thus keeping hand itself lightweight.
I am so excited to see this progress! It's so easy for me to lose interest in a project but I find it really inspiring to see someone else overcome that entropy and create the thing they envision. I know things aren't just free but I'd recommend looking at a newer 3d printer. I recently upgraded from a decade old Taz 4 to a Bambu P1S and it has just so much more potential. My only complaint is that I have to revise tolerances on any old parts I've designed because it is so much more dimensionally accurate.
he has a prusa mk4 which is a real good printer, though somewhat slow compared to bambulab standarts
Thanks for sharing the updates. Great progress! Looking forward to more videos.
I ran into the same volt/amp issues with PCA9685 and found some room using the Pololu mini maestro 28 channel board. Just a though if you wanted a higher voltage/amp range across your servos.
Great work. It works so much better than everything before! Congratulation and thank you for sharing
I feel like the goal should be servo's in each knuckle we been doing cables to fingers since the 70's
Can’t wait for the full video!
I wonder if they even exist yet, but I think using micro hub motors would be perfect for using as joints in fingers.
I think I will look into this more, maybe even a 3D printed version could work.
I know torque will not be great, but each finger joint wouldn't need to have a ton of strength.
On fingers you can remove five of servos because one of the each joints on our fingers is comply the other one.
this reminds me of the hand in T2; which was indeed fully functional and puppeted. i wonder why prosthetics haven't reached that level, when a movie prop pretty much proved it could work? they got new prosthetics that have successfully fused human nerves into them recently, so i bet we are close!
for resin printing you really should try water washable resin it is almost just better in every way, about the same price barely any smell at least not a harsh smell and the obvious water washable instead of ipa. its still brittle so that wont change but I am not sure of the change in strength but it is just as detailed
I'd also try out the formfutura tough resin. Depending on exposure settings, it'll be realy hard (and slightly brittle) or permanent flexible and rubbery even after post hardening.
Would love a video on how you design parts, both conceptually and on something like CAD
Even though they are rare - i love the updates on this project!
You can power the servo's directly, and only use the signal line from the driver board. Just make sure the ground for the board is the same as the servo ground. You could even use servo's with different supply voltages this way. Since the data input is PWM, you can get away with a lower voltage on the signal line than the power for the servo, but don't make the difference too big.
Is it possible to use smaller servo's, or are these the smallest you can get that still are strong enough to be functional? I mean, I'm just interested for animatronics purposes, so for my needs they don't need to be that strong. But I'm still very interested in actual functionality, and how much force is needed ;)
dc motor and pot for custom position control is best using arduino
Was looking to see if this comment had already been made.
The way servo driver boards are marketed is a bit misleading and acts like its as complicated as controlling a DC motor or stepper even though all you need is the enough PWM signal lines for the actual control and the power delivery can be handled by something external and more chunky :)
Common ground is important!
@@SKYWURX If you look up the datasheet for the driver chip, you'll see that it isn't even made for servo's. It's originally a PWM LED board ;)
@@nestorovski1993 You can use an arduino to connect to the driver board, but there's no arduino that I know of that has even nearly enough IO pins for all the servo's needed. And even if it had, the PWM timing has to be spot on. You'd need an internal timer for each servo, and arduino's have AT MOST 4 reliable internal timers.
For example u can use 4 dc motors using l293d shield and atraching pots a0-a3.
You are a madman ! really glad I found this channel
I think an important milestone to this project is the strategic requirement to fullfilling the hairy palm process. A community of testers in every adult entertainment video provider will surely accept this challenge and support further development. Greetings from an Engineer.
Wow! What amazing work. Thank you for sharing this.
Aww mah gawd it all looks so good
In a previous video you mentioned you used ultra high molecular weight polyethylene fishing line for the cabling. Has that changed / do you have a link or specific spec you use? And how do you manage to terminate each end while keeping tension?
Cheers
The "springs on linear guide" cable tensioner thingy sounds like you way want to consider a 3D printed flexure, as you can do the spring and the constraint as a single piece
Great project, wish you all success in next designs
Better to have some working prototype then no one so congrats bro. Keep going for people that need it and for cyberpunk future for all
Thank you for sharing this is really insightful !
Ayyyy, I think the spring was my comment. You can run the PTFE through the spring, too. We get custom spring wound to fit the PTFE liner we use. Always a pleasure to see another video from you!
Amazing as always. You and your work continue to inspire.
Congratulations on the huge achievement 😊👍🎉
Cool, cool, cool!
Super cool!
Keep on keeping on.
I'm using onshape while watching this video! Thanks for Pro!!
Hope this one gets a full tutorial on how to build it, I wanna set em up at my work bench too help since my current 3 fingered grabbers suck at holding stuff for me
I think it's really crazy just how much effort it takes to replicate the behavior of just one of our hands? It really goes to show how amazing nature is..
Cool stuff.
I wonder about the purpose of the transistor on the servo board.
Seems too small to be used as a switch for servo power, why did it blow? Hmm...
Could just buss all the power connections together and just used the board for signal distribution.
why is it that you use a motor for each joint of the finger? that's not how our hands work, although i'm not an expert on this, but i think each finger has two bending type, from the base (which is connected to the rest of the hand) and one for the whole finger to bend inward on itself, which can be achieved with a cable mechanism but for the whole finger, i think this would result in a sufficiently good and accurate movement while reducing the servo numbers (and so you can use more powerful/bigger/ servos
(sorry if my explanation was not fluent english is my second language)
Are you using a PID controller for each axle? Might wanna check out MPC in order to get rid of the "wavyness" in each finger...
Amazing stuff as always! Keep that momentum going!
Can the camera read the position of the artificial hand the same way it reads the position of your hand? If yes, you can calibrate the artificial hand in real time until its actual position matches the actual position of your hand. Let Unity try to adjust the coefficients to make the hands match.
Can't wait to see the finished product with synthetic skins and all and maybe even a Dr. Octavia style extra appendages from Into the spider verse.
If you used linear actuators instead of rotational servos I think you would be able to compact things a lot.
Very cool video.
I also Finished my first hand yesterday. I got some inspiration from one of your old videos. :)
But mine cannot move in with the precision that yours is able to 😂
Maybe a recommendation for a decade from now, but look into electrorheological fluid as a replacement for the actuation system, should be the ultimate way to shrink things down I think. TiO2 nanofluid can be made with pretty low tech and a microwave.
I think adding rubber onto it it for better grip is a good idea.
Can i know what metal tubes have you used for passing the threads?
The hand looks great. That said I think it's overengineered and too complex. Humans do non have explicit control of each and every joint via separate means, there is basically one tendon/sinew per entire finger motion. Removing excess complexity would allow for more robust and easier to control build.
Проделана невероятная работа, я впечатлён! Очень, очень сложный механизм. Я думаю, все-таки, в быту обычная двухпалая клешня намного функциональнее и проще в изготовлении. К тому же, если использовать этот механизм в качестве протеза, то к нему нужен какой-то интерфейс управления.
are you planning to add touch sensors on the fingertips ? ,,, would you kindly share with me information about how effective the extension spring cables by numbers if you have, and can you please share the supplier for this type of cables ?,, thank you very much
Could you post the 3d print files for that quadraped fox robot I'm having a hard time finding tutorials sorry
This is very interesting I wish I test all the possibilities so people have a arm to do everything to draw write and do everything
This video just popped up in y recommendations, so I don't really know all the history behind it, but congrats on getting the thing to work!
P.S.: also some valuable general advices about keeping moving.
Awesome to see this progress!
C# is nice. Also, losing momentum periodically is part of the process.
This is the first video I'm seeing of yours but I think having rubber covers over the things and thumb would help with gripping things
I really liked your views with the creative process too!
Thanks for sharing, Will! Awesome work and dedication. Following.
this plus a batter and Neural interface you Might be able to fit it to someone missing their arm. Some universe where a bunch of modifications were done.
Millions of years of evolution and humanity is speed runing it
maybe the compliece could be built ilnto the into the rotation of the pulley on the servo with a simple rubber bushing between the shaft and the pulley. it would tie the two sides togther. give some complience but hopefully be quite durable?
also im sure there are some easy to drive high amperage servo boards floating around.
love this!! this is really inspiring and beautifle props to you !
Eu adoro ver esse projetos, já quero montar, o lado ruim é ue gastar dinheiro 😂