3D Printed Biomimetic Mechatronic Hand: Control Glove Explained



it's been about six months since I uploaded anything it's all on this channel not including that quick open evenin demonstration and for the past year or actually a little bit longer than that I've just been working non-stop on this Bionic hand project so some of you will have seen I uploaded a video recently that was just a demonstration of the hand and the control glove and a few different parts of it and that was intended for university open evenings to play and just give people an idea of what this project was I wanted to make a little bit longer video and just kind of explain some of the finer details so for anyone who hasn't seen any of the other videos this is a project that I did as my individual project in my third year at University and the idea of it was essentially just to try and design a Bionic hand that was as close to a real human hand as possible in terms of motion and ranges of motion in all the joints and different sort of grip types I could get into and so for that reason it was called the biomimetic there UNAC and as in it tries to imitate biology as closely as possible the main application of this idea being that if in the future we can control prosthetics as easily as we can control our own limbs then how closely would you be able to design a bionic hand to imitate a real hand in order to restore as much function to the amputee as possible or alternatively there's also applications and maybe hazardous materials handling or telepresence robotics so because there were so many degrees of freedom and joints in the bionic hand I also needed to design along with it a control glove that would measure the angle of all the joints in the wearer's hand and then use that to control the bionic hand as finely as possible this video is going to talk about the control gold mostly and then I'll power a separate video about the Bionic hand so what I'm going to do is just go through the video and try and explain everything that I can so it starts off with this sort of overly simplified visual explanation of how it works and it's designed just to be as as non-technical as possible just so that the most amount of people possible can understand it and appreciate it it's obviously missing quite a few steps for instance it's not as simple as just plugging the sensors directly into the Arduino and because there were so many sensors or was about 25 in the glove they have to go through a multiplex in circuit so essentially rather than send 25 individual signals down 25 different wires and they had to be put through a circuit that would alternate between four different signals at a time so it allowed the Arduino to have an input for all 25 senses and but only go through about four or five pins so it seemed like it would be quite complicated at first but the multiplexing say it's actually really simple but because there were so many inputs the soldiering of it was an absolute nightmare as you can see it was a big job to solder all these and it's probably a really horrific circuit board layout but it seems to do the job anyway it's in two way so then we go straight onto the control glove I think in my head I thought that this would be a really simple part of the project because obviously I've used potentiometers to control servers before and I thought I'll be simple just attach a potentiometer to each joint in the hand and then it'll be super simple to control the hand actually obvious is a lot more complicated than that if you need to actually get and if you need to actually allow the user to be able to use the hand dexterously then you need to mount the potentiometers in a way that they don't obstruct motion whatsoever so I made a little demonstration of how each drawing there's a potentiometer attached to a little jewel lever that goes over the top of the joint and allows you to move your finger with as much dexterity as possible and still be able to detect the full range of motion in the potentiometer and I think at the very least it looks really nice and it really looks cool when you see the movement of my hand while I'm wearing the glove so yeah I was really happy with the way looks not like really madness for this sort of project so I'd nail your design for this which used the kind of the bigger potentiometers that you screw in and ball in and that Elliott is I intend out to be a fair bit more robust because in the second iteration I wanted to make it a lot more compact and then it up using surface mount resistors which are a lot more flat and compact also really less robust so I think that mechanically the system works fine at least in terms of all the linkages and how the linkages move with the finger if you really think hard about it the rotation of the potentiometer as compared with the rotation of the person's finger and about the join you're looking at won't be entirely linear and in certain positions the rotation of the potentiometer will move more or less relative to the finger for example in this little clip you'll see that the first lever seems to move a lot more and a lot faster when the fingers nearly completely straight compared to when the fingers more been I think it would be possible to get a perfectly accurate angle of the joint based on the angle of the potentiometer if you worked it all out mathematic in mid like an algorithm in the Arduino but you just take up so much processing power and it was just one really necessary on this scale and this early in the development so it was fun really it was barely noticeable the discrepancy so mechanically I think it worked fine but the main issue was that these components were surface mount which means that they have really tiny little contact designed to be put on a surface and never moved so whenever you solder something and this sort is a little bit more brittle than the way I was before because obviously why it's flexible and salt is just like a solid lump of metal so when I initially started moving my hand around in the glove quite a few of the contacts broke because obviously the wires need to be flexible as the fans moving around and bought any movement of the contacts that I've soldered would just break really easily because there was small and brittle so my solution was to try and pop them in so you see on electronics sometimes they just fill up a circuit board with with resin to protect it from what are damaged or whatever I tried to do a similar thing using this two-part epoxy is epoxy sculpt actually it's just like a two-part epoxy resin that you can sort of mold with your hands and I just pushed it into these gaps definitely made some difference but it was still just not quite enough and a few contacts still snapped after had already done that so it made it a really sort of shaky foundation to be able to develop the software on top of but regardless as you see later on in the video it is working with with the Bionic hand and also with like a simulation in unity so all of the finger joints and half of the thumb joints just have one axes of motion so they're just a really simple hinge but the senton drink types in the hands are a little bit more complicated at that and required a different sort of mechanism so the first one to notice would be the joint that the knuckles of the hand these are called mCP joints and rather than being a hinged room they have two degrees of freedom listed my Bionic hand the motions actually a little bit more complicated to describe than just two axes pretending what slightly as the move from side to side but it's such a tiny little movement that it made sense to just simplify it to be two axes of motion so in the original clunky glove two potentiometers for the knuckle joint will mounted just behind it and larger dual lever went over the whole of the knuckle well then because the potentiometers when centered around the the origin of the movement in the knuckle obviously I had to use like a little ball socket drain and it was all a little bit dodgy whereas in the newer redesign I designed it so that the potential might could just be directly over the knuckle I wanted me to worry about the levers being out of alignment the sensor for the wrists used the same kind of idea with a dual leaver on top of a rotating base but it was just on a larger scale really as you can see there were no real problem with this wasn't the contact breaking so this is another area where I would need to design the point where the contacts attached to the potential I might have to be a lot more robust and still offer flexibility with the wire so I can move base of the film is something that I thought about for absolutely ages and I had a lot of different ideas of how I could do it I'm not entirely satisfied with this one that I went with but it does seem to work fairly well although you can see in the tests that it's not exactly perfect the first burn in the thumb that's sort of buried in your palm is covered in mopping in muscle all the way around so that means that as you move your thumb around the the shape of the shape of that area completely changes shape as the muscle contracts and extends so this made it really really difficult to try and design a way for a sensor to attach to it so what I eventually went with was a little section right near the knuckle of the thumb and right near the wrist as there's a can of the areas with the least muscle and then I used little potentiometers attached to pulleys actually with a little torsional spring so that it could measure the movement of the thumb in two different axes from left to right and up and down and the pulleys are sort of unwind to allow the thumb to move and this unwinding would be measured by the potentiometers and used to control the thumb so they say I'm not really happy with this design it definitely needs some reworking to be a little bit more robust but detecting motion in the film as acutely as that was really challenging especially because the thumbs probably the most important finger to allow you to to do anything really and finally there was the sensor to measure the flexion of the palm via the movement of the metacarpal bones inside the palm I would say this part is the biggest sort of working progress sensor because I haven't really figured out a way yet to measure it without really compromising the stability of the knuckle sensors further up so it just uses it's kind of like a lever that goes across the palm of the back of the palm and attaches at the knuckle so overall with this glove design I would say there's nothing wrong with the mechanisms in concept but in order to make it work repeatably and and reliably I would need to find more reliable potentiometer news throughout the design and probably not a service mount at all or if it is it needs a really clever way of keeping the wire completely held still better than my epoxy idea of keeping it perfectly still where the contacts are but then allowing it to be flexible past that point and probably needs better quality quality wire and better quality solder and probably thinner wire as well because there was so many wires on the thing that it did actually start to obstruct the motion a little bit the other thing as well was that all these tiny little things were really really noisy so the signal was super shaky which was really unfortunate on air too so I write a program to smooth it out I assume just because there's so many wires crossing over each other and maybe interfering with each other slightly or maybe it's just the quality of the soldered connection and whereas the the original design that had they used the big clunky screwin potentiometers didn't have any noise at all and that was really smooth in other to make the control of work I would need to sacrifice some mobility in exchange for just more robustness and now you might be thinking like why do you even need that many potentiometers you could probably approximate the movement of the whole finger with just one potentiometer for like how open or close the finger is like humans are actually that dexterous that it's easy to isolate each individual finger joint at a time and that is absolutely correct although ii kinda like the whole idea of this project was just design a bionic hand that had like the maximum dexterity that a human could ever have to be able to the idea was really just like if in the future we can control prosthetic hands really easily like we can somehow wired directly into the brain or whatever how near could you get the design of the Bionic hand to a real human hand and could you get to the point where you'd be able to play the piano and pin and you know do all these fine things with your hands so I do think that design in the control glove on this level is meant that if I wanted to make a simplified version that did just use one or two potentiometers for each finger and I'll be able to just design a super robust what a super effective and robust one without too much trouble because I've gone through the effort to try and really design this of a complicated one and I sort of have a really good idea of how we could do it I mean quite a few people have made control girls felt like the the in move Bionic hand and stuff like that a Muslim used bend sensors which it looks like it works fine but the cost of fortune so by using potentiometers with this lever arrangement I have I could build a simplified glove that only really has like one degree of freedom pair finger for like almost no money at all and I do still think this really complicated one could work it just needs just more robust potentiometers a better way to manage the cables and it really just needs just sort of further iterations of the design there's so many things that you can't really notice in SolidWorks until you print it out and sort of test it with something so complex and organic as a human hand there's no way it'll really simulate it accurately in SolidWorks or any CAD software I mean you can try and get a 3d model of a human hand and try and fit it to the model of the control glove but you can never really fully appreciate how it feels and how naturally moved or anything like that so it is something that you just have to build prototypes of and make iterations of unfortunately but yeah I do think that given a few Moret iteration it could work perfectly well then it goes straight on to this bit in unity so I'd never used unity before this and I was sort of reluctant to learn everything from the ground up because they thought I wouldn't be using it that long so I tried to just kind of like cheat my way through the curtain just copy and paste bits but ended up having to learn most of it anyway so realistically it would have been easier to just start the at the bottom and back my way up and understand everything rather than try and jump in at the deep end and end up having to learn all the basic stuff anywhere so I found this hand model online and rigged it up in blender to of all the sim ranges immersion as my control glove yeah so what's not shown in that University open evening demonstration is all the setup that eventually got me to this really smooth they're looking and they looked a little bit horrific at one point but gradually I saw it we call the the values to be able to get it to look nice for the longest time I couldn't figure out what was wrong with the model that it just didn't move right and it only occurred to me right near the end that the burns as you place them in you name in blender sorry when you're rigging it up they have a direction and they have a sort of X Y Zed axis it's a just been kind of placing them willy-nilly and not been thinking about their rotation so when I went to program the hand in unity to move through a certain rotation at each joint found that it was going all over the place and twisting and going off at funny angles and I couldn't figure out why but obviously it was quite simple she needed to rotate the bones in blender to be around the axes that I wanted them to rotate so stuff a really long time but I'm happy that I went through all this because it's like another skill that I've learned is I have some idea of how unity works now and I think the video of it working is pretty cool and it also gave me a really good starting point for mapping the data from the control glove onto the Bionic hand so I'm going to talk about the Bionic hand in another video so come back with a part two I hope you've enjoyed this a little roundup of how the project went and I'll see you in the next video

46 Comments

  1. who the fook is that guy stevens said:

    I haven't seen anyone comment this but that transition tho @0:01 damn that was nice

    June 28, 2019
    Reply
  2. Yhago Vicente Santiago said:

    Sensacional,

    June 28, 2019
    Reply
  3. Raghav Thakur said:

    Please

    June 28, 2019
    Reply
  4. Raghav Thakur said:

    Can you please send me the pcb design in your project to control multi servo motor

    June 28, 2019
    Reply
  5. Sohaib Shahid said:

    Its really good well constructed bionic hand.it will be great if it become open source.lots of people can get benifits from that..

    June 28, 2019
    Reply
  6. italo garcia said:

    It s a beautiful project

    June 28, 2019
    Reply
  7. Michael Ellsworth said:

    Have you tinkered with the thumb pulleys? I recommend trying springs instead of the wire, more resistance should be a more sensitive reading… (Hypothetically… I'm new at this though…) LoL

    June 28, 2019
    Reply
  8. Family Greene said:

    at ~17:40– the end.. is it just me or were you using a right hand glove to control a left hand model?

    June 28, 2019
    Reply
  9. Sparky 6t said:

    how did you get the right dimensions for the model to fit your hand ? I have problems modeling a bionic hand that has the correct proportional dimensions and curves that would fit

    June 28, 2019
    Reply
  10. Omar Essilfie-Quaye said:

    Have you considered using a Leap Motion?

    June 28, 2019
    Reply
  11. Mr. Thomas said:

    nicely done though a few things to keep in mind are:
    1. the more you ty down wires the less stress on the solder joints. I recommend tying down with tyraps with plenty of slack on the cable
    2. for the movement of the base of the thumb and pinky, why not use a linear variable resister as that is kind of think you want to know. how much did it move of the plane of the hand.
    3. you could lose the finger tip potentiomitor on all fingers except the thumb and move that at the next joint in the finger, as I don't think very many people can move the tip of the finger without the second joint moving as well.
    4. learn your 3D printer so you can design better with less parts and a higher part strength.

    June 28, 2019
    Reply
  12. Cody Hurley said:

    Please take a look at my robot hand on my chanel. Keep up the hard work.

    June 28, 2019
    Reply
  13. Vikas Thakur said:

    Can you please send me circuit diagram

    June 28, 2019
    Reply
  14. Vikas Thakur said:

    Can arm be wireless

    June 28, 2019
    Reply
  15. Vikas Thakur said:

    It be like iron man hulk buster which copy movement

    June 28, 2019
    Reply
  16. Muhammad Hazim said:

    can i know what little part you using at 4:48 (what name it is ? )

    June 28, 2019
    Reply
  17. Andrew A said:

    Not a criticism, but you kept saying you'd want to fix cable management, so here are two ideas if you were to do something similar again in the future:
    1. Instead of running power/gnd from the arduino to EVERY potentiometer you could have two big wires for power and ground going to the plastic ring around the wrist. Then from there you can split it off to each potentiometer.
    2. look into enamel coated wire (usually used in motors) at least for the Power and ground, if not all three connections. I'm not sure how noisy that wire would be so maybe you could still use some stranded wire for the analog pin

    Anyway it was really impressive to watch! I stumbled upon your videos and I'm glad I did. This is a really impressive project. Even just making the glove would've been a feat on its own! Kudos to you

    June 28, 2019
    Reply
  18. Dylan Mc villain said:

    Nice job bro keep up the good work I hope that you finish this and that all of those in need will get this

    June 28, 2019
    Reply
  19. Jurgen E said:

    mazing project

    June 28, 2019
    Reply
  20. Shakapam said:

    I really love what you made with this project ! Have you tried to use a flex sensor on the glove instead of the angle sensor ? It coud save a lot of place, but I'm not sure it can be accurate enought for this kind of needs.

    June 28, 2019
    Reply
  21. NerdyPi said:

    I feel like it'd have been easier to use multiple cameras and a black glove with white dots to track the position of each finger section and use that for calculating joint angles and the like.

    However, many props to the engineering job on control glove. It'd be super useful for VR games.

    June 28, 2019
    Reply
  22. dekutree64 said:

    Very impressive! I'm making a control glove as well, though much more simple (for biomimetic bird wings, rather than hands). Do you have any ideas for how to measure wrist twist? I'm currently using two BNO055 absolute orientation sensors, one on the forearm and one on the back of the hand, to measure wrist twist and bend (transform hand quaternion by inverse of forearm, then convert to euler angles). It works pretty well and is sleek and durable, but these sensors are a bit iffy in terms of maintaining calibration while moving around a lot. It's also not super accurate since the back of the hand tends to wrinkle up when the wrist is bent upward, which lifts the sensor away from the skin at random angles.

    June 28, 2019
    Reply
  23. Eddie Shannon said:

    I wish you had a walkthrough video making one with all the components😭 I want to make a (muscle controlled) bionic hand with 3D printed parts but I’m new to this kind of stuff

    June 28, 2019
    Reply
  24. Hesla Engineering said:

    FINALLY i can leave my piano at home and play AIR PIANO now. just connect the arduino with a midi device.

    June 28, 2019
    Reply
  25. Victor Galvez said:

    An idea for the thumb: maybe you could track it using hall sensors and neodymium magnets in each "plastic rib" in order to map the distance and relative angle with the next rib, and use the pots with the final thumb part. Also have another idea to implement haptic feed.

    June 28, 2019
    Reply
  26. lineuve santos said:

    Best work!

    June 28, 2019
    Reply
  27. Eric Roberts said:

    i would like to know what program did you use to design in 3D

    June 28, 2019
    Reply
  28. TheRocketEngineer said:

    Part of a super realistic huminoid Android sometime soon?

    June 28, 2019
    Reply
  29. MrJefferson105 said:

    What program did you use to do the renders ? I am fascinated by them.

    June 28, 2019
    Reply
  30. Diego Perez said:

    Love this, really impressive! Though I would've thought that with the rate of progress being made in computer vision, you might not even need a physical "glove" soon?

    June 28, 2019
    Reply
  31. sean huybrechts said:

    why not make your own pcb with just the atmega chip and only the required essentials and write it with registers? your program will be working way faster with less delay between actions

    June 28, 2019
    Reply
  32. LuigiTime9 said:

    I really like this project, could you make it open source so that we can build our own versions and maybe even make it better?

    June 28, 2019
    Reply
  33. Dylan's Edits said:

    this looks so fun to code that I'm gonna cry

    June 28, 2019
    Reply
  34. Viro Science said:

    i would use motion tracking software to controle the glove i have some experience with motion tracking

    June 28, 2019
    Reply
  35. Thom Pascoe said:

    Gained a sub man. I've got a 3d printing company in Australia keen to talk if you need anything. Hit me up

    June 28, 2019
    Reply
  36. KKK77 VVV77 said:

    WHOAaa dope

    June 28, 2019
    Reply
  37. Paul Newton said:

    Anything thoughts on using conductive fabric or an immersive substrate for building a model trained on resistance|capacitive inputs along with visuals of the hand to interpret different resistances as hand gestures.
    see https://www.nickarner.com/blog/2017/7/10/touch-and-water-as-an-interface

    June 28, 2019
    Reply
  38. Jason Stokes said:

    Wouldn't the remote interface be easier and more reliable with a camera and a dotted glove used in motion capture? Love the robotic hand!

    June 28, 2019
    Reply
  39. Joshoowah said:

    Dude, you need some relief points for those wires; a peg or something to wind a bit of wire around–or to just secure it in place, so you can have extra slack–before the solder points, assuming there's enough flexibility in the wires and enough space to do so.

    June 28, 2019
    Reply
  40. ken pinoy said:

    If only leading bionic hand producer would support you. Man you got the most brillant idea.

    June 28, 2019
    Reply
  41. KuraIthys said:

    By the way, have you considered displacing the potentiometers through the use of cables?
    There'd be a lot less difficulty wiring them if they didn't have to be so near to the joints, nor would they have to be quite so small, and the hand structure in general would be less bulky.

    I'm sure there's downsides to the concept, but it seems like it'd make the construction less complex, and mean the actual wiring doesn't need to flex so much, the pots can be larger, and so on.
    (it would mean most of the electronics being on the forearm) – obviously this is similar to how a hand is connected via tendons with the actual muscles being in the forearm, but it may also be viable for the sensors themselves…

    my first inclination for this would be to use high strength fishing line, but there may be better choices…

    June 28, 2019
    Reply
  42. KuraIthys said:

    If you're using an arduino but with so many inputs, why aren't you using an Mega?
    Would probably make life a lot less complicated, since it has 54 inputs instead of 14…
    You'd reduce the need for intermediate circuitry quite a lot.

    June 28, 2019
    Reply
  43. Aimless said:

    shit, Too late for me to be the first dude to print this kind of hand.

    June 28, 2019
    Reply
  44. mark menezes said:

    what was used to programme it

    June 28, 2019
    Reply
  45. 四葉草のオオカミ said:

    great work!! keep it up, bro.

    June 28, 2019
    Reply
  46. Peter .S. said:

    To improve you wiring perhaps look in to silicone wire. BNTechGo on Amazon makes some amazing quality 30 AWG silicone wire that's absurdly flexible. Make sure not to skimp on cable management too, zip tie points are key.

    Also, to reduce the total number of wires, only wire the wiper and one end of the potentiometer. You can add a series resistor on your controller board to complete the voltage divider. The extra resistors will be less intrusive than 50% more wires.

    June 28, 2019
    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *