-
Hack Chat Transcript, Part 2
05/20/2020 at 20:04 • 0 commentswould the object being touched need to be conductive though?
Yeah, and it'd get pretty grubby pretty quickly
These $10 Sensor-Packed Gloves Could Give Robots a Sense of Touch
Machines are mastering vision and language, but one sense they're lagging behind on is touch. Now researchers have created a sensor-laden glove for just $10 and recorded the most comprehensive tactile dataset to date, which can be used to train machine learning algorithms to feel the world around them.
Touch is great, but having the ability to feel as if you have grabbed a hard cube by maybe restraining the give movement is SciFi I guess
You could just slide the glove right over the bionic hand !
That's what I was thinking XD Disney did published some recent research about accurate sensing in soft systems (at least for positioning. It can be extrapolated to pressure).
I remember seeing a design once where there were some servos on the wrist with cables attached to the fingers, and they locked up to restrict the motion of the fingers
Like batman Cape kind of thing
A mechanism that restricted finger motion based on what's in the bionic hand's grip would seem akin to trying to handle objects when your hands are "asleep". No direct contact sensing, just position limited.
So we've been talking about the hand so much, I don't want to forget about your other great stuff. How's the terrifying mouth going?
https://hackaday.com/2020/04/30/this-animatronic-mouth-mimics-speech-with-servos/
This Animatronic Mouth Mimics Speech With Servos
Of the 43 muscles that comprise the human face, only a few are actually important to speaking. And yet replicating the movements of the mouth by mechanical means always seems to end up only partly convincing. Servos and linkages can only approximate the complex motions the lips, cheeks, jaw, and tongue are capable of.
Nathan thats interesting, I wonder if that technology could be used in a silicone skin?
Yeah it could. That's kinda the point. You can technically extrapolate their optimization pipeline for other sensor setups.
Thanks the mouth is going well, currently it takes a typed sentence and converts it into phonemes using a python script. I'd love to try using it with motion tracking too sometime, I've been looking into open source mouth motion tracking
the mouth is great! i finished building it last week! my goal is to hook it up to a raspberry pi, running alexa. and have a sort of animatronic alexa
what do you do with those old projects? Is it just sitting on a shelf, ready to talk at people?
Nathan Yeah that's interesting because it could potentially also be used to find the joint angles in the fingers, which would eliminate the need for a rotary sensor in all the fingers
Rob that sounds great I'd love to see it working with an alexa
Yeah at the minute my mouth is in its rabbit sat on my desk looking at me :) Often I break them down and recycle the parts though
When I was researching that article, I was amazed to see how few muscles the human face actually has. You'd figure there's be more given the number of degrees of freedom a face has, but you might actually have more servos than muscles there.
Yeah I know what you mean but the muscles can move in really interesting ways, like the lips can protrude forwards but also retract back to expose the teeth, so I'd need more servos than muscles to get the same degrees of freedom
Thank you Nathan, now I got a good project to work on I mean rotary sensors aren't that hard to integrate. And you don't necessary need them in the fingers. The IRIM lab gets pretty good performance without them:
pneumatic muscles is definitely something I'd like to look into though, the Shadow Hand uses them too.
not anymore they dont
that crap was hell to work with
How? leaky?
oh they don't sell a pneumatic version anymore?
I guess there's also the trouble of having to lug an air compressor around with you haha
Leaky, inaccurate, a paint to maintain, etc.
So, the heart - was that for a specific reason? Like a client request? Or were you just seeing what you can do?
https://hackaday.com/2020/01/19/be-still-my-animatronic-heart/
Be Still, My Animatronic Heart
Fair warning for the squeamish: some versions of [Will Cogley]'s animatronic heart are realistic enough that you might not want to watch the video below. That'd be a shame though, because he really put a lot of effort into the build, and the results have a lot to teach about mimicking the movements of living things.
Yeah I'd always wanted to do an animatronic heart, but an artist in Romania bought one from me after I'd finished, I'm looking forward to see what he makes with it!
Do you plan on making an animatronic leg?
What got you interested in animatronics?
I actually worked on a prosthetic knee as a prototype but I don't think I'll take it any further, I feel like legs don't have the same cool factor as arms :)
Not sure really Matt, I've always been into robots but also biology and art - animatronics is like the perfect compromise of all three!
Not really, giving the boot to someone isn't as fun as giving them the finger
But also I've been super inspired by Karl Gallivan, Gustav Hoegan and others
Got to run, thank you Will, great links to follow up on. Keep up the great work!
Yea, it's a great way to combine all 3. The heart is really fantastic in a super creepy way
What was your first attempt at making an animatronic?
Well I've had the hand project in my head since I was about 17. The first "prototype" was made out of scrap and wittled wood with bike brake cables, It didn't work out so well haha
Well, we're at the top of the hour, so we'll have to let Will get back to work. I want to thanks him for stopping by and talking with us today - I really admire the build quality you manage, and your information sharing has been great too. And thanks to everyone else for turning out with such great questions and comments.
FYI, I'll post a transcript on the Chat in a few minutes, in case anyone missed anything.
@Will Cogley Great work!
Thank youThanks for having me guys, and thanks for the chat! I've got some very interesting links to look through !
Thanks for the insight into your projects. Thanks
@Dan Maloney for organizing the chat!
And thank youAnd don't forget - it's Hackaday Prize time again! We'll be talking about that next week:
https://hackaday.io/event/170038-2020-hackaday-prize-hack-chat
2020 Hackaday Prize Hack Chat
Majenta Strongheart will host the Hack Chat on Wednesday, May 27, 2020 at noon Pacific Time. Time zones got you down? Here's a handy time converter! It hardly seems possible, but the Hackaday Prize, the world's greatest hardware design contest, is once more at hand.
Thanks Will, veryninformative
-
Hack Chat Transcript, Part 1
05/20/2020 at 20:03 • 0 commentsOK, let's get started. I'm Dan Maloney, I'll be the moderator today. I want to welcome everyone to the Hack Chat today, especially our host, @Will Cogley, who does some really cool stuff with animatronics.
Will, can you tell us a little about your background and how you got into this?
Sure! I’m Will, and I make 3D printable animatronics designs and publish them open-source. I’ve had a lifelong fascination with robots, but also the human body and biology. My main long-running project is a biomimetic hand that I hope will have the same movement and functionality as a real human hand. I’d love to answer any questions!
Here's the hand Will is referring to:
https://hackaday.com/2018/10/18/mechatronic-hand-mimics-human-anatomy-to-achieve-dexterity/
Mechatronic Hand Mimics Human Anatomy To Achieve Dexterity
Behold the wondrous complexity of the human hand. Twenty-seven bones working in concert with muscles, tendons, and ligaments extending up the forearm to produce a range of motions that gave us everything from stone tools to symphonies.
When do you decide between stepper driven and servo-driven solutions?
When do you think you will work on the hand again?
wow, very impressive re. the hand!
Most of my projects are done on a tight budget so I almost always use hobby servos. With my bionic hand I'm currently starting from scratch and trying to optimise each element, stepper motors are definitely something I'm considering atm!
Nice
I'm actually working on the hand right now (relatively speaking), I'm going to release a video about it hopefully next week
What is the biggest design challenge you face atm?
It's been a while since I watched the hand videos - remind me what technique was used to make the plastic pieces? They don't look 3D-printed.
I'm trying to make my projects as accessible as possible so I'm using an FDM printer to make all the mechanisms and parts. Most of the parts in my new bionic hand design are so small they're kind of pushing the limits of what I can print reasonably so I'd say thats the biggest challenge I'm having right now.
Can you see the hand being completed and if so what would be the next step after you have a fully functioning hand.
...and your preferred filament type? PLA? PETG? ABS?
Yeah on that hand I used acetone-smoothing to get the ABS looking nice and shiny, it was actually kind of a nightmare since all the parts started to sag because of their thinness so I had to switch to hand polishing so they didn't totally collapse!
Wow, they came out great! Almost look injection molded. Gotta say, the hand looks bad-ass!
or rather bad-palm
Ba-dum tiss
Yeah I think I'm going to have a good working prototype relatively soon, there will always be room for improvement I imagine, but I'd really like to work on some different control methods next. Originally I used a control glove, and now I'm considering motion detection. The main goal would be mind-control though haha!
Gotta hand it to him. they look great
I'm sure other body parts will follow ;-)
only if you walk backwards /ducks
Have you ever looked at the vr hand thingy to controll it? I meant more like after a goo hand prototype would you consider mocing upwards and maybe develop an arm?
I've used mostly PLA up until now, mostly because of my printer's limitations, but I've recently been using ABS and its proving to be much better for sliding surfaces
Well, I mean that brings up a good point. Doesn't some of what animatronics attempts to do bleed over into prosthetic design? Like maybe drive it in terms of function and style?
Btw I really like you open sourcing stuff on your website. I can imagine if your hand prototype would be out many developers would start working on a vr styled controller.
I've seen some work with card-sized controllers with enough horsepower for image rec for things like rock-paper-scissors. Would be great to have a hand pose recognizer that drove the bionic hand.
Yeah in the original design the forearm at least was a convinient way to house the actuators - in the new design I'd like it to be more functional and aesthetic
What are your favorite "default" servos?
Is there a plan if this works well to create a kickstarter campaign to make injection molded kits?
Yeah there's definitely a bleed over between prosthetics and animatronics. I try to borrow from both worlds :)
I bought a pack of 100 MG90s servos from aliexpress about 5 years ago and I haven't run out yet, I use them for just about all of my prototypes haha
Could see an app for use as tele-ASL replicating a visual signer's poses for a blind/deaf person to read by feeling the bionic fingers move
I like the hand drawn sketches in the video. I usally also start with those sketches before starting 3-D drawing it
Thats an awesome idea, thats kind of why I like the idea of open sourcing it because there's so many different applications
MG90s! I got a bunch too. Some of them are pretty crunchy though and over time the pots can wear out.
Yeah they're not ideal, and almost definitely knock-offs.
How do you drive them? PCA9685?
Thanks Bindhammer, yeah I put all my best sketches in the video but most of them are far rougher than those haha
Yeah I use a PCA9685 most of the time
What controllers you use to drive them?
Arduino/RPi ?
@Will Cogley, you'd probably like Eric Strebel's design videos. I know he'd love the look of your stuff: https://www.youtube.com/channel/UCFzr6ae9Uv-a_tCmHmkw49g
@Boian Mitov He uses an arduino
Yeah I use an arduino, but some projects I've used a python console on my PC feeding into arduino
I see CircuitPython in your future!
Thanks Dan I was not aware of Eric Strebel
How do you manage power? I've seen servos cause low voltage and mess with micro controllers.
Hmm I'm not familiar with CircuitPython, is it like a tiny raspberry pi?
Well with the PCA 9685 driver you can power the servos separately to the arduino board
it's just a stripped-down python that runs on microcontrollers - still like programming an arduino, but python insteap of C++
Python env for multiple hardware platforms
Interesting I'll definitely look into that
If you don't want to spend too much time on the software development, you can alternatively try Visuino ;-)
AdaFruit features it in a lot of their box projects
To answer John's question, I usually budget 1A for each standard size analog servo. Digital servos can easily hit several amps.
@Will Cogley it's like you program Arduino, but with Python
I've only recently started getting into python so that would probably be a really useful next step actually
https://learn.adafruit.com/welcome-to-circuitpython/what-is-circuitpython
Welcome to CircuitPython!
You may have seen CircuitPython mentioned on the Adafruit site or on Discord. What is it? Why would you use it? Once you have a CircuitPython board, what do you do next? This guide has you covered from the moment you plug in your new board. Check it out!
Read this on Adafruit Learning System
The hand mimic is great, how about some sensory feedback. Do you have something in mind. Like tactile feedback
Do you do any wheeled robotics? I am looking for a good skid-steer routine for the Arduino.
https://www.brookshiresoftware.com/vsa_overview.htm
Personally, I use VSA for my storytelling robots:When I use servos they always have jittering , random small motions on powerup, is that normal? I tried many ways to avoid this, but with no help...
Yes sensory feedback is definitely something I'd be interested in. I've seen a glove which used servos to do it but very clunkily. I had thought maybe tiny solenoids ?
You said you were looking at motion detection as the next input control setup. Were you looking at something like a leap motion controller or something else?
Some quick&dirty systems just use haptics from phone buzzers to stim the fingers
I was trying to figure out a way to mimic pressure sensing, something like those pressure mat things
Yeah leap motion is exactly what I was looking at, it looks like it has all the degees of freedom I would need
Oooh for feedback, you could totally use an ultrasonic array.
the company leapmotion merged with manufactures those
https://ieeexplore.ieee.org/abstract/document/1041717
Maybe build your own whisker sensor?Use a strain gauge on the bionic hand and map it to stronger vibration in finger vibrations?
I wonder if little piezo transducers in the fingertips of the gloves could be driven to provide some haptic feedback. Your fingertips are pretty sensitive, after all.
Leap motion is fine but hand should always be parallel to it, or it lose tracking for some seconds, I had used it for a while
When I make very active characters, I tend to strip a lot of gears. I started using servo savers to limit torque and that seems to help. Any thoughts on these?
yes piezo might be a good idea. How would an ultrasonic array work?
I had a theory for a long time that each finger would have an oil-filled bladder to measure the pressure/force put on the finger.
@Louis I have a bunch of different wheeled robots, and maybe able to recommend you suitable one. Will send you private message...
But the response of oil flow might be slow for real time
Interference patterns of ultrasonic waves create pressure that you can sense on your hand
Synthetic 'Skin' Is Bringing a Sense of Touch to Virtual Reality
Virtual reality's ability to create artificial worlds has come a long way in recent years. Now new technology could make th ose worlds even more realistic by simulating one of the most crucial ways in which we experience things: touch.
Can the ultrasonic thing be scaled to a glove
https://www.ultraleap.com/haptics/
Haptics | Ultraleap
Using our patented algorithms, the ultrasound speakers are triggered with very specific time differences. These time differences mean the ultrasound waves arrive at the same point in space, at the same time.
that's really cool but not very portable I'd imagine
@yawar.saeed it's more of a no contact approach
For sensing, I'm working on shape sensors that might be ideal for controlling animatronics:StackPath
yea, probably not very portable. What are the use cases you're looking at?
Paul they look great, I had planned on using bend sensors the first time around but they didn't fit within the university's budget unfortunately haha
We will be releasing dev versions later this year...
I was trying an approach of defining zones in the hand and fingers because brain can autocorrect the rest
Not entirely sure Matt, I just like the idea of things being wearable haha
Will you add skin or leave it exposed?
Nah I like the robotic aesthetic
but actually thinking about it a silicone skin could really help with grip
Will be interesting if the surface of the hand can be covered with a touch/pressure sensitive layer
Enough realism with the heart huh? :-p
Have you tried grabbing an egg? It something soft to optimise handling
print with nijaflex instead of abs
Yeah my first thought was just to use flexible filament for the outer shell parts, probably wouldn't be as nice and shiny though
what about a reverse capacitive touch? have the same mesh on the hand instead of the thing being touched?
About Us Contact Hackaday.io Give Feedback Terms of Use Privacy Policy Hackaday API