Artificial muscle control

I had this idea a while ago, and while I think it would be cool, it's probably not very practical.

Muscle stimulation using external electrodes has been around for almost as long as we've had electricity. Apart from being a cool experiment, it has some uses in rehabilitation, but I won't go there since it leads into the pseudoscience of stimulation weight loss etc.

What I'm interested in fine-grained muscle control. A few years ago some academics wired up a guy's arm to allow a robot to control it. read about it /watch the video here.

What I have in mind is something like this, but more useful. e.g. imagine having a camera on your shoulder, and electrodes on your arm, you could program the system to watch for incoming objects e.g. a ball and move your arm to catch them. or a system to make your legs walk and avoid obstacles so you could read a book while doing walking in the park. etc.

Aside from this system being somewhat cumbersome, the main issues I foresee are:
not enough precision to be useful - i.e. can only give general commands to large areas, so you fall down when trying to walk.
muscle exhaustion - because stimulation is rather imprecise, the muscle uses more energy than normal

I don't think this will ever be useful in the real world, but I think it could be a fun/interesting experiment

(Ideal daydreams: maybe with arrays of implanted electrodes, control could be precise enough to make it useful)


  • edited October 2014
    I've done some experiments with this using electrodes above the skin surface. There are definitely methods of creating "fine" muscle control, even without implants, but they're kinda nit picky, and may or may not disappear when you move. And you'd be covered in wires..

    On another note, I wonder whether the technology they're applying in cars that drive themselves could be of use.
  • edited October 2014
    I believe their would be an issue with attempting to walk vs the stimulation. I have had electronic muscle stimulation as rehabilitation from an injury and when their is a significant amount of current it is very very difficult and painful to move because your muscles seem to "lock up". (imagine trying to move your hand while being shocked by an electrical current)

    I hypothesis that instead of moving out of the way, one would simply fall over due to the "lock," pain and surprise.

    However I can see similar technology be used as a "warning signal" (like sensors on the bumpers of newer model cars) and it could interface with an implanted magnet. (similar to the bottlenose project) I had an idea of setting up sensors in the blindspots of my car so whenever something was within range it would power a coil which would interact with my magnet. I would be able to feel if another vehicle approached me/warn me if I was backing up to close to an object. Unfortunately, I know nothing about creating software, electrical engineering or any other skill required to perform such project myself.
  • One thing to consider is how the muscle is activated normally vs via external electrical stimulation. When you contract your bicep for instance, the impulse starts at one end and moves through the muscle. This causes the smooth movement that we associate with curling and arm. When you have two or more electrodes on a muscle and you are using them to activate it, the impulse is simulataneous and from both ends. This is what gives you that weird spazzy "lock" described from EMS.

    If you are looking to actually drive an arm or leg in any sort of functional fashion, I believe that this will be the (initial) issue that you need to overcome.
  • May have a strip of electrodes running down the arm/muscle. Fire them in pairs, and start with the Electrodes 1 and 2, then 2 and 3, etc. That way, the pulse is simulated... Sort of.
  • long story short: precisely stimulating muscles is very easy. knowing how much you want to stimulate them is very hard.

    short story long: something like catching an object takes a lot more than just a camera and some wires. you can somewhat control the force a muscle is providing. but in order to precisely position your arm you need to have to control the force of many muscles. also the force you need depends on the current position, velocity , the desired speed you want to move the arm with, the arm's mass etc. then you need to build control loops which actually control the muscles force. since your arm does not come with plug-and-play sensors you either have to measure all those by external sensors (or you tap into the nerves providing that data). both is difficult. the actual control loops and kinematic model you need to calculate the forces is another task to solve. if you are up for the task, i recommend to start with a robotic arm where all required data is easier to access.
Sign In or Register to comment.