Sensors for human exoskeleton engineering project

[awally88]

Hey everyone,

For my final year engineering project my group has been looking at making a human exoskeleton, for all those who have seen the movie 'Iron man' we are basically making that suit but worse as its real life . Anyway I'm working on the electronics/control system for it and I've been researching sensors at the moment, we will need to be able to model all the movements of the human body, its going to be quite a large project for the team but we're starting early and we have over a year and a half to get this done.

I was wondering what other ideas people had for the sensors to map the human body. Initially we are just going to do a prototype of a single arm and I thought that I would just use a fabric bend sensor in the joint of the arm and then use accelerometers/gyro to correct the data however since I haven't used them before I'm not sure exactly how everything is going to work.

Any ideas you have or links would be awesome

Awally

[stan]

http://www.robotshop.us/force-sensors.html

I would use flexible bend sensor ...see link above..

[awally88]

If that sensor is what I think it is it only senses force in the circular section. Its really just a pressure sensor and can only be used to sense a force on the circular end section, it doesn't measure the flex of the sensor.
However I could be wrong about this.

[aludra_55]

check out Sparkfun for their sensors as well... 

from what it sounds, you want a sensor to "sense" what the body is trying to do
and then amplify the strength of that body part by using servos, motors, air muscles,
etc to increase the output force...   

if you're taking an arm, then I recommend accelerometer to measure the angle of the
arm, gyro to measure the rate of change... you could also use gyro to make your
accelerometer more precise because accelerometer gets noisy reading if the device is
in motion...   you could use force sensor inside the armor suit to measure the amount of
force the body is pressing against the wall of your suit...   

for motors you could use servos...  they have high torque digital servos... you could also
use gearmotors for some more torque...  you could try using cylinder pumps or air muscle
composed of hoselike tubing that is pumped with compressed air, etc...   are u gonna
make a mechwarrior?..  =)

[awally88]

That exactly what we are doing aludra. We are making something similar to this

Sarcos

This is a multi million dollar suit though so ours is probably not going to be this extreme.

We will be using hydraulics for motion, we did a study into it and it seems to be the most efficient way of getting the amount of force we need. Servos and other motors that give the force we need are too large to be as portable as we would like. Air muscles and other pneumatics looked interesting but can be hard to control I believe. I'm currently thinking that we will use the accel/gyro to do this, maybe also using either a fabric bend sensor or something like http://www.robotshop.us/images-scientific-bi-directional-flexible-bend-sensor.html or http://www.robotshop.us/images-scientific-4inch-stretch-sensor.html. I'm going to talk to some lecturers at uni and see what they have sitting around that I can play with before I spend thousands of dollars!

[dunk]

i'm not convinced accelerometers and gyros are the best solution.
if these sensors are attached to the exoskeleton the sensors would only tell you what direction the exoskeleton is moving in, not what direction you want it to move in.
you would have to start the exoskeleton moving before the motors kicked in.

you could leave a lot of space inside the exoskeleton and attach the accelerometers and gyros to your body (instead of attaching them to the exoskeleton) but that seems like an ugly solution.

i would use pressure sensors inside the exoskeleton so it can tell when your body presses against it and matches movement in that direction.


dunk.

[awally88]

I was thinking of mounting the sensors to the body, having a wetsuit or something similar that would fit snugly  on the body, however it seemed a bit of a hack to me, that's really why I was asking. 

Pressure sensors might be a good idea, I was wondering about how they would work when Stan said it before but I had a flash of understanding when you said it dunk (Sorry Stan!)

The only thing about using pressure sensors is that they wont be able to sense rotation, for the prototype the design is going be simplified not to need rotation however I believe that we may need to consider it at a later date. Did you have any ideas about how to deal with rotation?

[SmAsH]

do you mean rotation of the arm joints or the whole are rotating? like just from the elbow downward? you could have a pot on the elbow joint sensing when the arm is rotated... but i don't know if an external load would twist the arm...

[aludra_55]

use electrodes to put on the skin on top of the muscles like the tricep, bicep,
forearms, etc...   the electrodes would then measure the tension in particular
muscles and you could use the signal to determine what the person was
trying to do and the exoskeleton would drive the specific hydraulic cylinder
to perform what the person wants to do...

[SmAsH]

but then that would take a while to put the suit on wouldn't it? i guess you could try having a bar at the end to hold that is connected to a circular plate that rotates on a pot head (not that type)...
and electrodes are kind of hard to use aren't they?