Robotic exoskeleton

.:John:.

In the future, I am planning on creating an Exoskeleton-like suit (no not Iron Man) to help me work with heavy materials.

Plan so far:
Linear Actuators from http://servocity.com/html/12v_linear_actuators.html
Aluminum frame
12v SLA battery
3 actuators per arm
2 actuators per leg

A few questions:

1. What would be better to measure the angle my arm is at, a potentiometer or a bend sensor?

2. Could a linear actuator be controlled by the Motor Mind B?

3. How do you program the MMB?

Thanks!

Post Edited (curious_roboticist) : 5/6/2008 12:55:49 AM GMT

pharseid

· I've been fascinated with the thought of doing this since reading about the Hardiman when I was a kid (it was a military project to produce a strength enhancing exoskeleton). In Circuit Cellar a couple years ago there was a project that used a light exoskeleton to measure joint positions for a virtual reality project and they used potentiometers. But do you really want to measure joint position for this? The standard method in exoskeletons is to measure the force you exert on the exoskeleton and use that to determine the inputs to actuators. More advanced is to actually measure the nerve activity, but I've read that even in those cases they use pressure measurement as a check. Also, your arm has 5 degrees of freedom, plus those of your wrist and hand, you may find that only allowing 3 degrees of freedom is overly constraining. One of the reasons given for the failure of the Hardiman project was that the skeleton didn't have enough articulations. I would recommend building an unpowered exoskeleton that limits your movements to those you intend to allow in the powered version and see if those are enough to allow you to do your intended tasks. Also be aware that the exoskeleton should be somewhat loosely connected to your body, because if the load is tranfered to your joints you can be injured (say you're trying to pick up 1000 pounds and power to the actuators lags a bit.)

· Finally, there have been a couple of contests for exoskeleton weightlifting. You might look at these for inspiration.

-phar

Franklin

Take a look...http://youtube.com/watch?v=Qr-prmlYcTM&feature=related

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Stephen

Whit

In Popular Science (this month) there is a big article on the military versions of these - pretty amazing stuff.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Whit+


"We keep moving forward, opening new doors, and doing new things, because we're curious and curiosity keeps leading us down new paths." - Walt Disney

FireHopper

bend sensors are fairly fragile though..
theres a few different ideas.. a standard pot, a linear pot. and even a bit of fibre optics used to sense bends..

take a small bit of fiber, a led at one end, sensor at the other..

at the middle of the fiber, sand the fiber it self some..
then take the led, give it some juice, not a lot, then go and check the output of the sensor.. when you bend the fiber, it should reduce the output some.. not sure by how much. depends on how much light is lost to the abrasion of the fiber at the bend..

Gadgetman

2 actuators on the legs is not going to get you anywhere at all.
This stuff will be heavy, and that means you MUST have good support(read: strong actuators) in the ankle joints.
They will need to be able to tip the blade of the foot back and forth, side to side, and to swivel it somewhat.

Unless you leave a lot of slack inside the skeleton so that your limbs can move, positional sensors won't be much use. You will need pressure sensors.
you'll also need 'outside' pressure sensors.

Low pressure inside = slow movement.
High pressure on inside + no pressure outside = fast movement
High pressure inside + high pressure outside = Slow movement, high torque(lifting/pushing)
No pressure inside + High outside = Keep it powered to lock the joints. (Holding)

You may also want pressure sensors in the foot blades to adjust the ankle joints, possibly a gyro to help keep your balance. (This kind of stuff is heavy, after all, and falling flat on your back isn't the way to impress your buddies

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Don't visit my new website...

.:John:.

I was was thinking of a slightly different Exoskeleton than all of you described. However, I think I will attempt yours, as there is less chance for injury if the suit fails.

-I was not intending to augment the wrist. However, that may come later.

-These are the actuators I was planning on using. However, they only move at 0.5 inches per second at full load(115 lbs.) If someone can find a more powerful and/or faster actuator for a similar price, please let me know.
servocity.com/html/12v_linear_actuators.html

-How long would the signals take to be processed and sent to the actuators? Even with a small amount of lag, the suit could feel like moving through water.

-robotshop.ca sells actuators that lift 400 pounds! However, they only move at 0.5 inches per second with no load. Could I hack them safely to make them turn faster?

-What kind of pressure sensors do you suggest, and where would be the best place to put them?

Post Edited (curious_roboticist) : 5/7/2008 12:47:57 AM GMT

pharseid

· You really need to think about the point of attachment and what sort of speed versus force you want. For instance, to bend the arm at the elbow, you would anchor the actuator to the upper arm and then attach at some point, say 1.5 inches from the elbow. Then at 15 inches from the elbow (where I'm guessing the hand is), you're moving at 5 inches a second, but the force is only 10 pounds. You'll come to notice that the power/weight ratio of electromechanical actuators is lower than biological muscle.

-phar

Gadgetman

Have you considered other types of motors/actuators?

Geared motors, steppers, hydraulics...



▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Don't visit my new website...

.:John:.

I actually think I will start with something like the Robogames video posted by Franklin. One actuator per leg and one per arm. Would that be feasible?

pharseid

·· Are you thinking of the pneumatic actuator guy? That was pretty ingenious, but what could you do with that setup? He could only pick things straight up, it you wanted to actually move material it wouldn't be very helpful. (Also I can lift as much without an exoskeleton as he was lifting in the videos.) If you just want to move heavy material you should get a block and tackle or a fork lift, if you're interested in the whole exoskeleton thing, you might start by isolating a particular body part to augment, like a power arm. Or just build a force suit that measures the forces you're exerting and add augmentation later. In any case, this is quite a difficult project.

-phar

skylight

To measure the bend of your arm wouldnt an encoder be better

DiablodeMorte

You might want to consider Air Muscles(http://www.shadowrobot.com/airmuscles/overview.shtml) I've never used them but I hope to this summer. They seem fairly easy to build if you are going lone DIY or pretty cheap if you want to just buy. They are supposedly very similar to biological muscles in their power/extension curve and in other aspects... I'd love to see someone start a Air Muscle exoskeleton: I'd even help if I could, I've always wanted to build an exoskeleton.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Current Projects:
Robot Control Via Skype API - Dev Stage(50% Complete) - Total(25%)
Robot Localization Via Xbee's - Research Stage
IR Tracking with Propeller - Research Stage

Gadgetman

skylight said...
To measure the bend of your arm wouldnt an encoder be better

Yes, but WHY would you want to measure the bend at all?

On a robot that might be useful, particularly on startup, but on an exo-skeleton, the computer doesn't need to know the position of most limbs, even at startup.
It's only there to assist the wearer by applying extra force when he moves his limbs.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Don't visit my new website...