|
Post by Bones on Oct 15, 2007 23:05:21 GMT 10
Hi Everyone, I have an idea for everyone to have a think about. To see if it will work or not.
Now walking robots have a lot of problems falling over, even the top notch ones like asimo can fall over. If they put one foot out of place, even slightly then the next couple of steps compound and then Splat.
I am thinking of fitting a gyroscope into the robot. Like the old Bicycle gyroscope you learnt in school, but smaller. Maybe in the chest somewhere.
Now I am thinking that the gyroscopic properties would slow the tipping/tilting of the robot, but will still allow the forward and sideways plane movement. As it takes the foot off the ground to place it forward then the Gyroscope will try and keep the body upright on the back foot.
Once the gyroscope has spun up then it would only need a small amount of power to keep it at that speed I'm thinking. The only problem I see is that the torso of the robot would always try and stay in that upright position. This would make it look like it has a pole stuck up it's bum. lol.
I am thinking this idea may help lessen the constant balancing act code that is needed to to take one step after the other without falling over.
So what does everyone else think?
Rod
|
|
|
Post by Sandgroper on Oct 21, 2007 1:53:06 GMT 10
Hi Bones,
Sound like it would work.
I remember reading somewhere about ring-laser"gyros". They were set up so that any change in direction caused a detectable doppler-type interference pattern in opposing laser streams. No moving parts.
The robot doesn't have to look like it's got a pole up its.... It only has to keep its centre of gravity over its feet.
By the way, have you seen the donkey robot on YouTube? Nice piece of balance work there.
|
|
|
Post by Robotman on Oct 22, 2007 10:56:30 GMT 10
I would be surprised if the idea hasn't already been tried many times in many labs in many pasts.
Probably exactly because Asimo doesn't have a gyro in it would lead me to suspect that such a passive system won't work - the effort required to unbalance the robot for taking a step would probably "topple the gyro", and that's plan A scrapped.
I like Sandgroper's idea about using a "ring-laser gyro". I just see one slight problem: it is **expensive**. Really **expensive**. Probably so **expensive** I would need to re-mortgage my house. And that's also probably why Asimo doesn't use one of those either.
My thought is that a walking machine needs active control over its muscles, with the gryo being replaced by a 3-axis device - similar to what's in our ears - for balance control. (I have seen a basic capacitive sensor design that is a fluid in a drum with hemi-circle metals plates on the outside. The fluid moving around on the inside changes the capacitance between the two metal plates. Put three of these together to make "vestibular tubes".)
I don't think the maths needs to be "heavy" either. Surely someone prepared to build the hardware could knock up a software neural net (or equivalent) that is capable of learning how to walk using the hardware it was "born" with. The goals would be: 1) learning to just stay standing still, 2) learning to return to standing after being mechanically pushed off balance, 3) learning to return to standing after being pushed off balance *by software*. More tricks like learning to pick itself up off the floor and return to standing could be implemented later.
Each robot would the same hardware, but with slightly different tolerances, so it would have to learn its "body" like a baby. I've been saying this for awhile now (and not just to myself either) - just like recently some "guru" of robotics research has come out and said: "Duh, we ain't getting nowhere wid all dis hard-coding stuff, so let's make it learn like a baby does. We is too clever, huh?"
***k me.
|
|
|
Post by SDFDS on Aug 14, 2008 18:25:10 GMT 10
|
|
|
Post by SDFDSF on Aug 14, 2008 18:27:19 GMT 10
|
|
|
Post by SDFDS on Aug 14, 2008 18:28:58 GMT 10
|
|
|
Post by FDSFDS on Aug 14, 2008 18:30:49 GMT 10
I also know that when that decision is made you want to be on the road to a training program as soon as possible. free online games.Just remember that choosing a Training program for yourself and the dog is as important as choosing. play war games.a college. free online war games.You will need to know a lot about. online games.the program and what you really need from a specific Guide Dog school.
|
|