Society of Robots - Robot Forum

Electronics => Electronics => Topic started by: Asellith on February 07, 2008, 01:36:04 PM

Title: Wheel Encoders and error rates
Post by: Asellith on February 07, 2008, 01:36:04 PM

There is a lot of error from slipping with a wheel encoder so my thought was to minimize the error what if you used a caster that was not driven by a motor and attached the encoders to that. A simple and easy solution would be to mount an old style mouse to the bottom of a robot and read the encoders from that. Would that eliminate some of the error in the readers or introduce more? My thoughts are that as the motor rotates the wheel if the wheel doesn't move but slips this intorduces error however if the wheel only moves when the robot moves it would have less error. There is still some error because the robot would move and the ball might not. But would it be less then the other way around?

A better way to do this would be 2 omni wheels at right angles with encoders attached.

Title: Re: Wheel Encoders and error rates
Post by: krich on February 07, 2008, 10:35:54 PM

You might look at an optical mouse for this.

 Agilent ADNS-2610 (http://cp.literature.agilent.com/litweb/pdf/5988-9774EN.pdf)

Title: Re: Wheel Encoders and error rates
Post by: Rebelgium on February 08, 2008, 09:26:45 AM

The downside, and the reason I decided not to use this method on my robot, is that you can only use this method on a perfectly even surface. And preferably no tiles ( the grooves between tiles make the mouse ball float, giving error. Also using an optical mouse can cause problems on shiny floors...

The best way imho to eliminate slipping errors when stopping and accelerating is to make your robot do these actions smoothly.
Use PWM to let your motors accelerate smoothly.

Title: Re: Wheel Encoders and error rates
Post by: gamefreak on February 12, 2008, 08:34:21 PM

There was a way to measure distance in thr Robot magazine, I beleive it was back EMF

It sent the PWM to the motor and then measured the amount of voltage coming back from the motor, Using this they could determine if the wheels were floating or pushing against something solid.

Title: Re: Wheel Encoders and error rates
Post by: JesseWelling on February 13, 2008, 01:44:07 AM

You could also use PID and tune for optimal slip vs. step response performance.

Since the Admin is a Mech-E he might know a little bit more about that though  ;)

Title: Re: Wheel Encoders and error rates
Post by: Admin on February 16, 2008, 06:56:53 PM

I think the encoder on the caster idea will be even worse.

Think what will happen if the robot slips in such a way as for it to rotate around the caster - the encoders say nothing happened yet the robot still moves!

Of course a 3rd encoder could fix that, but thats just overcomplicating things . . .