Low Res Optical Encoders for Speed Control

[KrazyCarl92]

By the way, we were able to get the robot shooting about 50% last night without speed control. This was using the camera feed to the driver station for manual aiming and we had driver voltage control on the shooter motors. We may work toward integrating a simple non-PID speed control algorithm and vision processing from there. The vision processing part was complete during the build season, we just need speed control that works in PWM or to get our CAN working.

[Ether]

Originally Posted by KrazyCarl92

I could post pictures if anyone is interested.

If it's not too much trouble, please do.

[KrazyCarl92]

Originally Posted by Ether

If it's not too much trouble, please do.

posted here, there are 3 pictures:
http://www.chiefdelphi.com/media/photos/38088?

[Ether]

Originally Posted by KrazyCarl92

We may work toward integrating a simple non-PID speed control algorithm

Were you thinking of trying a bang-bang controller, or perhaps a take-back-half?

[Ether]

Originally Posted by KrazyCarl92

posted here, there are 3 pictures:
http://www.chiefdelphi.com/media/photos/38088?

Thanks. You got a lot of hits. I guess there was some interest :-)

[KrazyCarl92]

Originally Posted by Ether

Were you thinking of trying a bang-bang controller, or perhaps a take-back-half?

We haven't decided exactly yet. Haven't hooked the encoder backup since we stopped using it in competition, so that's the first step along with verifying that it works.

I had thought of a bang-bang approach where we base the original voltage on motor curves, then if the speed drops X% below the target speed, raise the voltage by Y%, or if the speed exceeds the target by X%, lower the voltage by Y%. X would be the max allowable deviation from the target speed and Y would be an experimentally determined voltage correction. If its between 100-X % and 100+X %, we would apply no voltage correction.

I'd also like to see what the students come up with. After reading about take back half control here:
http://www.edn.com/design/analog/432...ence-algorithm
that seems like a good option. It looks like a good opportunity to explore different speed control options so we may try multiple approaches to see what gives us the best results (at least that's what I would like to see).

[Ether]

Originally Posted by KrazyCarl92

I had thought of a bang-bang approach where we base the original voltage on motor curves, then if the speed drops X% below the target speed, raise the voltage by Y%, or if the speed exceeds the target by X%, lower the voltage by Y%. X would be the max allowable deviation from the target speed and Y would be an experimentally determined voltage correction. If its between 100-X % and 100+X %, we would apply no voltage correction.

There's a lengthy discussion (and paper) of bang-bang here, if you haven't seen it yet:

http://www.chiefdelphi.com/forums/sh...d.php?t=105679


Here's a thread (and paper) on take-back-half:

http://www.chiefdelphi.com/forums/sh...d.php?t=105965