Field-Oriented Mecanum Drive
 

An interesting idea I had:

http://web.mit.edu/first/www/semiOLcontrol.pdf

Should give a non-drifting absolute heading. The trig for rotating frame of reference is done off-board to save processor time, sent as analog signals of the sin and cos. This way, all the math done on the RC is straight multiplication.

It looks closed-loop, but it's actually more of an open-loop controller, in that you still trust the Mecanum algorithm to do what it is supposed to do. So, proper weight balance and such still important. The only direct feedback is to cancel out undesired rotation, which seems to be the biggest problem with Mecanum drive.

Feasible? I know I've seen field-oriented drive before. Any tips? Anyone tried something similar with success?

Re: Field-Oriented Mecanum Drive
 

This looks pretty feasible.

I like the idea of outsourcing all of the heavy math to a co-processor. The flow chart makes it very clear.

If we hadn't vowed not to use mecanum this year, I think I'd give this a try :D

Jacob

Re: Field-Oriented Mecanum Drive
 

Looks slick.

Using look-up tables for the trig functions, it's not hard to do this in code if you want to go that direction.

Re: Field-Oriented Mecanum Drive
 

Team 190 did a field oriented control system with mechanum drive for the 2k4 competition. One of the most important things is keeping all 4 wheels on the ground at all times. I wasn't too involved with the system that year, but some of the others on the team might be able to fill you in.

Re: Field-Oriented Mecanum Drive
 

Thanks. I got an email from Mason (190 alum, do you know him?) about it and he gave me some good tips. Looked really cool in '04 (or was it '05? or both?).

Re: Field-Oriented Mecanum Drive
 

Looks nice. We got pretty decent field-oriented control with a magnetic compass last year. (Although the bot went crazy every time we drove by this one locked door in the hallway at school.)

Re: Field-Oriented Mecanum Drive
 

Yea, was 2k5, I was getting years confused.