Quote:
|
Originally Posted by Mike Betts
Hutch,
Please read this...
Mike
|
I would never say to use the built in sine function... I don't even use it on desktop computers (I do a lot of work with real time 3D graphics). A simple approximation will handle sine quite readily. With robots, you can even get away with a 7th order taylor series if you don't want to write any sort of decent approximation - physical innacuracies will be greater than approximation inaccuracies then.
And I'm guessing that normal stuff like adds and multiplies are relatively cheap for floats, even on this hardware... Which, in my opinion, makes them pretty much the way to go. While you may get a little more speed with integers, I've never had a problem with the controller's speed and it not finishing the slow loop in 26.2ms while doing normal robot stuff. And using floats saves you that 2K look up table.