[Robotgroup] Drive a Droid proposal... How to detect the

[Bryan Bishop]

On Wednesday 09 July 2008, brooksdesign wrote:
>   For the 4 second backup idea it probably would not be a problem but
> I was also thinking about that issue and the possability you could
> have some sort of dead reckoning that gets verifyed and recalibrated
> after every couple of moves to build up a data base or neural net to
> learn what to expect from differant surfaces. Way more than what
> would be needed for this event but something to think about for the
> future.

Alternatively there's the "semi" preset move idea, where twisting the 
joystick around really does make it move in that direction, but it's 
really more like executing a specific program that does something 
similar. So it wouldn't be completely 360, but instead analog, and all 
of this error stuff would ideally be now positioned to the joystick and 
the user figuring out how much of a push on the stick will make it do 
what (a nonsignificant learning curve of something like 2 seconds). The 
trick would be testing the droid before :-) so that various maneuver 
sequences can be picked that will not cause problems but match up with 
the 16-direction joystick fairly well, etc.

- Bryan
________________________________________
http://heybryan.org/