I'm looking into things to improve my AVC performance for the next competition. My current idea is using a secondary processing unit for some OpenCV object recognition (specifically, the barrels). Last year, I used a single sonar and forced my rover to drive much more slowly through the barrel section before speeding up again (and just hoping I could avoid hitting the hoop or ramp through waypoints and/or sheer luck).
Now I am thinking about using an RPi (enough power?) to do OpenCV object detection in order to be able to keep near full speed on the rover.
A few concerns and questions come to mind:
- What's the best way to get the RPi to tell the APM how to navigate? My thought was to just have the RPi emulate a sonar and tell the APM it needs to turn right or left?
- Best ways to reduce imaging issues from vibration? I would imagine this would be a pretty big issue, especially given I'm trying to stay in the Micro class.
- Does this even make sense? Especially while trying to stay in the 10"x6"x4" Micro size limits?
Any feedback would be greatly appreciated. Thanks!