Quote:
Originally Posted by Flex..(Re-Flex)
The IR sensor's zones ranges decrease as you turn farther from the front of the sensor. You can actually increase the IR sensor's accuracy greatly by straddling two zones(like line following with a light sensor).
|
Yep.
In 2010, one of our more brilliant programmers could detect the direction of the IR beacon to within a +/- 2-3 degree accuracy regardless of the starting orientation. I know his basic algorithm (took 2-3 seconds overall), but I do not know the implementation or the tweaking he did to get it so accurate. I'll also say that he put the sensor on a servo so he could accurately adjust the sensor's starting orientation relative to the robot every match (actual orientation determined during off-match calibrations).
Go
here for a basic understanding of the sensor.
1.) Spin in place until the IR beacon is detected at direction 1 or 9 -- note the reading of the beacon and (if available) heading sensor. If the sensor is detected at the start, then go in either direction to the edge (either 1 or 9) and then turn back towards the beacon. The point of doing this is to calibrate the edge conditions.
2.) Keep spinning in the same direction, keeping track of the 'maximum' for all 5 registers of the center sectors (4-6).
3.) Once the IR beacon is lost, stop turning, then start turning back the other way.
4.) The basic way to get back to center is to continue turning until the center values match their maximum.
That year they could shoot a 3" diameter whiffle ball into a 12" hold that was 24" off the ground, even at a steep off-angle, in autonomous. Pretty nifty.