exposure periods

[wireties]

It just occurred to my team that we may be moving our turret while taking pictures (after the first capture). Is anyone out there doing this? Have you played with the exposure times? I can't get to the robot to check this - what is the default exposure time? The data sheet says is is selectable from 1/5000 s to 1/4 s but gives no default value. One can compensate (plug in the width of the rectangles given the height of the left or right side depending on which way the turret is moving) I reckon but is it worth it?

Comments, opinions?

TIA

[Mark McLeod]

One of the defaults is Auto exposure. That'll vary the exposure time.

[Tom Line]

Consider taking a picture, verifying target position, then translating that to some other sensor like a potentiometer to control your turret. This may also help with other problems like balls obscurring the target, or even other robots getting in the way of the target. Of course, you have to go on the assumption that you're not moving your robot, but that's a pretty fair assumption when most people shoot.

[wireties]

Originally Posted by Tom Line

Consider taking a picture, verifying target position, then translating that to some other sensor like a potentiometer to control your turret. This may also help with other problems like balls obscurring the target, or even other robots getting in the way of the target. Of course, you have to go on the assumption that you're not moving your robot, but that's a pretty fair assumption when most people shoot.

we do exactly that and we normally are not moving the robot - the problem is that the first capture occurs, we tell the turret to move and (we think) the turret might be moving when the next capture occurs - we can always force the sequence I guess

thanks

[Tom Line]

Originally Posted by wireties

we do exactly that and we normally are not moving the robot - the problem is that the first capture occurs, we tell the turret to move and (we think) the turret might be moving when the next capture occurs - we can always force the sequence I guess

thanks

Ah, I see. We get around that by turning off vision altogether once we have a verified target locked in. We save the values into another global variable, then zero out the target array and turn off vision entirely.

In fact, the only time vision turns on is when the shoot button is pushed, and then only for 3 frames (the number of frames we've decided we want to check to insure we see the target in the right location).