• Welcome to Redshift Project Depot.

Sideways boulder catapult

Started by Louis L, January 22, 2016, 10:07:30 AM

Previous topic - Next topic

Louis L

Last night, I described to BatChen and Ed where the catapult vision tracking algorithm was headed. Part of that discussion involved the notion that we have a "confidence" number attached to each shot. We also talked about adding drive fine tuning to the auto algorithm to move the robot closer or farther if we wanted to get a high probability shot. All of this is difficult to do not because of the controls needed but because there are other robots on the field and the robot can't account for them.

After I got home, I was thinking some more about the practical aspects of this - looking at it from a driver's perspective. What would the driver have to do to relocate the robot in the event it wants a higher probability shot? The upper Goal is taller than it is wide so we're more likely to miss horizontally than vertically. Depending on the shooting arc, we may well have a wide range of distance we can shoot from - if we're not at too steep an angle horizontally. So it's likely that horizontal movement is going to be critical. But without mecanum wheels, this will also be the weakest direction of movement for our robot if we point the catapult towards the front of the robot.

The solution is to point sideways, at right angle to the robot. This would have us drive in semi-circles around the tower to shoot. We can make small adjustments on the horizontal plane by simply driving forwards or backwards (and the vision tracker can do this easily). If movement of the shooter is limited we may only shoot from one side (left or right) so that just means we train to drive in both directions (or spin in place which is easy to do with the 6-wheel drop center drivetrain). This layout would still allow us to get to the safe-spaces against the walls if we need to get there to shoot. In fact we may not even need a rotating shooter!?

On the down side, the act of launching the boulder will rock the chassis more. That can affect consistency, especially if we're pinned against a wall or another robot. Any other negatives for this arrangement?