Improve Ball Indexing

[iamr-gh]

Magazine currently has a failure state where it will create an air pocket as it "skips" a ball because the magazine ran and the ball didn't come in. There are 2 main ways I see to improve this:

1. Mitigate failure state by tuning in IRs better, so they don't trigger when a ball isn't in position.
2. When failure state occurs, detect it using the intermediary IR sensors(for example, TFTF would mean a hole, FTTF would be goal for 2 balls) and run magazine (slowly) backwards until hole is eliminated. Make sure there isn't another failure state where this happens cyclically.

If anybody has a better solution please write it out, but I think you either do the above, or you implement some weird timing logic, which would slow down whole system.

Implement both, test your solution originally by hand feeding, then eventually with balls added in by intake. Start with magazine speed slow and then turn it up as your system works.

[aballard-mars]

I have a couple thoughts, but they'll require some testing.

1. Consider changing the polling frequency for the IR inputs. They are currently 30ms (I thought they were 20.) They are analog devices, so they shouldn't have the concerns with polling too frequently that the wheel encoders have. However, the impact this has will be limited by the loop frequency on the motors in the magazine subsystem. That may need tweaked as well. I'm not sure if we can get the computer to respond quickly enough to detect gaps from adjacent balls, especially if they are compressed against each other, but we can try.

2. We may get better threshold values if we run the magazine and graph the raw inputs across a series of balls. Even if the balls are touching one another such that the IR never sees an empty gap between balls, hopefully we can thresholds that ignore the ball edges. Keep in mind, this may need to be variable by sensor also. For instance - we may need the top sensor to trip when it sees ANYTHING to prevent feeding a ball to the turret but the middle or input sensors to only trip when a ball is directly in front of the sensor.

[wmstanton]

We discovered that changing the threshold value for the Bottom IR sensor from 1 to 2 helped to reduce the gap between balls and the robot is holding 4 balls for the couple of trials we've ran.
There is still about a 1 inch gap between two balls in the middle of the magazine, which is preventing the robot from holding five balls. Mechanical is currently looking at solution. Need to commit the change to the threshold. Also should look at if using middle IR sensors in the code will allow us to close the gap between balls.