Post-Qualifier Code DebuggingTags: think and control
Task: Debug code after the Allen Qualifier
After the qualifier, along with articulation plans, we had a long list of bugs in the code that needed to be sorted out. Most of them were a direct effect of not being able to test the code until the night before the qualifier. In hindsight, there were some issues which needed to be debugged in the turntable and turret.
The first one that we tackled was the turntable wind up and delay. This was one of the bigger problems, as it led to the instabilities seen at the qualifier. These included random jerking to one side, inconsistent speed, and most importantly the delay. As described by Justin, it was a 2-3 second time period in which the turntable did nothing and then started moving. This was especially important to fix for stacking, as quite obviously precision and careful movements are key to this game.
So we started at the source of what we thought was the discrepancy— the rotateTurret() method. This was under scrutiny, as it was the lowest level call, or in other terms the only code that assigns new tick targets to the motor. In the rotate methods that are called by other classes, we assigned a new value to a different value called currentRotation. Once one of the rotate (right or left) methods were called, then the new value would be assigned to currentRotation. Then where the update() method for this turret class was called in the loop, it would call rotateTurret(), which would them assign currentRotationInternal to currentRotation, and then subsequently call the setTargetPositon() giving currentRotationInternal as it’s new tick value target position.
We also started going through the demo mode that was written last year. We have this idea for a great cool demo mode that will be documented once it’s in progress. However, to get there we need a working IMU. We technically had an IMU that worked at the competition, though it was never properly used or calibrated. So, we decided to look into getting the IMUs running. We started by looking at the current demo code and seeing what it could do. Most of it was outdated. But, we did find what we were looking for- the maintainHeading() method, in which we called another method, driveIMU. We then wrote a new maintainHeadingTurret() which works pretty well. Granted, we need to adjust the kP and kI values for the PID, but that is quite easy.
Continue tuning PID values in both the turn-table and turret.