Today’s pool test was hopefully the last pool test focused largely on cleaning up functionality before we start testing new code. Overall, it went very smoothly.
Small Testing Items
Today we spent the first 30 minutes of the pool test working on getting joystick controls running to let some new members drive the robot. We ran into some strange issues with the joystick, which we now believe to be related to windows computers. Once we tested on a mac, the issues resolved themselves.
We also started testing different configuration settings for our extended kalman filter and DVL, to see if we could improve our localization. We tested different update frequencies for the Kalman filter, and characterized our current error. Our main test was moving the robot 3m away from a wall, and then back to see if it remained zeroed. We noticed that most of our error occurred when the DVL occasionally disconnected, causing the Kalman filter to assume a constant velocity for ~0.5 seconds.
Mechanically, we had no issues. The humidity in the electrical stack slowly rose throughout the pool test to around 47%, but was still well below an allowable level.
Main Testing
The main goal of our pool test was to confirm that our competition code branch is ready to merge, and to check out some changes in taskplanning package logic / debugging. We successfully completed a barrel roll and coin flip task. When we attempted the torpedo task, the task planning system worked correctly, but we lost connection to the front Depth-ai camera preventing us from locking onto the torpedo banner. We hope to fix this issue and merge our coding changes over the next week.
Conclusion:
We had a very successful pool test! Next week, we hope to continue looking at the DVL and start testing the sonar package some freshmen have been working on.