For this set of pool tests, we dealt with some of the lingering effects of last week’s pool tests, which unfortunately hindered our progress. However, we continue to push forward this summer.
07/05/2025
When we arrived at the pool, we quickly ran into an issue. At the lab, we tested Crush with a battery as usual to confirm the robot still functioned as expected before heading to the pool. However, we failed to actually put the battery into its holder, instead simply plugging it in for power. Therefore, when trying to use the battery in the holder, we found that the 3d print of the holder had deformed, shrinking the size of the holder. Since the holder was designed to snugly fit the battery, it did not easily fit into the capsule. We suspected that this was due to the heat gun used on the wire/3d print area from last week when fixing the burnt fuse holder. For the purposes of the test, we were able to fit the battery in after removing the capsule cover, and monitored the battery temperature frequently on our UI and we regularly took the robot from the pool to manually check the condition of the battery as well. In the future, the print can be filed down to ensure an easier fit.
We then continued on to working on PID tuning. However, we ran into an issue where our depth topic was not receiving any messages, which we suspected to be an issue with the pressure sensor. As the robot did not accurately detect its depth, any attempt to use our task planning system to submerge the robot resulted in a downwards thrust being applied for an infinite period of time. This essentially did not allow us to tune the PID z values, which were crucial to tuning both x and y values. Due to this setback, combined with our initial issues with the battery and limited remaining time, we decided to end the test earlier and head back to the lab.
During our time at the lab, we continued to take IMU readings for crush to measure EMI and drift over time.
07/06/2025
Before heading to the pool, we worked to debug the pressure issues from before. With some debugging support remotely from members of the CS and electrical teams, we were able to narrow down the issue to the peripheral Arduino, which takes readings from the pressure sensor, processes them, and outputs them. With some additional time, we found that the issue was that the pressure sensor was not plugged into the Arduino. It is likely that, while debugging the loss of power from last week’s pool test, the capsule cover caught onto the wire from the pressure sensor and unplugged the robot. After plugging the wire back into the Arduino and running some tests, we headed to the pool.
At the pool, the z values for the PID tuning were tuned, and preliminary values for x were tuned as well. We noticed that when Crush moves in the positive x direction, the nose tends to dip downwards, which causes the robot to unintentionally dive deeper.
Action Items
- Continue to tune the x values for PID (potentially a mechanical solution for stability)
- Continue to tune the y values for PID
- Begin work on the gate task and confirm the status of the style task