My b gang, caught pretty much Ebola, had to skip last week. But ts gone now, and I’m sure y’all missed me greatly. Anyways, we are so back! This week’s pool test went decently.
Pre-Pool Test
We made it to the pool test at pretty much 2pm exactly. Oogway was pretty much ready to go at the start of the meeting, which is excellent — and exactly how it should be. We took the sketchy route again, but this time we had a Good Samaritan help us down the steps, so the robot should not have experienced much force. As Jill commented, our new cart has surprisingly good suspension and the robot moved very little in our journey. Nothing but the best for Oogway.
Sonar
Unfortunately, some of the task-planning code was bricked on the Sonar branch, so we spent a good bit trying to debug that. Eventually, we just decided to hold the robot down and collect some sonar readings. Unfortunately, every time we ran sonar, sonar crashed, and the output denoised image was blank. This is likely due to some new configuration setting in the sonar request that has not been fully set up and tested, so we will be sure to ticket that.
Torpedoes
They shot! Our updated torpedoes design (despite having the left and right being mislabeled, nice going mechanical) worked without a hitch, and we actually made it through the hole one time!
Task planning
Task-planning once again worked without a hitch on the robosub-2025 branch. We ran quite a few tasks which included extensive utilization of the new base task structure that Hung created, and it all seemed to work out quite well, so we will plan to merge that branch in (finally!).
Notes on bugs
We had to restart Oogway quite a few times. The modem, DepthAI camera, and USB cameras would randomly disappear. Also, the gyro would disappear from the systems status panel (i.e., show as red when it should be green) even though it was fully working. We’ll keep an eye on these issues, and if they persist next week, we will ticket them.
Sid and Srinath also did see some bubbles come up from the heat-shrunk portion of the battery cable. After closer inspection, we found no water. Thus, it shouldn’t be a concern — we’ll just note it here. In fact, after the pool test, I received a call from Big Saag, gushing over over how good the vacuum seal on the capsule is.
Controls
There was a newly proposed wrench matrix that was created using a Python script, but that was already what was in the CSV for Oogway, so we didn’t need to do any additional testing for that. We were going to do further characterization of the DVL, but the point person for that was unable to make it, so we’ll save that for next week.
Concluding thoughts
Despite being way behind schedule, it is a relief to finally be able to merge robosub-2025 back into main, and have a new base with 1 semester and 1 competition’s worth of new code. We look forward to proceeding with refining the torpedoes task algorithm on Oogway, and our next big challenge will be the 8 thrusters on Crush. For a quick preview, here’s 6-7 things we need to do: set thruster configs, fix wrench matrix, adjust covariance matrix, add physical offsets, tune PID, and re-tune bro zero bias. Good times!
Thank you Sid for swimming! And to Jill for managing our Insta 🙂 Go follow @duke_robotics!
