Thank you for joining me on this Sunday edition of the pool test update! We were extremely happy to have our team president Saagar “Big Saag” Arya and the rest of his team join us with acoustics, and new CS members join us with sonar!
I will not even comment on the arrival since we are now in a consistent swing of things, arriving on-time at 2pm and entering the water quickly! With that, we had two main tasks we wanted to accomplish at the pool today.
Acoustics
Originally scheduled for 30-45 minutes, we ended up going for around 90 minutes. Classic underestimation of time required by first-time pool testers 🙂
We started off with some odd connection gremlins and our robot drifting — turns out, our battery drained a lot faster than we thought it would, and we stuck out this pool test while under 15V for most of the time. Even though the nominal battery voltage is under 15V, we often operate between 15.5V-16V, as that is what we find our robot to operate most stably in (certainly not ideal, but these batteries can handle them no issue).
There also was originally some connection issues with the acoustics that we had to get figured out. This wasn’t as bad, since Saagar and I pre-tested sonar on the Wednesday before the pool test. This was a great idea, and a pattern that we hope to continue to get into, to do as much as we can on land to make sure stuff works to maximize our time in the water.
Sonar did receive some great data though — even though our first (of four) hydrophones appear to have a lot of noise, we were able to identify the pings! Certainly the data looked cleaner without the thrusters running than with, but given that human eyes were clearly able to identify a peak, we have strong hope that code can, as well. The signal, however, was weak, and combined with some strong noise, it is clear that these hydrophone are not the best. Moving forward, we will let the acoustics team cook and try to use some techniques (e.g., filtering, variable gain, using an amplifier), to better detect any pings, but acoustics is showing signs of life, which is super exciting!
Also given that we needed to hold our robot still but couldn’t guarantee it due to low voltage, thank you to those that held the robot still as we collected data! Certainly not an easy task, especially for our swimmer in the ice-cold water.
Sonar
Unfortunately, the sonar team had less time to test than we would have liked, although the most critical part of the test was enabling the sonar team to connect to the robot and understand our testing procedure. This is a crucial step towards younger members taking more leadership on our team, not to mention seeing the results of their code on our robot in the water is also really cool. The code had a few bugs to iron out, but the code did successfully run. We had some trouble with visualizations that we hope can be resolved by the next pool test on Sunday.
Concluding thoughts
A lot of sensor fusion this pool test! Unfortunately, due to sonar taking longer than expected, we were unable to do some task-planning and test / characterize our state and DVL. However, the data and experience we gained was invaluable.
The battery voltage was low throughout this pool test, and we were in a crunch for time, so we did not switch the battery out. However, in the future, the few minutes lost from switching the battery is likely worth it, due to the gremlins we see that creep up whenever the battery voltage is low, and it may result in bad data collection. It’s always good to reset the robot in the middle of the test anyways.
Also, even though I will not comment on coming to the pool test, I will comment this once (hopefully the last time!) on returning from the pool test, simply because of Big Saag’s strength. His ability to get friction on a steep hill and push our entire robot cart up a slippery and steep slope with pine needles and mud is simply astounding. It made our walking path a lot shorter, so Big Saag’s bigness was absolutely crucial here. Turns out, we don’t need Big Saag’s car when we have Big Saag himself.
Looking forward
Moving forward, we will have a lot of concluding projects from CS to test, so we should be more efficient to ensure everybody gets the chance to test. We will also have 8 thrusters on Crush, so once we get that ironed out (yay PIDF tuning!), we can have two robots testing, which will make us much more efficient.
We have one more pool test this semester, scheduled for 11/22 (next Saturday), to be led by President Emeritus Hung Le. Thank you to Hung for stepping up, as I will be in London, England competing at a debate tournament.
This will also be my last pool test update post for a while, since I will be studying abroad next semester in Copenhagen, Denmark. Returning from studying in Tokyo, Japan for a semester is Mathew (CS Lead) and Patrick (Co-President), who I know will run fantastic pool tests. Thank you very much for reading my weekly updates from our rocky starts to now better processes and meaningful progress, and I trust pool tests will be in good hands with Hung next week and Mathew and Patrick next semester. With task planning taking up most of the semester, I imagine you’ll be getting much more frequent updates (hopefully 2 pool tests a week!), and I trust in Mathew and Patrick’s leadership and the rest of the CS subteam’s ability to run successful pool tests with two robots (!) to keep y’all entertained. I look forward to reading along with you, dear reader, as we knock out task after task, from torpedos to octagon, across both robots. Watch out for us at comp >:)
Thank you to Bumjoon Cha for swimming!