Robotic Oceanographic Surface Sampler

  ROSS was a gasoline-powered water-sampling robotics platform built around a Mokai jet-drive kayak. It's purpose was to continuously, and often autonomously, gather water data over extremely long distances and/or in locations where human-safety concerns would make gathering it manually too risky. There were a variety of sensors it could be outfitted with depending on the needs of the exact research project and destination, but some common ones were an ADCP for gathering 3D water current vector data, a CTD for measuring water conductivity/temperature/depth, and a high-precision GPS for generating meaningful 3D plots of the sensor data. These kayaks have been deployed to places like the Indian/Pacific Ocean mixing line, and along the active LeConte glacier terminus in Alaska, gathering novel data on how vastly different bodies of water act when mixing.

  In its original configuration, the Mokai kayak's throttle and steering were already drive-by-wire, which made it an excellent starting point for automating. It was also designed for easy transport, breaking down into three major compartments that could easily fit in the back of a short-bed pickup. For our custom hulls, Mokai also thickened the plastic significantly and provided a bare minimum of electronics. This barebones platform was then modified by our team to include a storm-surge-rated intake and exhaust for the engine, a keel to improve rough sea stability, a large alternator, and a plethora of mount points the electronics, batteries, fuel, sensors, and radios.

  In terms of the electronics and software for this project, the kayak itself was centered around a Pixhawk flight controller flashed with a modified Rover variant (this was before a dedicated boat option existed). One pelican-case electronics box housed this controller, a small NUC with UPS, wifi router, radio control receiver, satellite modem, and quite a few custom PCBs for interfacing with external electronics and implementing glue logic/safety overrides. A second box housed nothing but sealed lead-acid batteries, which were charged by the alternator on later revisions of the platform. The PC ran a custom python script, which interfaced with a Matlab GUI over a remote radio link. The kayak could also be overridden with an FrSky RC controller, when at close range, and additionally allowed for direct control without the PC needing to be in-the-loop. To see some of the custom hardware inside of these boxes, check out Nick McComb's design pages for them here! For even more context on ROSS, and history from before I joined the project, check out his summary page.

  I first started on this project by doing what I thought was a one-off help session for Nick, working on an issue he was having getting ROSS's engine to start and shut down properly. I had more experience with engines, and engine control, so I quickly realized that a beefier and high-voltage-rated relay was needed to avoid arc-welding the contacts closed during shutdown. He rolled out a new board revision with those changes and it was the final version used for the rest of ROSS's lifetime. This little taste of the project, and some wishful prodding from Nick, was enough for me to join the team part-time.

  While the original plan for me was to re-write the GUI for ROSS in Python using Qt, it turns out they needed my help on the electrical and firmware side more than anything, so most of my time at the lab was focused on that. I hand-assembled so many of Nick's circuit boards at this lab that I still can pick his out of a lot from design aesthetic alone! I also helped with plenty of wiring harness builds, electrical box fabrication, embedded firmware development, and of course, plenty of electrical and software debugging. One thing that this project taught me very quickly was how difficult it was to make reliable hardware in a high vibration, electrically noisy, and salt-laden environments. The number of PCBs we went through, alongside wiring harnesses, was pretty incredible considering the lengths we went to in order to protect them.

  A very unique aspect of this team/project, and a large part of why I was drawn to it, was that it was about as hands-on as you could possibly get. Doubly so for an undergraduate student! Not only did I get to design and repair a real robot, but it was actually being used for proper scientific research! We would regularly go to Newport, OR for testing and have to make crazy additions and repairs on the fly. This got even more extreme during my deployments to the LeConte glacier, as you had to get creative and fix things with what you had on-hand due to how remote we were. These are experiences that graduate students rarely even get to have, so I'm extremely thankful and fond of the time I spent here. Huge shout out to Nick, again, who made it possible in the first place! Also be sure to check out the scientific paper on this project below!