Students’ robot aims to streamline marine research and aid the recovery of underwater ecosystems impacted by climate change.
Written and Photographed by Justin Horne
Ten years ago, an unprecedented marine heatwave hit the West Coast of the United States, wiping out nearly 95% of the sunflower sea star population in the waters off California. This mass die-off had a cascading effect on the ocean ecosystem. Without sea stars to keep their population in check, native purple sea urchins surged in numbers, devastating kelp forests. The result? Vast sea urchin barrens: lifeless underwater landscapes where kelp once thrived and are now unable to support the native species that once lived there.
Further north along the Pacific Coast, the waters along the State of Washington have fared somewhat better than California’s due to colder temperatures. Nonetheless, researchers at the University of Washington’s Friday Harbor Labs, a marine biology research center in the San Juan Islands, are racing to understand how to restore balance to these fragile ecosystems.
To aid the scientists’ work, students in the MS in Technology Innovation (MSTI) program at the University of Washington’s Global Innovation Exchange (GIX) have developed an underwater robot and an AI-powered dashboard designed to improve the process of tracking sea star populations. The team, which includes MSTI graduate students Kaiwen Men, Sam Cole, Haokun Feng, and Ziqi Gao, worked with the UW Applied Physics Laboratory and Friday Harbor Labs to experiment with real-world testing and development of their technologies.
The Problem: Manual Tracking is Expensive and Inefficient
Accurate measurements of sea star populations off the Washington coast are critical for understanding the health of the undersea ecosystem. “Right now, a scientific diver has to go down, take pictures, and write down observations manually,” explained Cole. “It takes a lot of training to be a scientific diver, so there aren’t many of them. They use a PVC pipe frame to define their survey area, note the species they see, then surface and record the GPS coordinates where they popped up. They keep doing that over time see how the population is changing.”
AAUS Scientific divers (one of the highest certifications of divers) undergo challenging training, and their services are expensive, costing up to $500 per hour. Given the complexity and cost, data collection is limited, making it difficult to track population changes accurately.
The Solution: An Affordable, AI-Powered Underwater Robot
The team’s solution is a low-cost, semi-autonomous remotely operated vehicle (ROV) that significantly increases measurement accuracy and speed. While ROVs with cameras have existed for decades, they are typically designed for industrial applications like oil rig inspections. Smaller, affordable ROVs have been available to hobbyists and researchers, but there has been little incentive to adapt them for the specific use case of sea star tracking. “There’s no incentive for industry or anyone to go and make this product,” Cole explained. “Researchers and schools just don’t have big budgets.”
The GIX team’s simplified system enables far more researchers to conduct surveys, allowing a greater amount of seafloor to be surveyed and reducing overall research costs.
Upgrading the Tech: Custom Sensors and AI Integration
To achieve these results, the GIX team has customized a research ROV which includes several novel features:
A downward-facing camera for capturing footage of the ocean floor
A GPS system for surface-level tracking
An altimeter (sonar sensor) to measure the distance from the seabed
Machine learning software to detect and count sea stars and urchins automatically
They’ve also built a custom machine-learning dashboard to identify and track sea stars sensed by the ROV. Unlike human divers who must manually count and record observations, the robot autonomously collects visual data and uses AI to analyze species populations. “In our user interface, the user will choose their starting and endpoint for the ROV. And then on that line we’ll start recording data. We have a camera on the bottom, and then that data gets run through a machine learning model that counts the sea stars.” “Our detection problem is actually simpler than some,” Cole noted. “Sea stars have a very distinct shape, whereas something like counting sticks in a forest (like Team Aigen) is much harder.”
From Garbage to Urchins
The project wasn’t always focused on sea stars and urchins. “We started this off as a trash collection project,” Cole explained. . “But they always told us, ‘If you find something more meaningful, do that.’” One conversation changed their direction entirely. “A diver told us about her work in Monterey California removing fishing lines from the seafloor,” Cole recounted. “She mentioned a group trying to combat urchin overgrowth. Their solution was to go with hammers and literally just kill thousands of urchins.” The group had spent months clearing a small 100m² patch of ocean floor, and within a year, kelp growth had rebounded by 90%. “That blew my mind,” Cole said. “But getting permission to do something like that is incredibly difficult. You’re basically asking to remove a native species, and without solid data proving it helps, regulators won’t allow it.”
Field Research at Friday Harbor Labs
To refine their system, the team traveled to Friday Harbor Labs in the San Juan Islands to test their prototype in real-world conditions. “Friday Harbor Labs is a big deal in the marine biology world. They’re the first place in the world to raise sunflower sea stars from the juvenile stage."
“It was freezing, the water was rough, and waves made it hard to control the ROV,” Cole recalled. “At one point, the propellers got tangled in kelp, and we almost lost it. We learned a lot about what the ROV could handle.” A scientific diver accompanied them on the expedition, showing them how the Lab currently does surveys. “It took him almost two hours to get everything prepped, and his camera rig alone was worth $30,000,” Cole said. “It wasn’t even an extreme setup, that’s just what you have to do.”
“The robot and our sensors are only about $6,000. And then with that, you don't need an expert to go in the water. You can even have a student who knows how to pilot it. And then we just open our laptop, calibrate the sensors, and then throw this thing in the water. None of us are divers, but with this technology, we can do kind of the same thing.”
What’s Next?
The project is sponsored by the University of Washington’s Applied Physics Lab (APL), which provided the original BlueROV2 from Blue Robotics. “Aaron Marburg, our sponsor, has been invaluable,” Cole said. “He’s an engineer through and through, so anytime we have a technical problem, he has a solution. And if we need a part with a two-week lead time, he’ll just say, ‘Oh, I have an extra one, come grab it.’”
“The faculty at GIX were invaluable, especially in the beginning when we just literally had no idea what direction we should be exploring,” Cole said. “They also have great connections. There were times we’d wish we could talk to someone at an agency, and our professor (Dr. John Raiti) would say ‘Oh, I know five people that work there, go talk to them.’”
With the robot functional, the next step is refinement and expansion. More field tests are planned for late February, when they hope for better weather conditions. “We have every piece of the puzzle, but they’re all scattered. We’re mainly working on bringing everything together.”
For Cole, the project is personal. “I grew up in Northern California, and in middle school we took these yearly field trips to the Monterey Bay Aquarium. I remember going there and the kelp forests were just thriving. It was one of the only places in the whole world like that. That ecosystem is basically gone. And now, I get to work on a project that can positively impact that and maybe bring that back? That’s really cool.”
Ready to work on projects like these? Learn more about the MS in Technology Innovation or apply to be a student.
Interested in sponsoring a project? GIX capstone projects offer corporate sponsors up to six months of access to international and interdisciplinary teams of graduate students in the University of Washington’s project-based Master of Science in Technology Innovation. Learn more here.
Back to News
