WPI students developing “killer” autonomous underwater robot

AUGUST 31, 2018 —Worcester Polytechnic Institute (WPI), Worcester, MA, recently reported that its undergraduate students are developing an autonomous underwater robot to hunt an invasive species of fish that poses a threat to native fish, coral reefs and other marine ecosystems in the coastal United States.

In a multi-year effort launched last fall, WPI student teams focused on developing a robot designed to autonomously hunt for and harvest lionfish. The project’s goal is to reduce the invasive species’ impact on marine ecosystems from the Caribbean up to the coasts of Florida and Georgia. Additionally, since the fish is a pricey delicacy—fetching as much as $20 per pound—local fishermen could use the robot to generate more income and hunt more efficiently and safely. They currently scuba dive and spear the fish, which have poisonous spines that are painful to touch.

The first student project team, completing their MQP (Major Qualifying Project), worked this past academic year to develop several systems that will enable the submersible robot to distinguish lionfish from other species and spear them.

Lionfish, a colorful aquarium fish native to the South Pacific and Indian oceans, have become a serious problem in the Caribbean and western Atlantic. The National Oceanic and Atmospheric Administration (NOAA) calls them the “poster child for invasive species.” With no predators outside of their native waters, the population is expanding at an astonishing rate, putting additional stress on coral reefs already struggling from the effects of climate change, pollution, and overfishing.

Lionfish also are disrupting native fish populations that are commercially, recreationally, and ecologically important to the region. According to the Ocean Support Foundation, the species, which has a stomach that can expand up to 30 times its normal volume as eats, can reduce juvenile fish populations on a reef by nearly 90 percent in as little as five weeks. Lionfish have been found with more 50 species of juvenile fish in their stomachs.

A graduation requirement for every WPI student, MQPs are team-based design or research projects that give students professional-level experience. The 2017-18 project team consisted of Joseph Lombardi, William Godsey, Brandon Kelly, Nikolay Uvarov, and Andrey Yuzvik, all members of the Class of 2018 who majored in robotics engineering. The faculty advisors were Kenneth Stafford, teaching professor and director of the university’s Robotics Resource Center; Bradley Miller, associate director of the Robotics Resource Center; and Craig Putnam, senior instructor in computer science and associate director of WPI’s Robotics Engineering Program.

“There are economic and environmental benefits to this, and the fish are delicious,” said Kelly, who focused on the robot’s computer vision system.

There are commercial robots that could be used to harvest lionfish, but they are directed by remotely through a tether connection, which can damage fragile reefs. The WPI robot would be untethered and would hunt for fish on its own.

“The goal is to be able to toss the robot over the side of a boat and have it go down to the reef, plot out a course, and begin its search,” said Putnam. “It needs to set up a search pattern and fly along the reef, and not run into it, while looking for the lionfish. The idea is that the robots could be part of the environmental solution.”

The past year’s MQP team used machine learning, advanced computer vision libraries, neural network software, and computer vision models to develop the robot’s computer vision system, which is the key to distinguishing lionfish from the other fish and aquatic species in the area. 

They showed the system thousands of images of lionfish of different colors, taken from different angels and in different lighting conditions, to train it to recognize a lionfish with 95 percent accuracy.

The robot does not yet have a propulsion unit. That will be provided by a commercially available submersible robot that the WPI system is designed to attach to and control.

During the 2018-19 academic year, a second MQP team will focus on the robot’s global navigation system, which will enable it to autonomously establish and navigate a three-dimensional search grid as it seeks lionfish.

“It’s great to have a project that has relevance to society,” Stafford said. “This is a real-world problem and the students attacked it. WPI’s responsibility is to develop students with the confidence and competence to take on global problems. We are giving them the tools to attack these problems.”