Listen to this article

As a native of Connecticut and a boater, Northeastern University student Colin McKissick is well aware of an invasive plant that is wreaking havoc in the state’s bodies of water.

Native to Australia, Africa, and parts of Asia, the hydrilla plant found its way to Florida in the 1950s, when it was used to bed aquariums because it doesn’t need much nutrition or light to grow.

Since then, hydrilla has been labeled the “world’s worst invasive aquatic plant” as it spreads and grows rapidly and is difficult to control. The plant can now be found in many parts of the U.S., but Connecticut has been hit particularly hard by the noxious weed.

A 2020 survey of the Connecticut River commissioned by the Connecticut River Gateway Commission found hydrilla in 200 acres in the river’s lower third. Its dense strands make it difficult for native aquatic plants and marine life to thrive, and it often clogs boat propellers.

McKissick, a fifth-year Northeastern student, has experienced this firsthand while boating on the Connecticut River.

“Just going up on the river to get to the boat ports, a couple of times, our propeller would get clogged up with the plant, which is wild because you wouldn’t expect a plant to gum up an 80-horsepower engine,” he said.

---

Learn from Agility Robotics, Amazon, Disney, Teradyne and many more.

---

Northeastern team designs robot to detect aquatic weeds

Enter the Hydrilla Hunter, an autonomous robotic boat outfitted with a hyperspectral camera designed to detect and identify the invasive plant. McKissick helped develop the boat with a dozen other Northeastern engineering students as part of two capstone project classes.

Their goal is to provide the boat to plant scientists at the Connecticut Agricultural Experiment Station to help them more quickly identify and survey where hydrilla can be found and stop it from growing further.

The project is a collaboration between Northeastern’s electrical engineering department, the mechanical engineering department, the Robotics and Intelligent Vehicles Research Lab, and the Connecticut Agricultural Experiment Station. 

Students working under Charles DiMarzio, associate professor of electrical and computer engineering, created the internals of the device, which include an imaging system, a renewable battery, and communication systems.

Students working under Randall Erb, associate professor of mechanical and industrial engineering, developed the boat’s housing and navigation system.

“We came up with a solution to tackle this, which is to automate the detection of the hydrilla and notify the scientists of its location to extract it before it takes over the Connecticut water bodies,” says McKissick, who worked on the electrical and computer engineering side of the project.

How the Hydrilla Hunter works

The robotic boat works in a three-step process. 

First, the user pinpoints where on the map the robot should go with a homebase system separate from the robot. As it hits those waypoints, the robot scans the surface below for hydrilla. If it detects any, the user can pin the location where the plant was detected.

The robotic boat weighs 62 lb. (28.1 kg), can travel at speeds of up to 1.3 mph (2 kph), and can operate for 90 minutes on a charge. It can either be controlled remotely or operate autonomously, explained Daniel T. Simpson, a fourth-year student who worked on the mechanical engineering side of the project. 

“I can manually control it and tell it to move forward, backward, and I can flip a switch and the robot’s software will say, ‘OK, let me look at the GPS waypoints I was told to go to, and let me start going through those points,’” he said.  

Jessica Healey, a fourth-year student working in the mechanical engineering group, said the mechanical and electrical engineering teams worked closely together to develop the project. 

“Throughout the semester, we would meet up monthly, sometimes more frequently depending on what was going on, and just touch base with each other,” she recalled.

Hyperspectral perception helps distinguish plant types

Methods currently used to survey for the invasive plant involve scientists on boats searching for several hours a week using heavy underwater cameras. Distinguishing the plant can also often be a challenge because it looks similar to native species. 

That’s what makes the robot’s hyperspectral camera ideal for this kind of situation, noted Lisa Bryne, a fifth-year student who worked on the electrical and computer engineering side of the project. Hyperspectrical cameras work by capturing a range of wavelength greater than what the human eye can comprehend. 

“These plants look incredibly similar, and the data in the infrared is really valuable to be able to distinguish the plants,” Bryne said.

Experiential discovery drives Northeastern robotics researchers

The idea for the project was born out of discussions the students had with Taskin Padir, professor of electrical and computer engineering and head of Northeastern’s Robotics and Intelligent Vehicles Research Lab.

Through the lab, Padir had already drafted a National Science Foundation proposal with Jeremiah Foley, a plant scientist at the Connecticut Agricultural Experiment Station about using robotics to help solve the hydrilla problem. 

“We’ve been thinking about this problem from an environmental robotics perspective for a while,” Padir said. “It’s a [relatively] unknown yet important problem.” 

Foley said he has big plans for how he’ll like to use the system. Ideally, the station would like to hire a number of technicians to bring the robot to bodies of waters in Connecticut where fishermen typically work, he said. They sometimes unintentionally carry pieces of hydrilla with them where they fish between bodies of water. 

“Rather than getting out to a water body and having us drive around for hours on end, we can send a robot in, and my technicians can do it,” said Foley. “I can stay back in the lab and collaborate with them.” 

Solving these kinds of problems follows the stated mission of Northeastern’s Institute of Experiential Robotics, of which Padir is the director. 

“We always talk about four pillars of experiential robotics, and one of them is experiential discovery,” Padir said. “That doesn’t happen in the lab. It happens outside, when we reach out to stakeholders, when we try to understand the problems that need to be solved. We usually don’t approach the problem by saying ‘Oh we have a robot here. Let’s solve your problem.’” 

“What we do is try to understand the problem, what the bottlenecks are, and come back to the lab to try and create a solution toward solving that problem,” he added. 

The Northeastern students took Padir’s suggestion and ran with it, working directly with Foley to help develop a useful robotic tool. 

“What’s cool about our project is that we actually had a stakeholder say, ‘Hey, we have this huge problem, can you help us engineer a solution?’ That’s where we came in,” said Arjun Fulp, a fourth-year student who was in the university‘s electrical engineering capstone group.