Eight weeks ago, local seventh and eighth graders in an after-school science club got an up-close look at real scientific research while visiting Northeastern’s Marine Science Center in Nahant, Mass. The field trip, attended by students from the Buckingham Browne & Nichols School in Cambridge, Mass., commenced a semester-long curriculum designed by Northeastern doctoral candidate Daniel Blustein as part of his effort to teach youth how to translate the organizational beauty of nature into control systems for robotic devices.
“I want to inspire students to get into science because I think science is awesome,” said Blustein, who spends most of his time designing robotic lobsters and bees in biology professor Joseph Ayers’ lab. The Biomimetic Underwater Robot Program takes its own inspiration from the neural networks of real animals to design more adaptable robots, he said.
The program consists of two sections. First, the students learn some basic animal biology. They practice observing organisms in their natural habitats by taking objective notes about the animals’ behavior, and they learn about neural networks by playing games in which the students pretend to be neurons passing a signal (in the form of marbles) from one person to the next.
In the second half of the program, they begin to learn basic computer programming and robot design. Using Lego Mindstorms as the platform, the students design a robot that must maneuver through a simple maze. The trick, however, is that its commands don’t come from a remote controller, but rather from the environment around it—just as with real animals.
Blustein has developed a visual programming language to approximate neural networks that interfaces with the standard Lego NXT software. The students can drag and drop bits of this visual code on the computer to tell their robots to do things like moving right if they encounter a wall on the left or moving backward if they encounter a wall up ahead.
This is how most animals work, one of the middle school students said during the final club meeting in which they raced their robots. Cues on one side of the body cause a physical response on the opposite side. Lobsters collect this information with their antennae and claws, but the Lego robots do it with sensors.
Some students designed their robots to respond to the walls using touch sensors. Others used light sensors to track the black tape running along the perimeter of the maze. Still others used a combination of the two.
In the end, the Lego robots encountered some significant challenges as they made their way through the maze. One got caught in a spinning loop, or “doing a jig,” as its designer put it. Another was “doing the right thing, it’s just doing it backwards!” exclaimed a student.
Despite the setbacks, a palpable excitement charged the room as the students cheered on each other’s robots and anxiously watched as the victor crossed the finish line.
Blustein’s own interest in marine life began with regular childhood trips to the National Aquarium in Baltimore. He later became interested in research while working with octopuses at the Seattle Aquarium during college. “Science isn’t complete until you form some kind of communication with the public,” he said. “You need to extend what you do, to share it with others.”