Computer scientist Peter Stone’s CS 378 class was launched in the spring of 2007 under unique circumstances. Stone had been asked by the College of Natural Sciences to lead a “research stream” in the Freshman Research Initiative—to teach undergraduates by bringing them directly into the midst of his life as a researcher.
He was also, coincidentally, already helping Austin Robot Technology, a local group of programmers and engineers, develop the artificial intelligence to prepare their self-driving car to compete in the Defense Department’s $2 million “Grand Challenge.”
So CS 378, “Autonomous Vehicles: Driving in Traffic,” was born. Stone’s undergraduates were able to work at the cutting edge of artificial intelligence and robotics research, and “Marvin,” the autonomous Isuzu SUV, was given a shot in the artificial brain that helped propel him further in the competition than all but about a dozen of the 89 teams that entered.
“I can’t draw any parallels to any other class I’ve been in,” say Tarun Nimmagadda, a senior computer sciences major. “We were involved in making history. We participated in a race that is going to be heralded as the big-bang of serious field robotics.”
The students met twice a week for class, but it was outside of class where the work got done. Students worked in smaller groups that dealt with various aspects of the project, and, almost every weekend, students went to the J.J. Pickle Research Campus to test that week’s software on the car itself.
Within six months, the students had helped Marvin pass a series of tests and earn entry to the National Qualification Event in Victorville, Calif. Although Marvin ultimately washed out, he made it further than most of the other teams’ cars—many of which were engineered with much larger budgets, used professional programmers, and specifically barred undergrads from touching the code.
And the class itself was a success by any measure, says Stone. “These could be the students who end up making a mark on the world in the future,” he says, “and I want this to be the kind of course that can be the inspiration that really sets it off.”
CS 378 is back in swing this spring, with a new class of students being inducted into the mysteries of Marvin’s brain, a process that will be repeated for at least the next few years. And each class will build on the one before, says Patrick Beeson, the graduate student who’s taking over the class from Stone, with former students sticking around as research assistants and peer mentors.
“Last year, they had a car that could steer and break and hit the gas, and they had to start from scratch,” says Beeson. “Now, we have a basic platform, and we’re able to focus on much more nuanced problems. The improvements will continue.”
Comments