On a May afternoon, Prairie Rose Goodwin ’12 stood before a crowd in the Interdisciplinary Robotics Research Laboratory (IRRL) in the basement of Olmsted Hall. She extended her arm, pointed her finger at one of three plastic cups on a table, and commanded, “Pick that up.” A nearby robotic arm dutifully came to life, reached out, grabbed a cup, and hoisted it in the air. The crowd cheered. Welcome to Vassar’s 2012 Robotics Challenge.
As robotics competitions go, most follow a similar format: teams focus on a standard task, and compete to see who can solve a given problem the best. Vassar’s own competition, which began in 1998, has followed suit. “The faculty would sit around in the fall and come up with a crazy idea for a competition,” explains IRRL director and cognitive science professor Ken Livingston. “Then at the beginning of spring term, we’d set the rules, give students a computer ‘brain,’ some batteries, and a small budget of about $250 per team.”
Past competitions have included tasks such as firefighting (professors created a faux house plan, with a burning candle in one room, and students’ robots had to find it and put it out), road navigation (robots had to drive a road, stop at signage, and navigate a course from start to finish), predator-prey (faculty members built a "rabbit," while students built hunters that had to chase and catch it), and most recently, an artificial ant hill (where "seeds" were dispersed throughout a room, and robots had to identify and collect seeds and deposit them in the ant hill).
But the 2012 edition of the event was different. The competition was rebranded as a “challenge,” with a few notable changes: instead of teams focusing on a common problem and competing to solve it the best, each team conceived its own problem, solution, and business plan. “The whole field of robotics has exploded in the last few years. It’s a full-fledged rapidly growing part of the economy,” says Livingston. “We thought it would be good to start preparing students in a more direct way, of being able to move into a world where robotics is not just a game, but actually an enterprise.”
And so, “we let the students do the work of thinking about what the interesting questions and problems are,” he says. “That’s where the real creative juices happen. It’s a whole other thing to invent your own problem.” Students were tasked with conceiving an idea, researching it, finding out what others have done to tackle it, and then giving it their own novel take.
First, though, they had to convince the faculty that it was a worthy project. “One evening the faculty all sat in Olmsted 301, effectively as venture capitalists, and the students came in as teams with PowerPoint presentations, and had to pitch to us,” Livingston explains. Three groups comprising three students each made their cases. In short, Livingston and his colleagues were impressed.
The teams were cleared to begin the hard part: delivering on their proposed projects. Then, on an afternoon in May, professors, past competitors, students, and the teams crammed into the lab in the basement of Olmsted Hall for the big reveal, where the teams demonstrated just how far they’d come with their projects.
Max Hoffman ’14, Max Lapides ’14, and Evan Cesanek ’13—a veteran of the 2011 competition—showed off their quadcopter, a helicopter with four sets of propellers-cum-rotors. Interest in quadcopters has exploded in recent years, they noted, especially for use in filming and military surveillance. Such copters usually use external navigation cues, such as GPS, but the quadcopter team wanted to design a craft capable of navigating autonomously based on its own sensory data, as biological agents do, such as by recognition of landmarks. After explaining some of the finer points of the copter’s software and aerodynamics, they powered up the motors. With the quadcopter tethered to prevent it from flying out of control and causing a catastrophe in an enclosed room full of people, the copter achieved modest liftoff.
The second team—comprising Abubakar Awumbila ’12, Nishon Subedi ’12, and Tariq Sanda ’12—designed a teaching robot. Finding inspiration in the LEGO Mindstorm and other tech toys, the team saw a gap in the market and sought to build a software-robot system aimed at teaching kids (target age: 9 to 13 years old) how hardware and software interface, and how inputs and outputs are related. They created a customizable robot and tutorial that guides a user through building the robot and writing software algorithms that make it move and navigate in its environment.
Finally, a third team made up of Goodwin ’12, Erik Snow ’12, and Ben Conant ’12 created a robotic arm that could pick up objects—in the case of their demo, blue and red plastic cups—in response to a person’s gestures (such as pointing) and verbal commands. As Goodwin explained, “You point to something and say ‘pick that up.’ The intention is obvious to another person, but not to robots. How can you help them understand natural language and humanistic controls?” The team set out to answer that question using a Kinect for Windows sensor system (microphones, plus infrared and visual cameras) coupled with a robotic arm powered by calibrated servo motors. Soon, the gathered crowd implored, “Let us see it!” Goodwin and her teammates delivered.
“They all made really good progress; everyone learned an amazing amount,” says Livingston in response to the presentations. “We tried to make sure they weren’t getting in over their heads, tried to point them in the direction of things that were manageable and doable. They were all overly ambitious. We knew that when we heard the pitches. But there was an enthusiasm about it, where you don’t want to dampen things. The motivation is different when it’s your idea, and you could see that in the students.”
In the end, the trio of teams went above and beyond the call of duty, says Livingston. Many late nights were spent in the lab in Olmsted’s basement. The result, he says, was some “pretty sophisticated robotics engineering stuff, the kind you normally wouldn’t encounter unless you were in an engineering program or until you were in a graduate degree program.”
Unlike previous years, there were no declared winners and losers in the 2012 Robotics Challenge. The main focus, says Livingston, was on problem solving: “conceiving, building, and producing something that does something you care about at the end, rather than just winning the game.”
This year, the projects of the Challenge will have an opportunity to continue. “Unlike a competition, where it’s finished and you walk away and it’s done, we hope we can sustain these projects over time,” explains Livingston. In fact, one student has already approached him about continuing the robotic arm work. Both Lapides and Cesanek will be back to work on their quadcopter. (Hoffman would be, too, except that he’ll be at Dartmouth for one year as part of the Vassar-Dartmouth dual-degree engineering program.) And the professors, curiously enough, may be interested in picking up where the teaching robot team left off.
“We hope to give people a sense of what it’s like working on a task in the world outside college. Very few things really finish,” Livingston says. “The iPad is never done. There’s always a next version. We want to create some sense of that continuity of investment in a project across time and across generations of students.”