"Albert" Debuts in March Madness for the Mind
Posted: April 13, 2006
Laura Trutoiu '08 hopes that in her lifetime the world will see voice-activated cars for use by physically or visually disabled adults who are now unable to drive. The cars, she imagines, will respond to commands such as, "Take me home." She is collaborating with a Five College team of students on such a project and recently traveled to the tenth annual March Madness for the Mind, a showcase for the nation's top student inventors, to demonstrate the team's work-in-progress, a robotic vehicle named Albert.
Describing March Madness for the Mind, Trutoiu said, "It's impressive to see how well people with different backgrounds and training can collaborate to come up with and implement these amazing ideas. It was definitely a worthwhile experience, and I have only one regret: there were more sessions that I wanted to attend than there was available time."
The project is the brainchild of Hampshire College student Ben Einstein, who was inspired by the 2005 DARPA Grand Challenge, an annual race of self-driven vehicles across 132 miles of difficult terrain in the Southwestern desert. While that race is sponsored by the U.S. Department of Defense to hasten the development of robotics and autonomous vehicles for use in battle, Einstein observed that DARPA technology could be applied to work being done at the Lemelson Assistive Technology Development Center at Hampshire College. The Lemelson Center develops technology to assist physically and visually disabled people.
Einstein's team will eventually produce a full-sized vehicle, officially known as ERNEST (Electronic Rationale and Navigation Embedded Systems Technology). So far, the team has a smaller prototype of ERNEST named Albert (for obvious reasons). Albert is 25 inches long, 14 inches wide, and weighs about 35 pounds. Mounted on its front are its "eyes," two cameras that operate in stereovision--the way human eyes work--to focus on a single point simultaneously. Albert differs from other autonomous vehicles now being developed because it operates only on visual cues, not on other sensory devices such as infrared or sonar. According to Einstein and Trutoiu, it is extremely difficult to program a computer to visually discern important features in the landscape. Albert is now programmed by a color cue to follow a path marked by a red ribbon.
Einstein got started on the project last fall after speaking with Colin Twitchell, director of the Lemelson Center, about his interest in creating an autonomous vehicle for his senior thesis, or "Division-III" project. Twitchell agreed to be his Div-III adviser and put him in touch with several professors including MHC's Claude Fennema, professor of computer science, who specializes in computer vision technology. "He was super interested," said Einstein. "He said I should talk to Laura because she would be very interested in working on the project."
Trutoiu, a computer science and mathematics major with a keen interest in artificial intelligence, said the project "perfectly fit what I want to do. It involves a lot of thinking outside the box. It's not narrow like some computer science projects." Einstein and Trutoiu have been working with two other Mount Holyoke students, Farahnaz Ahmed '06 and Susannah Larrabee '07, and the four are taking a class this spring with Fennema in computer vision and robotics that is centered on the ERNEST project. Trutoiu plans to do her senior thesis on issues relating to computer vision.
While in Portland, Einstein and Trutoiu also showcased a related project, an adaptive steering wheel for a vehicle that allows the vehicle to be operated by a single finger. Einstein, who qualified for his pilot's license while in high school, designed the system based on the yoke of an airplane. The system would enable physically disabled people to drive and could be ready for use in as little as five years.
Einstein and Trutoiu got a lot of positive feedback at the conference and returned to Massachusetts ready to resume their work. Albert's next task will be to follow a sidewalk, a far more complex software challenge. Einstein and Trutoiu are grateful to Fennema, Twitchell, Hampshire College computer science professor Jaime Davila, and others at Mount Holyoke and Hampshire who have supported the project.