Dr. Dan Lee was introduced by Ken Glass (M&T 82) to speak about his pioneering work exploring how to apply understanding of how biological entities learn to program increasingly sophisticated and adaptive behavior into robotic systems.
He began with a demonstration of a face recognition task which, while very easy for the audience and even primates to complete, he explained it is very difficult to program a robotic system to perform. He drew contrast between this behavior and other seemingly complex tasks such as playing chess which can be approached using brute force computation. He next went on to describe his current work with Penn’s GRASP lab including a live demonstration of a two foot tall humanoid robot capable of balanced walking, picking itself up from a fall, and the ability to follow sound moving across the room.
This robot is the second generation of Penn’s robotic team which has over the past several years been successful at fielding teams of robot dogs in international robotic soccer championships. Such competitions introduce some friendly competition into the challenging task of making robots capable of responding appropriately to their environment and working collaboratively. Dr. Lee admitted that the sophistication of play is still at a junior league level, but this represents a leap forward from primary school recess level of play just a few years ago. Whether the matches will approach the thrill of human soccer anytime soon is unlikely, but it is clear that Dr. Lee and his team are enjoying the challenge of keeping their robot teams one step ahead of their international competition.
The lecture concluded with another project representing the edge of the possible in autonomous robotics. In 2007 Penn entered the DARPA Urban Challenge which charged teams with the task of developing vehicles able to navigate city streets and interact with other traffic without human intervention. The Penn team’s car “Little Ben” was a Prius outfitted with a roof rack of sensors and a trunk of computers outfitted with software allowing it to make driving decisions up until now considered squarely in the realm of human drivers. Little Ben did his namesake proud and the Penn team was one of only six groups to finish the course despite the comparatively small budget of the team.
Overall Dr. Lee’s talk was an inspiring summary of a rapidly developing realm of technology in which M&T students are already contributing in research and are sure to play a role later in the financing and execution of bringing autonomous robots into homes, offices, sports matches, and roadways.