Robots gaining independence
Their "learning" is the result of Lumelsky's pioneering efforts in the new field of sensor-based motion planning. This process equips robots with a plastic "skin" covered with hundreds of infrared lights and sensors that help the machines sense their surroundings and plot their movements accordingly . . . without human direction.
Potential applications of sensor-based motion are varied, ranging from underwater robots that could guide themselves over fault lines and obstacles, sending back key information about ocean floors, to space-traveling robots that could hover around the outside of a space station monitoring for problems.
Lumelsky has even taught one of his robots to dance. In a video he made to demonstrate this process, the machine, dressed in black, dances swiftly with its human partner, a ballerina clad in pink. As the ballerina approaches, the robot gracefully moves away; as she moves away, the robot follows her. "It was a reaction between two beings," Lumelsky notes.
Sensor-based motion planning is a multidisciplinary field, drawing on experts in engineering, mathematics and computer sciences, explains Lumelsky, who earned his doctorate in applied mathematics and undergraduate and graduate degrees in electrical and computer engineering. He has also done research in pattern recognition and artificial intelligence.
In 1975, when Lumelsky moved to the United States from his native Soviet Union, he was already recognized internationally for his research in pattern recognition and artificial intelligence. He worked for the Ford Motor Company Research Center and General Electric before joining the faculty at Yale in 1985. In 1991 he joined the UW-Madison Department of Mechanical Engineering.
Lumelsky can be contacted at any of the following:
Vladimir J. Lumelsky
304 Mechanical Engineering Building
1513 University Avenue
Madison, WI 53706