LEADER IN DEVELOPMENTAL ROBOTICS RESEARCH
Accepting the challenge: Assistant Professor Alexander Stoytchev designed and built an upper-torso humanoid robot in less than two months to conduct the ribbon cutting at the Electrical and Computer Engineering Building Addition: Phase I’s dedication ceremony October 2, 2008. “They asked me two months in advance, but I had to prepare for a conference in California, so the real work didn’t start until mid August,” Stoytchev says.
Building the robot: Stoytchev and his students designed the steel fixture for the robot from scratch. They built a complete prototype of the steel fixture out of wood, and prototyped the robot’s body using copper wires and shelf-lining material. The plastic covers for the robot’s body were printed using two rapid prototyping machines that can print 3-D plastic pieces of complex shapes. They also developed the software to control the robot.
Vision of the future: The robot will be used as the platform for Stoytchev’s Developmental Robotics Laboratory. Stoytchev’s goal is to make robots more intelligent and more adaptive than today’s robots, which can only function in limited and highly constrained domains such as factory assembly lines. Stoytchev says that human-inhabited environments are dynamic and often unpredictable and that places a premium on the intelligence that needs to be built into the robot’s control algorithms. “We believe that in the near future robots will work along side humans in homes, hospitals, and universities. This could revolutionize the global economy similar to the way the personal computer did over the last 25 years,” Stoytchev says.
Robots learning like infants: Developmental robotics is one of the newest branches in robotics. This area approaches problems by looking for insights and inspiration in developmental learning patterns of humans and animals. Stoytchev says robots could learn a thing or two from how infants learn. “Infants explore and learn about the world using their intense curiosity drive,” Stoychev says. “In other words, the child’s behaviors act as ‘questions’ to the object about its functionality. The object ‘answers’ these questions by producing perceptual changes that the child can detect. Robots should learn in the same way.”