Perception and Learning
Like people, but with the help of algorithms and AI, robots can learn using a variety of approaches, from examples, through experience or from human teachers. When a robot enters a new environment, it must quickly understand the context and make decisions on how to operate most effectively. Our research develops algorithms for robot perception through new innovations in robotic vision, audio sensors and haptic perception, and improved semantic scene understanding.
Our researchers are also pioneering new developments in supervised learning, unsupervised learning and reinforcement learning to enable robots to learn to operate effectively in new contexts and environments.