Striving to build ever-more intelligent systems, Pieter Abbeel is paving the way to the next generation of robots that can learn and improve to function outside of traditional manufacturing settings. Abbeel helped start the field of robot learning by cleverly combining learning and optimal control for hard tasks including helicopter flight, manipulating cloth, and surgical knot-tying. He then applied deep learning to robotics by incorporating aspects of reinforcement learning, imitation learning, and unsupervised learning to enable AI-powered robots that have the potential to sense and react to their environment and meaningfully reason through the consequences of their actions in real time. These robots are now playing a role in warehouse order fulfillment, logistics, and supply-chain applications critical to e-commerce.
An IEEE Fellow, Abbeel is a professor and director of the Berkeley Robot Learning Lab at the University of California Berkeley, Berkeley, CA, USA.
