Pieter Abbeel is a Belgian-American computer scientist, roboticist, and professor at UC Berkeley. He is the Director of the Berkeley Robot Learning Lab and the co-founder of Covariant, an AI robotics startup.
Robotics and Reinforcement Learning
Abbeel earned his Ph.D. from Stanford University in 2008, where he worked with Andrew Ng. His doctoral work pioneered apprenticeship learning (inverse reinforcement learning), enabling autonomous helicopters to perform complex aerobatic maneuvers by observing human pilots. At Berkeley, his research combined deep neural networks with reinforcement learning, creating robots that can learn to fold clothes, manipulate objects, and navigate unknown environments.
Covariant
In 2017, Abbeel co-founded Covariant to commercialize robotic learning. The company developed the "Covariant Brain," a universal AI system that enables industrial arms to pick, pack, and sort arbitrary items in warehouses, solving one of the hardest challenges in physical robotics.
