Robots that Learn to Learn
Reinforcement learning and imitation learning have seen success in many domains, including autonomous helicopter flight, Atari, simulated locomotion, Go, robotic manipulation. However, sample complexity of these methods remains very high. In contrast, humans can pick up new skills far more quickly. To do so, humans might rely on a better learning algorithm or on a better prior (potentially learned from past experience), and likely on both. In this talk I will describe some recent work on learning-to-learn for action, where agents learn the imitation/reinforcement learning algorithms and learn the prior. This has enabled acquiring new skills from just a single demonstration or just a few trials. While designed for imitation and RL, our work is more generally applicable and also advanced the state of the art in standard few-shot classification benchmarks such as omniglot and mini-imagenet.
Pieter Abbeel (Professor, UC Berkeley EECS) works in machine learning and robotics, in particular research on making robots learn by watching people (apprenticeship learning) and how to make robots learn through their own trial and error (reinforcement learning). His robots have learned: advanced helicopter aerobatics, knot-tying, basic assembly, and organizing laundry. His awards include best paper awards at ICML and ICRA, Young Investigator Awards from AFOSR, ONR, Darpa and NSF, the Sloan Fellowship, the MIT TR35, the IEEE Robotics and Automation Society Early Career Award, and the Dick Volz Best U.S. Ph.D. Thesis in Robotics and Automation Award. Pieter also founded covariant.ai and Gradescope.