Deep learning has enabled countless successes on a number of applications, including computer vision, speech recognition, and machine translation. While many of such systems are already deployed, the field of robotics has yet to reap the benefits that other areas have experienced -- in general, standard deep learning methods are not directly applicable to robotic learning. In this talk, I will discuss how deep unsupervised and supervised learning techniques can enable robots to learn manipulation skills from raw pixel inputs. In particular, I will show how robots can learn mental models of the visual world and imagine the outcomes of their actions, and how unsupervised learning can be used to allow robots to build internal representations of moving objects. I will end by sharing my vision for the future of deep robotic learning and hypothesize what machine learning advances we need before reaching human-level AI in robotics.
Chelsea Finn is a PhD student at UC Berkeley, studying machine learning for robotic perception and control. She is interested in how learning algorithms can enable robots to autonomously acquire complex sensorimotor skills. Before joining Berkeley AI Research, she received a Bachelors in Electrical Engineering and Computer Science at MIT.