Autonomous Driving Decision Making with Deep Reinforcement Learning
Urban autonomous driving decision making is challenging due to complex road geometry and multi-agent interactions. Current decision making methods are mostly manually designing the driving policy, which might result in suboptimal solutions and is expensive to develop, generalize and maintain at scale. On the other hand, with deep reinforcement learning, a policy can be learned, improved, and adapted to new tasks automatically without any manual designs. In this talk, I will introduce our recent works about urban autonomous driving decision making using deep reinforcement learning techniques. I will talk about how we design and learn a suitable representation for the driving environment, and how to learn a good driving policy automatically.
Jianyu Chen is currently a Ph.D. candidate in Mechanical Engineering at University of California, Berkeley (UC Berkeley). He works with Prof. Masayoshi Tomizuka in Mechanical Systems Control (MSC) Laboratory at UC Berkeley starting from 2015. Prior to UC Berkeley, Jianyu received the B.Eng. degree in Mechanical Engineering from Tsinghua University, Beijing, China. Jianyu's research interests focus on designing decision making systems for autonomous vehicles, with approaches ranging from classical robotics motion planning and control to learning-based techniques such as deep reinforcement learning, imitation learning, and unsupervised learning.