Reinforcement Learning: Value Learner

Matthew Robards from the Australian National University recently finished up his work at Willow Garage working on reinforcement learning methods for robotics. In particular, created a value learner that can a robot distinguish between good and bad outcomes. Please watch the video above for an overview or view the slides below (download PDF) for more technical information. You can download the software from the value_learner package on ROS.org.