Learning to Drive a Bicycle using Reinforcement Learning and Shaping

Learning to Drive a Bicycle using Reinforcement Learning and Shaping
ICML-98 ( gzipped Postscript - 726kb )

Abstract: We present and solve a real-world problem of learning to ride a bicycle. We solve the problem by online reinforcement learning using the Sarsa(lambda)-algorithm. Then we solve the composite problem of learning to balance a bicycle and then ride to a goal. In our approach the reinforcement function is independent of what task the agent tries to learn to solve.