Lego Mindstorms Robots as a Platform for Teaching Reinforcement Learning

摘要

This paper examines the suitability of Lego\nMindstorms robotic kits as a platform for teaching the\nconcepts of reinforcement learning. The reinforcement\nlearning algorithm Sarsa was implemented on board an\nautonomous Mindstorms robot, and applied to two\nlearning tasks. The reasons behind the differing results\nobtained on these two tasks are discussed, and several\nissues related to the suitability of Mindstorms as a\nplatform for teaching or experimentation with\nreinforcement learning are identified.