Demo of Cart-Pole MDP
Demo of Cart-Pole (pole balancing) reinforcement learning problem.
Solved in Model-Based Reinfocement Learning setting using: Vector-valued Matching Pursuit to learn features Vector-valued Linear Regression to learn dynamics
Demos: CP_1ter.mov % 1 iteration of learning (no learning) CP_20iter.mov % 20 iterations of learning (better) CP_30iter.mov % 30 iterations of learning (almost) CP_40iter.mov % 40 iterations of learning (perfect learning)