Reinforcement learning for control

shape