Reinforcement learning on autonomous humanoid robots door