Learning Skills to Navigate without a Master: A Sequential Multi-Policy Reinforcement Learning Algorithm