Deep Reinforcement Learning with Successor Features for Navigation across Similar Environments | Read Paper on Bytez