This is an attempt at implementing deep successor representations for DeepMind Lab and Vizdoom environemnts based on Deep Successor Reinforcement Learning and Successor Features for Transfer in Reinforcement Learning. The architecture used here is modified to make it work for different environments.