This is a reinforcement learning agent trained on double deep Q-network and delicate reward function that can automatically play laser hockey, which is similar to the Atari Game “Pong”, against computer (PD controller or other agents) or human.
Use Game_Start.ipynb notebook file to launch a game with the trained agent model or watch the demo video to see the performance directly.
Run run_PR_DQN.py script to train your own agent.
