For my final project in CS4100 I tried to make a reinforcement learning agent for playing Atari 2600 Frogger. I am using Q-learning with a deep neural network to keep track of the Q-function.
For my final project in CS4100 I tried to make a reinforcement learning agent for Atari 2600 Frogger using Q-learning with a deep neural network. The agent takes one of five actions and receives rewards from the Arcade Learning Environment.
Five experiments were conducted to explore different reward systems and training setups, including reward shaping, preventing exploits, and cropping the game screen to focus on the road portion.
Animated GIFs demonstrate sample playthroughs from the random baseline, Experiment 1, Experiment 4, and Experiment 5.
Reflections on the challenges of training a reinforcement learning agent for Frogger, including sparse rewards, environment quirks, and the importance of reward shaping in guiding the agent’s behavior.