Empirical Demonstration of Learning Capabilities

The final component of this project is to investigate the behavior of an RL AGENT under different conditions and different parameter settings. The results of these investigations MUST be described in a written report, to be handed in with the rollout of this project. The goal of this part of the project is to understand how different RL algorithms interact with different environments, how they differ from each other, what factors of the environment do or don't affect learning, etc. Most importantly, the designer should learn about the properties of these algorithms by exploring them interactively.

Note: It can take a substantial amount of time to run some RL experiments. The agent may require tens- or hundreds-of-thousands of steps of experience to converge to its final model, which may require hours of wall-clock time. The designer is strongly encouraged to start the experimental work as soon as possible to ensure sufficient time to complete the experiments.