Mastering Rocket League with Reinforcement Learning A Proximal Policy Optimization Approach

Reinforcement learning (RL) is a subfield of artificial intelligence that involves training an agent to make optimal decisions by interacting with an environment. The agent learns through trial and error, receiving rewards for optimal decisions and penalties for suboptimal ones. The objective of this project was to develop an AI agent capable of playing the game Rocket League. Specifically, the agent was trained to compete in one-on-one (1v1) matches with the intention of optimizing its ability to score goals and prevent the opposing agent from scoring. While the developed agent is technically capable of participating in matches with more players, such as two-on-two (2v2) or three-on-three (3v3), its strategies and behaviors are principally designed for one-on-one (1v1) play. Consequently, it may not perform optimally in these other settings.