elirichmond

Designing the car physics, racing line generation, and reinforcement learning algorithm to pick the quickest the fastest way around the track

An RL based agent drives around the track and finds the best racing line to minimize total lap time

This project recreates Warzone in Unity and showcases the development of intelligent agents that compete against each other. Various agents were implemented, including naive, adversarial, and Monte Carlo Tree Search agents, with the latter showing the best performance. Code was added to procedurally generate the maps, and even the ability to play against the agents themselves as a player