MARS-Traffic: Multi-Agent Reinforcement learning for Simulation of Traffic
In this project we aim to explore the application of multi-agent reinforcement learning for traffic management. By doing small traffic simulations with multiple agents representing vehicles, the project seeks to analyze their interactions from a game-theoretic perspective. Agents learn and adapt their driving strategy through repeated interactions with other agents and the environment. The focus will be on developing small-scale traffic systems using multi-agent reinforcement learning and analyzing the resulting dynamics.