Reinforcement Learning for Vehicle Overtake

In this project, we trained a DDQN agent to solve the autonomous vehicle highway overtake problem based on a well-built gym-like highway environment.

Quick Demo

The green vehicle is the ego-vehicle, trained with the DDQN algorithm. Blue vehicles are the traditional human-driven vehicles.

Authors

Haishan Liu, phd candidate in UC Riverside

Acknowledgments

• high environmenthttps://highway-env.farama.org/)