All Tools
H
AgentsFreeOpen Source
HIGHWAYENV
Gymnasium environments for autonomous driving research
MIT
ABOUT
Full driving simulators are heavy for tactical RL, so researchers cannot iterate on lane-change and merge policies quickly. HighwayEnv gives compact Gymnasium scenes for highways, roundabouts, and intersections so agents can practice tactical driving decisions without spinning up CARLA or a game engine.
INSTALL
pip install highway-envINTEGRATION GUIDE
1. Train a DQN or PPO agent on highway-v0 to learn lane changes among other vehicles
2. Benchmark merge and roundabout policies before moving to a high-fidelity driving simulator
3. Prototype collision-avoidance rewards on intersection-v0 with Gymnasium wrappers
4. Teach multi-agent driving tactics by placing several HighwayEnv vehicles in one scene
TAGS
pythongymnasiumreinforcement-learningautonomous-drivingsimulationfarama