Skip to content

Latest commit

 

History

2 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Highway-RACER

best_policy_highway_eval_checkpoint_190000_episode_4.mp4

A research repository that integrates RACER (Risk-sensitive Actor Critic with Epistemic Robustness) with a continuous-action highway-env to train a distributional Soft Actor-Critic (SAC). The original RACER implementation by Kyle Stachowicz is available at: https://github.com/kylestach/epistemic-rl-release.

Setup

  • Python 3.11 (tested)
  • See requirements.txt for full dependency list

Create and activate a virtual environment, then install the dependencies:

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Training

Start training with the included script:

python -m racer.scripts.train_highway

During training you should observe that the average speed of the agent reaches approximately 35 m/s by around 30k steps.

Evaluation & results

Comparing standard SAC with the RACER distributional SAC shows better performance and faster convergence for the distributional variant:

Quantitative results

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages