Skip to content

Tensorflow-Agents RL on custom 2D-topdown racing game

Notifications You must be signed in to change notification settings

Lennart4711/TF_Agents_Racing

Repository files navigation

Agents Racing

Environment

Created a gym environment to train the model on. It contains a car with n sensors and borders that mark the track. The goal is to be able to drive inside this track. A track editor is provided here. You can try it yourself by running play.py

NEAT

Implemented a form of the genetic algorithm NEAT described in this paper in neat.py

TODO

  • Change the reinforce_agent
  • Better reward function than driven distance because it leads to turning in one spot
  • Implement more algorithms
  • Multiple agents in one environment
  • Better folder structure and order by changing how imports are handled

About

Tensorflow-Agents RL on custom 2D-topdown racing game

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages