DRL-Navigation

A customized grid map-based navigation simulation platform following the Gym API.

We reproduce a light RL training framework from OpenAi Five. As seen in the following, the structure of our framework is totally the same as the paper shown. 3 key ingredients in the RL training process, Forward Module, Backward Module, and Env Module are separated.