Skip to content

A modular python implementation of various policy gradient algorithms for use in control problems on experimental quanser robots. This repository includes implementations of Maximum A Posteriori Policy Optimization, Trust Region Policy Optimization and a draft for Soft Actor Critic.

License

Notifications You must be signed in to change notification settings

kosmitive/abstract_rl

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Abstract RL

Description

A modular python implementation of various reinforcement learning algorithms for use in control problems. There is a distinction between discrete and continuous action spaces. For example Deep Q Network is discrete. Nevertheless automatic discretization is supported by the framework already. So it can be used for control problems as well.

The following algorithmic papers were used:

Usually policy gradients algorithm can be successfully combined with value function methods into a actor critic architecture. Additionally the following methods are used for estimating the v and q function:

Installation

To install the abstract_rl package, start by checking out the repository and install it locally.

cd abstract_rl
pip install -e .

Usage

To create a new script utilizing TRPO or MPO simply include:

git clone [email protected]:kosmitive/abstract_rl.git
cd abstract_rl
pip install -r requirements.txt
pip install -e .

About

A modular python implementation of various policy gradient algorithms for use in control problems on experimental quanser robots. This repository includes implementations of Maximum A Posteriori Policy Optimization, Trust Region Policy Optimization and a draft for Soft Actor Critic.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages