Pyrado’s Documentation

Pyrado is a toolbox for reinforcement learning tailored domain randomization based on PyTorch and numpy. It is designed to be used with RcsPySim in the SimuRLacra framework. The implementations are focus on modularity rather than on performance.

Installation

Please choose one of the options described in the root-level readme file of SimuRLacra.

Where to start?

Check out the provided examples or run some of the existing scripts in Pyrado/scripts.

Content

Environments:

• environments
• barrett_wam
• mujoco
• pysim
• quanser
• rcspysim

Environment Wrappers:

• environment_wrappers

Domain Randomization:

• domain_randomization

Algorithms:

• algorithms
• episodic
• step_based
• meta
• regression

Exploration:

• exploration

Policies:

• policies
• feed_back
• feed_forward
• recurrent
• special

Spaces:

• spaces

Tasks & Rewards:

• tasks

Sampling:

• sampling

Plotting:

• plotting

Utilities:

• utils

• Module Index