dm_control
Google DeepMind's software stack for physics-based simulation and Reinforcement Learning environments, using MuJoCo.
What is dm_control?
Google DeepMind's software stack for physics-based simulation and Reinforcement Learning environments, using MuJoCo.
Overview of dm_control
What it is
dm_control is Google DeepMind's software stack for physics-based simulation and Reinforcement Learning environments, using the MuJoCo physics engine. It provides Python bindings to MuJoCo, a set of pre-built RL environments, an interactive viewer, and libraries for composing custom tasks. The package is open source under the Apache-2.0 license.
Key features
- Python bindings to the MuJoCo physics engine via
dm_control.mujoco. - A collection of Reinforcement Learning environments in
dm_control.suite. - An interactive environment viewer (
dm_control.viewer). - A library for composing and modifying MuJoCo MJCF models in Python (
dm_control.mjcf). - A component-based framework for defining rich RL environments (
dm_control.composer). - Additional libraries for custom locomotion tasks and multi-agent soccer tasks (
dm_control.locomotionanddm_control.locomotion.soccer).
Use cases
- Training and evaluating reinforcement learning agents in continuous control tasks.
- Developing custom physics-based simulation environments for research.
- Interactive visualization and debugging of MuJoCo simulations.
- Multi-agent reinforcement learning experiments, such as soccer.
Getting started
Install from PyPI: sh pip install dm_control Note: dm_control cannot be installed in editable mode. For rendering, at least one OpenGL backend (EGL, GLFW, or OSMesa) must be available. See the repository README for detailed rendering setup instructions.
License and pricing
Licensed under Apache-2.0. Open source.
Based on the project's public README and repository metadata.