dm_control: DeepMind Infrastructure for Physics-Based Simulation.
DeepMind's software stack for physics-based simulation and Reinforcement Learning environments, using MuJoCo physics.
An introductory tutorial for this package is available as a Colaboratory notebook:
Overview
This package consists of the following "core" components:
-
dm_control.mujoco: Libraries that provide Python bindings to the MuJoCo physics engine. -
dm_control.suite: A set of Python Reinforcement Learning environments powered by the MuJoCo physics engine. -
dm_control.viewer: An interactive environment viewer.
Additionally, the following components are available for the creation of more complex control tasks:
-
dm_control.mjcf: A library for composing and modifying MuJoCo MJCF models in Python. -
dm_control.composer: A library for defining rich RL environments from reusable, self-contained components. -
dm_control.locomotion: Additional libraries for custom tasks. -
dm_control.locomotion.soccer: Multi-agent soccer tasks.
If you use this package, please cite our accompanying tech report:
@misc{tassa2020dmcontrol,
title={dm_control: Software and Tasks for Continuous Control},
author={Yuval Tassa and Saran Tunyasuvunakool and Alistair Muldal and
Yotam Doron and Siqi Liu and Steven Bohez and Josh Merel and
Tom Erez and Timothy Lillicrap and Nicolas Heess},
year={2020},
eprint={2006.12983},
archivePrefix={arXiv},
primaryClass={cs.RO}
}
Requirements and Installation
dm_control is regularly tested on Ubuntu 16.04 against the following Python versions:
- 3.7
- 3.8
- 3.9
Various people have been successful in getting dm_control to work on other Linux distros, OS X, and Windows. We do not provide active support for these, but will endeavour to answer questions on a best-effort basis.
Follow these steps to install dm_control:
-
Download MuJoCo 2.1.0 from the Download page on the MuJoCo website. MuJoCo must be installed before
dm_control, sincedm_control's install script generates Pythonctypesbindings based on MuJoCo's header files. By default,dm_controlassumes that the MuJoCo archive is extracted into~/.mujoco. -
Install the
dm_controlPython package by runningpip install dm_control. We recommendpip installing into avirtualenv, or with the--userflag to avoid interfering with system packages. At installation time,dm_controllooks for the MuJoCo headers from Step 1 in~/.mujoco/mujoco210/include, however this path can be configured with theheaders-dircommand line argument. -
If the shared library provided by MuJoCo (e.g.
libmujoco210.soorlibmujoco210.dylib) is installed at a non-default path, specify its location using theMJLIB_PATHenvironment variable. This environment variable should be set to the full path to the library file itself, e.g.export MJLIB_PATH=/path/to/libmujoco210.so.
Versioning
dm_control is released on a rolling basis: the latest commit on the master branch of our GitHub repository represents our latest release. Our Python package is versioned 0.0.N, where N is the number that appears in the PiperOrigin-RevId field of the commit message. We always ensure that N strictly increases between a parent commit and its children. We do not upload all versions to PyPI, and occasionally the latest version on PyPI may lag behind the latest commit on GitHub. Should this happen, you can still install the newest version available by running pip install git+git://github.com/deepmind/dm_control.git.
Rendering
The MuJoCo Python bindings support three different OpenGL rendering backends: EGL (headless, hardware-accelerated), GLFW (windowed, hardware-accelerated), and OSMesa (purely software-based). At least one of these three backends must be available in order render through dm_control.
-
Hardware rendering with a windowing system is supported via GLFW and GLEW. On Linux these can be installed using your distribution's package manager. For example, on Debian and Ubuntu, this can be done by running
sudo apt-get install libglfw3 libglew2.0. Please note that:dm_control.viewercan only be used with GLFW.- GLFW will not work on headless machines.
-
"Headless" hardware rendering (i.e. without a windowing system such as X11) requires EXT_platform_device support in the EGL driver. Recent Nvidia drivers support this. You will also need GLEW. On Debian and Ubuntu, this can be installed via
sudo apt-get install libglew2.0. -
Software rendering requires GLX and OSMesa. On Debian and Ubuntu these can be installed using
sudo apt-get install libgl1-mesa-glx libosmesa6.
By default, dm_control will attempt to use GLFW first, then EGL, then OSMesa. You can also specify a particular backend to use by setting the MUJOCO_GL= environment variable to "glfw", "egl", or "osmesa", respectively. When rendering with EGL, you can also specify which GPU to use for rendering by setting the environment variable EGL_DEVICE_ID= to the target GPU ID.
Additional instructions for Homebrew users on macOS
-
The above instructions using
pipshould work, provided that you use a Python interpreter that is installed by Homebrew (rather than the system-default one). -
Before running, the
DYLD_LIBRARY_PATHenvironment variable needs to be updated with the path to the GLFW library. This can be done by runningexport DYLD_LIBRARY_PATH=$(brew --prefix)/lib:$DYLD_LIBRARY_PATH.
927 Jan 05, 2023
6 Nov 09, 2022
99 Dec 27, 2022
1 Dec 07, 2021
175 Dec 24, 2022
274 Dec 14, 2022
22 Aug 23, 2022
111 Dec 29, 2022
5 Nov 26, 2022
84 Jan 04, 2023
72 Dec 20, 2022
1 Aug 10, 2022
39 Dec 13, 2022
74 Dec 22, 2022
37 Oct 25, 2022
598 Jan 07, 2023
22 Dec 11, 2022
368 Dec 26, 2022
142 Jan 08, 2023
1.4k Dec 30, 2022