Skip to content

README.md

Latest commit

 

History

History
117 lines (86 loc) · 4.49 KB

README.md

File metadata and controls

117 lines (86 loc) · 4.49 KB

mujoco-c

MuJoCo (Multi-Joint dynamics with Contact) is a physics-based simulation engine with graphics and animation for the C target. This repo defines a base reactor and some example derived reactors. The MuJoCoBase reactor provides a single simulator with graphical animation. The derived reactors customize this base class for particular MuJoCo model files. The MuJoCo menagerie is included as a submodule, so be sure to update submodules:

git submodule update --init --recursive

Prerequisites

Linux (Ubuntu)

  1. Install GLFW, a graphics library used by MuJoCo
apt install libglfw3-dev
  1. Download a prebuilt version of Mujoco v3.2.6 and install it to /opt/mujoco. The following works for x86, for aarch64, change the download path accordingly (https://github.com/google-deepmind/mujoco/releases/tag/3.2.6)
wget https://github.com/google-deepmind/mujoco/releases/download/3.2.6/mujoco-3.2.6-linux-x86_64.tar.gz
tar xvf mujoco-*
sudo mv mujoco-3.2.6 /opt/mujoco

Alternatively, Mujoco can be built from source as explained under the section for MacOS below.

macOS

  1. Install GLFW, a graphics library used by MuJoCo
brew install glfw
  1. Build Mujoco v3.2.6 from source and install to /opt/mujoco
git clone https://github.com/google-deepmind/mujoco.git -b 3.2.6
cd mujoco
mkdir build
cd build
cmake .. -DCMAKE_INSTALL_PREFIX=/opt/mujoco
cmake --build .
sudo cmake --install .

It has also been reported that you can instead install the MuJoCo.app bundle, though we have not tested that.

All platforms

If mujoco is installed to a different location the mujoco.cmake must be updated accordingly.

Using this Library Using Lingo

Install the Lingo package manager, if you haven't already.

In your project, create Lingo.toml file with contents like the following:

[package]
name = "<your package name>"
version = "<your version number>"

[[app]]
name = "<AppName>"
main = "src/<AppName>.lf"
target = "C"
platform = "Native"

[app.properties]

[dependencies]
mujoco-c = {version=">=0.1", git="https://github.com/lf-lang/mujoco-c.git", branch="main"}

In your Lingua Franca application, instantiate one of the library reactors. For example, for the Frank Emika Panda robot, your .lf file should include:

import MuJoCoPanda from <mujoco-c/MuJoCoPanda.lf>
main reactor {
  panda = new MuJoCoPanda()
  ...
}

Then compile and run your app:

lingo build
build/bin/<AppName>

Library Reactors

  • MuJoCoBase: Base class providing navigation of the view and methods to update the scene and advance the simulator. This is not meant to be directly instantiated.
  • MuJoCoAdvance extends MuJoCoBase: Base class providing an advance input to advance the simulation to the logical time and update the scene. This refers to the hello basic demo model, which has a box and a floor.
  • MuJoCoAuto extends MuJoCoBase: Base class that automatically advances the simulation and outputs a tick for each step. This separates the updating of the scene, which is driven by a periodic timer. This refers to the hello basic demo model, which has a box and a floor.
  • MuJoCoCar extends MuJoCoAdvance: Simulator for the car basic demo model, providing a two-wheel vehicle and keyboard controlled driving. This version actively controls the simulator advance.
  • MuJoCoCarAuto extends MuJoCoAuto: Simulator for the car basic demo model, providing a two-wheel vehicle and keyboard controlled driving. This version lets the simulator advance automatically.
  • MuJoCoInvertedPendulum.lf extends MuJoCoAuto: Simple inverted pendulum model.
  • MuJoCoPanda.lf extends MuJoCoAuto: Simulator for a Franka Emika Panda robot.

Demos

Build the demos using lfc or make:

  • Basic: Rectangular object that falls to the floor.
  • Car: Simple drivable car.
  • CarAuto: Simple drivable car.
  • InvertedPendulum: Inverted pendulum demo.
  • Panda: Franka Emika Panda robot doing gyrations.