Physics engine benchmark for robotics applications: RaiSim vs Bullet vs ODE vs MuJoCo vs DartSim
Switch branches/tags
Nothing to show
Clone or download

README.md

SimBenchmark

Build Status

SimBenchmark is a project providing the reliable benchmark results of state-of-the-art physics engines used for contact simulation for various robotic tasks.

We evaluated the most widely-used physics engines for robotics and machine learning applications. The list of the engines is as follows:

You can see the details in webpage.

ANYmal PD Control Benchmark

Installation

SimBenchmark runs on Ubuntu 16.04 LTS.

Notes:

  • RaiSim and MuJoCo is proprietary.
  • Matlab R2018a is required for plotting (optional).

via Docker (recommended)

See docker/README.md

via bash script

  • Run install.sh script by ./install.sh
  • The dependencies are downloaded in lib directory.
  • To use visualization, add environmental variable "RAI_GRAPHICS_OPENGL_ROOT" that points to lib/raiGraphics
  • download MuJoCo v.1.5 in lib directory
    • lib/mjpro150
    • put mjkey.txt in lib/mjpro150/mjkey.txt

Test and Results

We designed the following tests for the evaluation

  • Rolling test: friction model test
  • Bouncing test: single-body elastic collision test
  • 666 balls test: single-body hard contact test
  • Elastic 666 balls test: single-body energy test
  • ANYmal PD control test: articulated-robot-system speed test for quadrupedal robot
  • Atlas PD control test: articulated-robot-system speed test for bipedal robot
  • ANYmal momentum test: articulated-robot-system momentum test
  • ANYmal energy test: articulated-robot-system energy test

Please see our webpage for more details.

Contact

  • Note that this benchmark is done by Dongho Kang and Jemin Hwangho who are the developers of RaiSim.
  • If you have any concern, please contact Dongho Kang