On October 18 (local time), AI development company DeepMind announced the acquisition of MuJoCo. Multi-Joint dynamics with Contact. As a result of the merger, Musoco is free to download by anyone.
Muzocco is a physics engine suitable for multi-joint dynamics simulation. Musoco was developed by Emo Todorov, a mobility control laboratory at the University of Washington, and has been commercialized for a fee through Roboti since 2015, but according to the merger, it can be downloaded for free under the Apache License 2.0.
Numerous contacts occur in the real world, such as foot contact with the ground when walking and pen and hand contact when writing. However, it becomes a complex task in computer simulation to change by factors such as softness, viscosity, and elasticity of the contact object. Mujoko, which reproduces various contacts and is efficient, is an important choice for robotics researchers, and DeepMind said that it acquired and freed Musoko to support such research. In addition, the Deepmine robotics team has been using Musoco for a long time and will continue to improve it with open source software while collaborating with the community in the future.
A study examining the use of robotics simulations published in January 2021 emphasized the importance of open source tools in advancing research. The author of the paper recommends building an open community-based library for models certified as developing and validating an open-source simulation platform, and DeepMind develops and maintains Musoco as an open-source community-based project according to the recommendations. Related information can be found here.