holosoma
- Software
- Humanoids and robot learning
- Framework
A framework for training and deploying reinforcement learning control policies on humanoid robots, plus converting human motion capture data into robot motions.
Made by Amazon FAR, United States (checked on github.com/amazon-far (opens in a new tab)).
Source on GitHub: github.com/amazon-far/holosoma (opens in a new tab). Facts read at commit b633c2f (opens in a new tab).
Licenses
- Code
- Apache-2.0 (opens in a new tab)
Read by Robotheca from LICENSE (opens in a new tab) (Apache-2.0) at commit b633c2f (opens in a new tab), on 16 Oct 2026.
- Commercial use
- Yes
What it does
- Trains locomotion (velocity tracking) and whole-body tracking policies using PPO and FastSAC algorithms.
- Runs across several simulators: IsaacGym, IsaacSim, MuJoCo Warp, and MuJoCo for inference only.
- Supports Unitree G1 and Booster T1 humanoid robots.
- Deploys the same inference pipeline in simulation and on a real robot, for sim-to-sim and sim-to-real transfer.
- Retargets human motion capture data to robot motions while keeping interactions with objects and terrain.
- Integrates with Wandb for video logging, automatic ONNX checkpoint uploads, and loading checkpoints directly.
Written by Robotheca from the project's README (opens in a new tab) at commit b633c2f.
Getting started
- Difficulty
- Advanced
- How to start
- The project's README (opens in a new tab)
Maintenance
- Status
- Active: a commit in the last 90 days
- Newest commit
- 14 Oct 2026 on main
- Latest release
- holosoma-v0.0.2 (9 Oct 2026)
Checked on GitHub on 16 Oct 2026.
Links
Robots on Robotheca that use it
No robot on Robotheca is linked to it yet. Links come from a robot's robot.yaml, from its dependency files read as text, or from Robotheca's editors.
Robotheca lists software as a reference: nothing is hosted or run here, and every fact links to where it was read.