Skip to content

HybridRoboticssupermimic

Repository stars
1,918Star
All-time
#27,374 all time
Fork
156 Fork
Repository created
since 2026
Website
View source

Humanoid whole-body motion tracking with one differentiable world model.

Commercial use OKApache-2.0; keep the license notice
Actively maintainedLast push 1 days ago

About

SuperMimic trains humanoid whole-body motion tracking with a differentiable world model.

It targets research and training scenarios needing G1 whole-body tracking, offering generalized and maximal coordinate task representations. Train a world model and policy as described, then play back a trained model.

Summarized by AI from the READMEREADME

As of 2026-11-22, HybridRobotics/supermimic has 1,918 GitHub stars and 156 forks. Written mainly in Python; licensed under Apache-2.0; last pushed on 2026-11-21.Record updated:

Raw observations · 12
Historical observations
Measured atRepository starsSource
2026-11-22 00:22 UTC1,918Source
2026-11-21 00:31 UTC1,897Source
2026-11-20 00:22 UTC1,857Source
2026-11-19 00:22 UTC1,809Source
2026-11-18 00:22 UTC1,749Source
2026-11-17 00:23 UTC1,660Source
2026-11-16 00:20 UTC1,554Source
2026-11-15 00:22 UTC1,499Source
2026-11-14 00:27 UTC1,435Source
2026-11-13 00:22 UTC1,304Source
2026-11-12 01:30 UTC1,034Source
2026-11-12 01:03 UTC1,029Source

First collected 2026-11-12Measured at 2026-11-22 00:22 UTCSourceLicense source

Compiled from public project information; not independently tested. · Repository Stars, not unique users · License tiers are derived from metadata. Check obligations for your use; unknown does not mean prohibited.