One frozen controller for humanoids: spectral skills enable motion composition and language control
daibond_alpha · x · 2026-11-10
Spectral Skills, a paper under double-blind review at ICLR 2027, learns what a high-level behavior model should send to a whole-body controller: a "spectral skill" command acquired by predicting how motion continues.
Key ideas and results:
- One frozen controller, one loop: on a Unitree G1 (sim + real), skills can be switched at runtime and steered along spectral directions (added/removed) without retraining, enabling skill composition and chaining.
- Efficiency: once the gradient is stable, pathwise SNR is 30× higher than score-function estimation, yielding 64× fewer imagined steps than PPO on the same model rollouts; a terminal critic keeps rollouts short.
- Language-directed control: a language-conditioned planner predicts skill sequences that the controller executes.
- The project page ships an in-browser interactive demo: policy in ONNX Runtime Web, physics in MuJoCo compiled to WebAssembly (50 Hz control on 200 Hz physics, 45 MB), letting users steer the tracker live.
The authors note the demo is a demonstration, not a benchmark. Paper and code are available.
More from Embodied
- Calibrex, an open-source robot sensor calibration checker, hits 50 GitHub stars — rsasaki0109 · 2026-11-10
- SpaceX says Grok detects aerospace foreign object debris with 99.9% accuracy vs humans' 94% — elonmusk · 2026-11-10
- HermesEnabler2 confirmed working on Apple Watch Series 10, unlocking Hermès watch faces — gregbarbosa · 2026-11-10
- Fable 5.2 hits 64.9% on RoboDojo, beating purpose-built robot AI systems — CyberRobooo · 2026-11-10
- Caltech robot builds a structure in orbit autonomously, a first for space construction — GregCook2011 · 2026-11-10
- Robot paints original watercolors and ships them with telemetry — a one-man robot art club — repligate · 2026-11-10