analiza aiqbench
PHASOR addresses the scalability bottleneck in robot learning by treating action embeddings as first-class representations rather than task-specific intermediates. By factorizing motion into phase and pose components, the method creates an interpretable, embodiment-agnostic manifold that enables effective knowledge transfer across diverse humanoid platforms. This approach offers a promising path toward unified action spaces that improve controllability and reduce the need for extensive retraining on new robots.
PHASOR: Phase-Anchored Universal Action Representations for Humanoid Embodiments
arXiv:2606.01851v2 Announce Type: replace Abstract: Learning a good action embedding space is fundamental to scalable robot policy learning, yet existing methods treat action latents as task-specific intermediates rather than first-class representations. The resulting latents are unstructured, embodiment-specific, and weakly tied to motion semantics, limiting interpretability, controllability, and transferability across robots. We position the action embedding space itself as a first-class design target, with downstream policy quality emerging from representation quality. Exploiting motion's intrinsic periodicity, we factorize it into a phase manifold that captures cyclic structure via FFT-parametric coefficients, together with a pose branch that conditions the manifold on non-periodic configuration detail. Combined with motion-semantic distillation, this factorized structure yields a cross-embodiment motion manifold that is interpretable and embodiment-agnostic by design. Anchoring multiple humanoid robots to a shared human-pretrained manifold then produces a unified action embedding space across diverse platforms, achieving strong cross-embodiment retrieval and consistent gains on downstream robot tasks.
Abstract:Learning a good action embedding space is fundamental to scalable robot policy learning, yet existing methods treat action latents as task-specific intermediates rather than first-class representations. The resulting latents are unstructured, embodiment-specific, and weakly tied to motion semantics, limiting interpretability, controllability, and transferability across robots. We position the action embedding space itself as a first-class design target, with downstream policy quality emerging from representation quality. Exploiting motion's intrinsic periodicity, we factorize it into a phase manifold that captures cyclic structure via FFT-parametric coefficients, together with a pose branch that conditions the manifold on non-periodic configuration detail. Combined with motion-semantic distillation, this factorized structure yields a cross-embodiment motion manifold that is interpretable and embodiment-agnostic by design. Anchoring multiple humanoid robots to a shared human-pretrained manifold then produces a unified action embedding space across diverse platforms, achieving strong cross-embodiment retrieval and consistent gains on downstream robot tasks.
Robotics (cs.RO)
arXiv:2606.01851 [cs.RO]
(or arXiv:2606.01851v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.01851 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 236 cuvinte
Subiecte conexe
UniT: Toward a Unified Physical Language for Human-to-Humanoid Policy Learning and World Modeling
arXiv · 15 oct. 2026
Building a Foundation Stack for General-Purpose Robots
IEEE Spectrum · 13 iul. 2026
Brookhaven Lab Intern Develops AI for Plant Digital Twins
Quantum Zeitgeist · 6 iun. 2026
Universal Pose Pretraining for Generalizable Vision-Language-Action Policies
arXiv · 30 sept. 2026