aiqbench analysis
This research addresses a critical gap in whole-body humanoid teleoperation by introducing a gated residual coordination framework that aligns body and hand movements without relying on explicit object observations. By applying bounded corrections to frozen motion-tracking and retargeting modules, the method significantly reduces geometric errors in wrist and fingertip positions during bimanual tasks. The approach demonstrates robust performance across diverse motion datasets while maintaining high success rates for complex interaction objectives.
Gated Residual Body-Hand Coordination for Whole-Body Humanoid Teleoperation
arXiv:2609.18763v1 Announce Type: new Abstract: Whole-body humanoid teleoperation commonly combines a motion-tracking policy with a separate dexterous-hand retargeter. However, independently generated commands do not explicitly preserve body-hand geometric relations, leading to mismatches in relative wrist poses and fingertip positions during bimanual interaction. We present a gated residual coordination framework that keeps both modules frozen and applies bounded corrections to their outputs. A motion-conditioned action gate allocates correction authority across joint groups, while reference-geometry-dependent reward gates emphasize relevant interaction objectives during training. To establish the nominal body controller on Agile One, we introduce multi-pose morphology calibration that jointly estimates triaxial scales and effector-local offsets, together with staged motion dataset curation for training a SONIC-based tracker. The residual policy uses human motion references, initial commands, and robot proprioception without explicit object or contact observations. In simulation, it reduces wrist and fingertip geometry errors by 39.2-56.3% over direct composition on held-out GRAB motions, while preserving whole-body tracking on AMASS, with success rates of 89.03% without residual coordination and 89.29% with it. Ablations characterize the contributions of reward gating, adaptive correction authority, and separate body and hand correction heads.
Abstract:Whole-body humanoid teleoperation commonly combines a motion-tracking policy with a separate dexterous-hand retargeter. However, independently generated commands do not explicitly preserve body-hand geometric relations, leading to mismatches in relative wrist poses and fingertip positions during bimanual interaction. We present a gated residual coordination framework that keeps both modules frozen and applies bounded corrections to their outputs. A motion-conditioned action gate allocates correction authority across joint groups, while reference-geometry-dependent reward gates emphasize relevant interaction objectives during training. To establish the nominal body controller on Agile One, we introduce multi-pose morphology calibration that jointly estimates triaxial scales and effector-local offsets, together with staged motion dataset curation for training a SONIC-based tracker. The residual policy uses human motion references, initial commands, and robot proprioception without explicit object or contact observations. In simulation, it reduces wrist and fingertip geometry errors by 39.2-56.3% over direct composition on held-out GRAB motions, while preserving whole-body tracking on AMASS, with success rates of 89.03% without residual coordination and 89.29% with it. Ablations characterize the contributions of reward gating, adaptive correction authority, and separate body and hand correction heads.
Robotics (cs.RO)
arXiv:2609.18763 [cs.RO]
(or arXiv:2609.18763v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2609.18763 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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.
Full text under the publisher's open licence · 273 words
Related coverage
DexWeave: Learning Dexterous Humanoid Loco-Manipulation from Human Demonstrations
arXiv · Sep 29, 2026
Towards a General Humanoid Loco-Manipulation Model via Egocentric Whole-Body Human Data Pretraining
arXiv · Oct 2, 2026
I Did This Without Your Knowledge
LessWrong · Oct 17, 2026
Iteratively Tuning hipBLASLt TensileLite Kernels: A Smaller Search, a Faster Kernel
AMD · Sep 9, 2026