aiqbench analysis
The paper applies reinforcement learning and policy distillation to humanoid manipulation, using the robot's arms and torso as contact surfaces rather than only its hands. The reported gains (90% task completion versus 18% for direct retargeting and 4% for a generalist whole-body controller) come from a learned sim-to-real pipeline, but they are measured on a single robot platform, so generalization across hardware and tasks remains untested. For ML readers, the useful takeaways are the any-to-any repositioning objective and the expert-to-single-policy distillation approach.
Whole-Body Dexterity
arXiv:2610.22184v1 Announce Type: cross Abstract: Current humanoid dexterous manipulation remains largely confined to end-effectors and simple grasps, falling short of humans' ability to engage the entire body by bracing and sliding bulky objects against the arms and torso. We introduce Whole-Body Dexterity (WBD), a general framework that uses the robot's body and appendages as contact surfaces to grasp and manipulate objects. Our sim-to-real pipeline first generates a dataset of feasible whole-body grasps, then trains reinforcement-learning experts on a general any-to-any repositioning objective by sampling initial and goal grasps from the dataset, and distills these experts into a single policy that uses proprioception and object geometry. We show that the distilled policy transfers zero-shot to a real Unitree G1, where it repositions a basketball and a cardboard box to regions around the torso and completes a long-horizon task that stows a box under one arm, carries it from a shelf to a table, and places it. The controller operates interactively, steered by coarse motion references from a human teleoperator. Across real-world whole-body manipulation tasks, the distilled controller completes 90% of trials, compared with 18% for direct retargeting and 4% for a generalist whole-body controller.
Abstract:Current humanoid dexterous manipulation remains largely confined to end-effectors and simple grasps, falling short of humans' ability to engage the entire body by bracing and sliding bulky objects against the arms and torso. We introduce Whole-Body Dexterity (WBD), a general framework that uses the robot's body and appendages as contact surfaces to grasp and manipulate objects. Our sim-to-real pipeline first generates a dataset of feasible whole-body grasps, then trains reinforcement-learning experts on a general any-to-any repositioning objective by sampling initial and goal grasps from the dataset, and distills these experts into a single policy that uses proprioception and object geometry. We show that the distilled policy transfers zero-shot to a real Unitree G1, where it repositions a basketball and a cardboard box to regions around the torso and completes a long-horizon task that stows a box under one arm, carries it from a shelf to a table, and places it. The controller operates interactively, steered by coarse motion references from a human teleoperator. Across real-world whole-body manipulation tasks, the distilled controller completes 90% of trials, compared with 18% for direct retargeting and 4% for a generalist whole-body controller.
Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Systems and Control (eess.SY)
arXiv:2610.22184 [cs.RO]
(or arXiv:2610.22184v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2610.22184 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 · 291 words
Related coverage
Google DeepMind Ships Three Physical AI Models For Whole Body Control, Dexterity And Multi Robot Collaboration
MarkTechPost · Jul 30, 2026
Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics
Latent Space · Oct 13, 2026
PanTact: Teaching Whole-Body Dexterity with a Wearable Contact Interface
arXiv · Oct 12, 2026
DexRoam: Learning Mobile Bimanual Dexterous Manipulation from Egocentric Whole-Body Human Demonstrations
arXiv · Sep 29, 2026