aiqbench analysis
HANDOFF introduces a distilled mixture-of-experts controller that simplifies the command interface for humanoid robots, enabling natural-language-driven tasks without requiring task-specific data or fine-tuning. Evaluated on the Unitree G1, the system matches state-of-the-art velocity tracking while expanding robust manipulation workspaces and integrating safety-filtered motion with fall-recovery capabilities. This approach addresses a critical bottleneck in real-world deployment by decoupling high-level planning from low-level control constraints.
HANDOFF: Humanoid Agentic Task-Space Whole-Body Control via Distilled Complementary Teachers
arXiv:2606.06493v3 Announce Type: replace-cross Abstract: For a humanoid robot to be deployed in the real world, the choice of command space (i.e., the interface between task planning and whole-body control) is crucial. Existing whole-body controllers typically demand dense kinematic or spatial references that planners struggle to synthesize from task semantics. We instead propose a compact, explicit interface that is intuitive, general, modular, and expressive enough for diverse loco-manipulation skills. To this end, we introduce HANDOFF, a single humanoid whole-body controller that follows this interface and is distilled via multi-teacher KL distillation under a context-conditioned gating scheme into a mixture-of-experts student from three complementary specialists: whole-body motion tracking with safety-filtered data, locomotion, and fall-recovery. On the Unitree G1, HANDOFF matches state-of-the-art velocity tracking and offers one of the largest robust manipulation workspaces. We further demonstrate hardware feasibility through multiple natural-language-driven task roll-outs, powered by a VLM-driven agentic planner with no task-specific data or controller fine-tuning.
Abstract:For a humanoid robot to be deployed in the real world, the choice of command space (i.e., the interface between task planning and whole-body control) is crucial. Existing whole-body controllers typically demand dense kinematic or spatial references that planners struggle to synthesize from task semantics. We instead propose a compact, explicit interface that is intuitive, general, modular, and expressive enough for diverse loco-manipulation skills. To this end, we introduce HANDOFF, a single humanoid whole-body controller that follows this interface and is distilled via multi-teacher KL distillation under a context-conditioned gating scheme into a mixture-of-experts student from three complementary specialists: whole-body motion tracking with safety-filtered data, locomotion, and fall-recovery. On the Unitree G1, HANDOFF matches state-of-the-art velocity tracking and offers one of the largest robust manipulation workspaces. We further demonstrate hardware feasibility through multiple natural-language-driven task roll-outs, powered by a VLM-driven agentic planner with no task-specific data or controller fine-tuning.
Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2606.06493 [cs.RO]
(or arXiv:2606.06493v3 [cs.RO] for this version)
Full text under the publisher's open licence · 165 words
Related coverage
HANDOFF: Humanoid Agentic Task-Space Whole-Body Control via Distilled Complementary Teachers
arXiv · Jun 9, 2026
HANDOFF: Humanoid Agentic Task-Space Whole-Body Control via Distilled Complementary Teachers
arXiv · Jun 6, 2026
Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics
Latent Space · Oct 13, 2026
HANDOFF: Humanoid Agentic Task-Space Whole-Body Control via Distilled Complementary Teachers
arXiv · Sep 30, 2026