Paper Overview
Field: Computer Vision / Robotics Authors: Dihong Huang, Zhenyu Wei, Zhuxiu Xu Published: 2026-06-26 arXiv: 2606.28323
Abstract (Original)
Dexterous manipulation policies can solve individual skills, but composing them to perform multiple tasks with a single hand remains challenging. Adding a new task on top of an existing manipulation skill often imposes conflicting demands on overlapping fingers and contact modes, causing destructive interference between preserving an existing manipulation outcome and executing a new one. We propose DexCompose, a role-aware residual composition framework that reuses pretrained dexterous policies for multi-task manipulation through explicit finger-level action ownership.
Key Points
- Problem: Composing multiple dexterous skills on one hand leads to conflicting demands on overlapping fingers and contact modes, producing destructive interference between skill retention and new-task execution.
- Approach:
- Given two pretrained full-hand policies, DexCompose collects successful post-task states from the first skill.
- It performs release tests over candidate finger masks to identify which fingers are necessary for maintaining the established skill state.
- Two asymmetric residual modules are trained:
- A bounded residual stabilizer for task retention.
- A context-aware residual that adaptively freezes the downstream policy only within the action subspace allocated to the new task.
- Evaluation: Tested on 16 composite dexterous manipulation tasks, covering four object-retention skills and four downstream interactions.
- Results: DexCompose achieves a 77.4% average composite success rate, demonstrating that structured action ownership combined with a dual-residual design is a promising direction for dexterous skill composition beyond traditional policy chaining.
*Auto-collected on 2026-06-30.*