A Minimalist Retargeting-Guided Reinforcement Learning Recipe for Dexterous Manipulation

2026-07-13

AI Summary

This work addresses the challenge of effectively transferring motion retargeting–based reinforcement learning to dexterous manipulation tasks involving complex contact dynamics. To this end, the authors propose the REGRIND framework, which generates robot reference trajectories from a single human demonstration by preserving hand–object spatial structure and contact relationships. A residual reinforcement learning policy is trained to track object-centric keypoints, and system identification enables zero-shot sim-to-real transfer. This approach represents the first successful application of minimal-retargeting-guided reinforcement learning to dexterous manipulation, achieving human-like, fluent tool use—such as scissor operation and screwdriver rotation—on two multi-fingered robotic hands. Hardware experiments validate its capability to generalize across rich-contact scenarios from simulation to reality without fine-tuning.

Abstract

Recent work in humanoid whole-body control has found success with a simple recipe: retarget human motion to robot kinematic references, then train policies via reinforcement learning (RL) to track them. But how does this recipe transfer to dexterous manipulation? The answer is not obvious, as manipulation involves complex, contact-rich dynamics and requires delicate regulation of contact modes and forces. We present REGRIND, a minimalist retargeting-guided RL pipeline that learns dexterous manipulation policies from a single human demonstration. REGRIND retargets human hand-object motion to a robot reference that preserves hand-object spatial and contact relationships, trains a residual RL policy in simulation to track object-centric keypoints along that reference, and transfers the resulting policy zero-shot to hardware with careful system identification. The resulting policies produce fluid, human-like behavior on two different multi-fingered hands across contact-rich tool-use tasks, including operating a pair of scissors and turning a screwdriver. Through systematic hardware experiments, we identify and analyze the key factors that govern sim-to-real transfer in dexterous manipulation, offering practical guidance for retargeting-based learning in contact-rich settings. Videos and code are available at https://yunhaifeng.com/REGRIND.

Problem

• dexterous manipulation
• retargeting
• reinforcement learning
• sim-to-real transfer
• contact-rich dynamics

Innovation

• retargeting-guided RL
• dexterous manipulation
• sim-to-real transfer
• contact-rich dynamics
• residual policy