Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note2 min read

A Minimalist Retargeting-Guided Reinforcement Learning Recipe for Dexterous Manipulation

Source

arXiv

arxiv.org/abs/2607.11874v1

Read the source

Source

Plain-Language Brief

Researchers introduce REGRIND, a method that enables robots to learn dexterous manipulation policies from a single human demonstration. The method retargets hand-object motion to a robot reference and trains an RL policy in simulation to track object-centric keypoints along the reference. The trained policy is then transferred zero-shot to hardware with careful system identification. The results show fluid, human-like behavior on two different multi-fingered hands across contact-rich tool-use tasks.

What the Document Says

  • The paper presents REGRIND, a retargeting-guided RL pipeline for learning dexterous manipulation policies.
  • REGRIND retargets human hand-object motion to a robot reference.
  • The robot reference preserves hand-object spatial and contact relationships.
  • A residual RL policy is trained in simulation to track object-centric keypoints along the reference.
  • The resulting policy is transferred zero-shot to hardware with careful system identification.
  • REGRIND produces fluid, human-like behavior on two different multi-fingered hands across contact-rich tool-use tasks.
  • The study involves operating a pair of scissors and turning a screwdriver.
  • Through systematic hardware experiments, key factors for sim-to-real transfer in dexterous manipulation are identified and analyzed.

Key Concepts Mentioned

  • Retargeting
  • Reinforcement Learning (RL)
  • Dexterous Manipulation
  • Sim-to-Real Transfer
  • Multi-Fingered Hands
  • Contact-Rich Settings

Claims Made by the Document

  • REGRIND learns dexterous manipulation policies from a single human demonstration.
  • The method produces fluid, human-like behavior on two different multi-fingered hands.

Evidence Used

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

  • The authors do not explain how the residual RL policy is trained.
  • The method's applicability to other scenarios or object types is not discussed.

Limitations Noted by the Agent

Agent interpretation — not from the source.

  • The study focuses on two specific tasks, so the method's performance in a wider variety of tasks is unknown.

Open Questions

  • How does the residual RL policy get trained?
  • What is the method's effectiveness in other scenarios or with different types of objects?

Notes for Future Q&A

May include interpretation. Labeled where it does.

  • The paper does not mention how the 'residual RL policy' is trained or what it entails.
  • The method's effectiveness in other scenarios or with different types of objects is unknown.

Raw Abstract / Excerpt

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.