FoLD: Force-Informed Learning for Dexterous Articulated Object Manipulation

2026-09-27

AI Summary

This study addresses the failure of transferring human demonstrations to dexterous robots caused by discrepancies in hand morphology and contact dynamics. To overcome this challenge, we propose an explicit force-guided framework that pioneers the computation of compensatory force fields from human demonstrations as priors. Through force-aware learning and residual policy optimization, the framework directly guides the robot to adapt retargeted motions so as to satisfy task-specific contact requirements. Experimental evaluations demonstrate that our method surpasses existing state-of-the-art baselines on public benchmarks and successfully validates skill transfer on a real-world robotic platform.

Abstract

Transferring human demonstrations to dexterous robots remains challenging because differences in hand morphology and contact dynamics often cause retargeted motions to fail at producing the intended object behavior. We present \textbf{FoLD}, a framework for learning dexterous manipulation of articulated objects through explicit force guidance. FoLD compute compensatory force fields from human demonstrations together with the robot's current interaction state, yielding a force prior that promotes the demonstrated object motion. This force prior informs a residual policy that adapts retargeted hand motions to the contact requirements of the task. We evaluate FoLD on a public benchmark for articulated object manipulation, where it consistently outperforms state-of-the-art baselines across tasks and embodiments. We further validate FoLD on real dexterous robot platforms, demonstrating successful transfer of human manipulation skills to robot execution. Here is the link of our project page: https://gghgghgghgg.github.io/FoLD-project-page/.

Problem

• dexterous manipulation
• articulated objects
• human demonstration transfer
• contact dynamics
• hand morphology

Innovation

• Force-Informed Learning
• Dexterous Manipulation
• Articulated Objects
• Residual Policy
• Sim-to-Real Transfer