AnyDexRT: Calibration-Free Dexterous Hand Retargeting with Few-Shot Human Guidance

2026-07-09

AI Summary

Current motion retargeting methods for dexterous hand teleoperation rely on handcrafted objectives, precise calibration, or global shape matching between human and robotic hands, resulting in poor generalization and extensive parameter tuning. This work proposes a calibration-free, few-shot guided retargeting framework that establishes fingertip correspondences through self-supervised learning, incorporates minimal human guidance in task-relevant regions, and refines pinch postures using a contact-state classifier. By eliminating the need for global shape alignment or manual parameter adjustment, the method significantly improves both retargeting fidelity and operational intuitiveness across diverse dexterous hands, achieving, for the first time, efficient cross-platform teleoperation.

Abstract

Teleoperation is a key interface for controlling dexterous robotic hands and collecting demonstrations for imitation learning. Its effectiveness largely depends on kinematic retargeting, which maps operator hand motions to feasible and intuitive robot hand motions. Existing methods often require hand-crafted objectives, precise calibration, or global shape matching between human and robot hand spaces, making them sensitive to hand-specific tuning and less reliable across different dexterous hands. We propose AnyDexRT, a calibration-free retargeting method for intuitive dexterous teleoperation across human-like dexterous hands. AnyDexRT combines self-supervised fingertip correspondence learning with few-shot human guidance to anchor the mapping in task-relevant regions, and further refines pinch-related poses using a contact classifier. Experiments on diverse dexterous hands and real-world teleoperation tasks show that AnyDexRT improves retargeting quality, reduces manual tuning, and provides more intuitive and efficient control than prior retargeting methods. Project website: https://chenxi-wang.github.io/projects/anydexrt

Problem

• dexterous hand retargeting
• teleoperation
• calibration-free
• kinematic mapping
• human-robot correspondence

Innovation

• calibration-free retargeting
• dexterous hand teleoperation
• self-supervised correspondence
• few-shot human guidance
• contact-aware refinement