Abstract

We present Glovity, a novel, low-cost wearable teleoperation system that integrates a spatial wrench (force-torque) feedback device with a haptic glove featuring fingertip Hall sensor calibration, enabling feedback-rich dexterous manipulation. Glovity addresses key challenges in contact-rich tasks by providing intuitive wrench and tactile feedback, while overcoming embodiment gaps through precise retargeting. User studies demonstrate significant improvements: wrench feedback boosts success rates in book-flipping tasks from 48% to 78% and reduces completion time by 25%, while fingertip calibration enhances thin-object grasping success significantly compared to commercial gloves. Furthermore, incorporating wrench signals into imitation learning (via DP-R3M) achieves high success rates in novel contact-rich scenarios, such as adaptive page flipping and force-aware handovers. All hardware designs, software, and datasets will be open-sourced to facilitate community advancements in robotic learning.

Wrench Feedback System

Finger Retargeting

Fingertip Hall Calibration

Haptic Feedback

Design Philosophy

Glovity's development is guided by three core principles that ensure both accessibility and high performance:

High Cost-Performance: The system achieves remarkable functionality at an unprecedented price point of approximately $300. This affordability is achieved through strategic use of minimal high-quality commercial components combined with 3D-printed parts, making advanced teleoperation technology accessible to researchers and developers worldwide.

Ease of Replication: All components can be easily obtained online or manufactured via 3D printing. The system adopts a modular, Lego-style design philosophy, enabling complete assembly in just a few hours with basic tools. This approach democratizes access to sophisticated teleoperation capabilities.

Versatility and Modularity: Glovity accommodates various hand sizes and integrates with diverse robotic platforms. The system operates at high frequencies (up to 100Hz) for wrench feedback, retargeting, and tactile feedback, ensuring responsive control across different manipulation tasks.

Wearable Spatial Wrench Feedback System

Four base feedback modes of the wrench feedback system

Imitation Learning

Adaptive Page Flipping: By integrating wrench data, the policy can learn to inference the next action based on the contact force and realize page turning at different heights.

Force-Aware Handovers: Integration of wrench data enables smooth object transfers between human and robot, with appropriate force modulation throughout the interaction.

Page Flipping

Force-Aware Handovers

Citation

@article{gao2025glovity,
  title={Glovity: Learning Dexterous Contact-Rich Manipulation via Spatial Wrench Feedback Teleoperation System},
  author={Gao, Yuyang and Ma, Haofei and Zheng, Pai},
  journal={arXiv preprint arXiv:2510.09229},
  year={2025},
}

This project represents a collaborative effort to advance the state of robotic teleoperation through innovative hardware design, sophisticated algorithms, and open-source principles. For more information, access to code repositories, or collaboration opportunities, please visit our project resources linked above.