At a glance

  • The thumb uses tendon-driven joints with actuators mounted on the forearm. Source
  • A palm-facing camera captures the hand and surrounding visual context. Source
  • A visual imitation policy maps egocentric images to joint positions. Source

Evaluation scope EgoThumb combines a wearable thumb with a vision-based imitation policy for hold-and-manipulate tasks. Source

The wearable thumb, forearm actuation module, palm-facing camera and control interface. Original figure

How it works

  1. The control policy learns to map images from the palm-facing camera to thumb joint positions. Source
  2. The authors evaluated four tasks with the demonstrator and two users whose hands were absent from training. Source

Evidence & results

  • In the training-area evaluation, the demonstrator averaged 0.90 success across the four tasks; the two other users averaged 0.925 and 0.875. Evidence
  • With expanded initial and target areas, average success across users fell to 0.17 for Transfer Ball. Evidence

Limitations & boundaries

  • The authors report that high-speed manipulation and robustness to varied hand approach angles remain untested. Source
  • The evaluation uses binary task success over ten trials per condition; larger evaluations and finer coordination measures are needed. Source

Takeaway

Performance drops on longer tasks when the initial and target areas vary, and high-speed manipulation remains unevaluated. Source

Resources

Figures: the paper's authors. Editorial source and image review: 2026-10-29. Findings are author-reported, not independently reproduced by Robotics Research Hub.

Original abstract

Supernumerary robotic fingers augment human dexterity by enabling a single hand to perform complex "hold-and-manipulate" tasks that typically require bimanual coordination. However, existing devices are often hindered by bulky hardware that limits interaction to large objects and rely on explicit hand motion triggers that lack environmental context. To address these challenges, we present EgoThumb, a fully open-source, vision-driven supernumerary robotic thumb. EgoThumb features a compact, dual tendon-driven finger whose actuation is relocated to the forearm, reducing distal weight while enabling precise active joint control. The system utilizes a visual imitation learning policy that maps egocentric images from a palm-facing camera directly to continuous joint positions. By implicitly decoding human intent from visual cues capturing both the hand and the object of interest, EgoThumb achieves robust autonomous manipulation and strong generalization across diverse users and environments. All hardware designs and software implementations are fully open-sourced at https://ego-thumb.github.io.

Citation
@article{Hou2026_261037582v1,
  title     = {EgoThumb: A Vision-Driven Supernumerary Robotic Thumb for Autonomous Grasp and Holding Assistance},
  author    = {Jianyu Hou and Jiawei Fu and Weikang Wan and Minghui Jiang and Jiajun Li and Hao Su},
  year      = {2026},
  eprint    = {2610.37582v1},
  archivePrefix = {arXiv},
  primaryClass  = {cs.RO},
  url       = {https://arxiv.org/abs/2610.37582v1}
}
Share