Scholarlink

<p>Current hand tracking technologies often suffer from inconsistent accuracy due to occlusion, electromagnetic interference, and ambiguous signals from soft sensors. These limitations hinder reliable motion capture for advanced human-machine interaction. To overcome them, we propose a liquid metal dynamic wetting-based flexible and wearable inertial skeletal tracking (FWIST) system. The system integrates 16 FWIST units on the hand to capture palm pitch/roll and finger bending angles in real time with high precision. It accurately reconstructs fingertip and joint trajectories, providing comprehensive hand motion analysis. In laboratory tests, a feedforward neural network achieved nearly 100% gesture recognition accuracy on a small dataset, proving the method’s potential. Future work will focus on verifying robustness on larger, diverse datasets. The FWIST approach offers promising applications in teleoperation, robot programming, drone control, and immersive VR/AR experiences.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A liquid metal dynamic wetting strategy for spatiotemporal monitoring of hand movements

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月6日

查看全文
  • Fei Zhan,
  • Nan Li,
  • Ruohan Zhan,
  • Lei Wang,
  • Chun Chen,
  • Guoqing Wang,
  • Feifan Guo,
  • Bowen Tian,
  • Hongbin Zhao,
  • Shuizhong Wang,
  • Guoyong Song

摘要

Current hand tracking technologies often suffer from inconsistent accuracy due to occlusion, electromagnetic interference, and ambiguous signals from soft sensors. These limitations hinder reliable motion capture for advanced human-machine interaction. To overcome them, we propose a liquid metal dynamic wetting-based flexible and wearable inertial skeletal tracking (FWIST) system. The system integrates 16 FWIST units on the hand to capture palm pitch/roll and finger bending angles in real time with high precision. It accurately reconstructs fingertip and joint trajectories, providing comprehensive hand motion analysis. In laboratory tests, a feedforward neural network achieved nearly 100% gesture recognition accuracy on a small dataset, proving the method’s potential. Future work will focus on verifying robustness on larger, diverse datasets. The FWIST approach offers promising applications in teleoperation, robot programming, drone control, and immersive VR/AR experiences.