Scholarlink

<p>This study aims to enhance assistive technologies by predicting head movement intentions in real-time using surface electromyography (sEMG) signals and machine learning algorithms. The primary motivation is to improve the responsiveness and accuracy of gaze tracking systems for individuals with physical disabilities. Six healthy adult males participated in the experiments, with their head and neck muscle activities recorded using a high-speed optoelectronic motion capture system (Vicon Vero/Nexus) and wireless sEMG sensors (Delsys Trigno). Reflective markers were positioned on the subjects’ heads and shoulders, and mini-size sEMG sensors were placed around the eyes and neck muscles. The experimental procedure involved the participants sitting 1.5 meters from a visual guide, performing head movements in four directions, and holding each position for three seconds. Four EMG feature sets were created for analysis, combining signals from different muscles and time intervals. Various machine learning models, including kernel naïve bayes, gaussian naïve bayes, bagged trees, and subspace KNN, were applied to predict head movement states. The subspace k-nearest neighbors (KNN) model applied to EMG set 3 achieved the highest classification accuracy of 78.0%. The study demonstrates the potential for sEMG combined with advanced computational techniques to significantly improve the real-time prediction of head movement intentions, offering valuable applications in human-computer interaction, virtual reality, and assistive technologies.</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

Neural Network-based Head Movement Prediction Using Electromyography Signals

  • 期刊论文

  • 发表日期：

    2025年2月5日

查看全文
  • Yundong Kim,
  • Jirou Feng,
  • Taeyeon Kim,
  • Gibeom Park,
  • Kyungmin Lee,
  • Seulki Kyeong

摘要

This study aims to enhance assistive technologies by predicting head movement intentions in real-time using surface electromyography (sEMG) signals and machine learning algorithms. The primary motivation is to improve the responsiveness and accuracy of gaze tracking systems for individuals with physical disabilities. Six healthy adult males participated in the experiments, with their head and neck muscle activities recorded using a high-speed optoelectronic motion capture system (Vicon Vero/Nexus) and wireless sEMG sensors (Delsys Trigno). Reflective markers were positioned on the subjects’ heads and shoulders, and mini-size sEMG sensors were placed around the eyes and neck muscles. The experimental procedure involved the participants sitting 1.5 meters from a visual guide, performing head movements in four directions, and holding each position for three seconds. Four EMG feature sets were created for analysis, combining signals from different muscles and time intervals. Various machine learning models, including kernel naïve bayes, gaussian naïve bayes, bagged trees, and subspace KNN, were applied to predict head movement states. The subspace k-nearest neighbors (KNN) model applied to EMG set 3 achieved the highest classification accuracy of 78.0%. The study demonstrates the potential for sEMG combined with advanced computational techniques to significantly improve the real-time prediction of head movement intentions, offering valuable applications in human-computer interaction, virtual reality, and assistive technologies.