Scholarlink

Abstract <p>Stroke rehabilitation often requires advanced assistive systems to restore upper-limb functionality. This study presents an electroencephalography (EEG)-driven, six-degree-of-freedom robotic exoskeleton designed to support motor recovery through real-time brain–computer interface control. EEG signals were de-noised using wavelet packet decomposition combined with a Gaussian Mixture Model for improved clarity and feature retention. Extracted features were classified using a Random Forest algorithm, achieving an average accuracy of 85% in detecting motor intentions. The proposed preprocessing led to a 12% increase in SNR and a 15% reduction in RMSE compared to baseline methods. Response Surface Methodology was employed to optimize key kinematic factors joint angle, movement duration, and limb velocity demonstrating significant effects on EEG signal fidelity and control precision. The results highlight the potential of integrating adaptive machine learning with exoskeleton systems to enhance neuro-rehabilitation outcomes.</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

An EEG-Driven Exoskeleton Rehabilitation Robot for Upper Limb Recovery Using Empirical Mode Decomposition: A Response Surface Methodology Approach

  • 期刊论文

  • 发表日期：

    2026年3月4日

查看全文
  • K. Vijayakumar,
  • D. Bubesh Kumar

摘要

Abstract

Stroke rehabilitation often requires advanced assistive systems to restore upper-limb functionality. This study presents an electroencephalography (EEG)-driven, six-degree-of-freedom robotic exoskeleton designed to support motor recovery through real-time brain–computer interface control. EEG signals were de-noised using wavelet packet decomposition combined with a Gaussian Mixture Model for improved clarity and feature retention. Extracted features were classified using a Random Forest algorithm, achieving an average accuracy of 85% in detecting motor intentions. The proposed preprocessing led to a 12% increase in SNR and a 15% reduction in RMSE compared to baseline methods. Response Surface Methodology was employed to optimize key kinematic factors joint angle, movement duration, and limb velocity demonstrating significant effects on EEG signal fidelity and control precision. The results highlight the potential of integrating adaptive machine learning with exoskeleton systems to enhance neuro-rehabilitation outcomes.