JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

GMM-HMM-Based Eye Movement Classification for Efficient and Intuitive Dynamic Human-Computer Interaction Systems

  • 0Department of Precision Machinery and Precision Instrumentation, University of Science and Technology of China, Hefei 230027, China; jc_xie@mail.ustc.edu.cn (J.X.); crf0114@mail.ustc.edu.cn (R.C.); lzm1224@mail.ustc.edu.cn (Z.L.); jh152728@mail.ustc.edu.cn (J.Z.).
Journal of Eye Movement Research +

|

July 25, 2025

|

July 25, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

This study introduces a new algorithm for real-time eye movement classification in assistive robotic arms (ARAs). The GMM-HMM model improves human-robot interaction accuracy and speed.

Area Of Science

  • Robotics
  • Human-Computer Interaction
  • Biomedical Engineering

Background

  • Eye-tracking technology enhances human-computer interaction (HCI) for assistive robotic arms (ARAs).
  • Current gaze-dependent methods suffer from the "Midas Touch" problem, limiting dynamic control.
  • Real-time eye movement classification is crucial for efficient and accurate human-robot interaction.

Purpose Of The Study

  • To propose a novel Gaussian Mixture Model-Hidden Markov Model (GMM-HMM) classification algorithm.
  • To overcome limitations of traditional methods in dynamic human-robot interactions.
  • To improve the intuitiveness and efficiency of assistive robotic systems.

Main Methods

  • Developed a GMM-HMM classification algorithm for real-time eye movement analysis.
  • Incorporated sum of squared error (SSE)-based feature extraction and hierarchical training.
  • Integrated the algorithm with an assistive robotic arm for gaze trajectory-based path planning.

Main Results

  • Achieved a classification accuracy of 94.39%, significantly outperforming existing methods.
  • Reduced average path planning time to 2.97 milliseconds.
  • Demonstrated effective and intuitive control in dynamic human-robot interaction scenarios.

Conclusions

  • The proposed GMM-HMM algorithm offers an efficient and intuitive solution for dynamic human-robot interaction.
  • This approach enhances the performance of assistive robotic systems.
  • Provides a robust framework for future advancements in HCI for complex real-world applications.

Keywords: