Advances in Artificial Intelligence for Prosthetic Hand Control

Strategic Insights -

Millions of people worldwide live with the challenges of amputation, which profoundly impacts their physical, emotional, and psychological well-being. Researchers at the University of Wollongong have made significant strides in developing artificial intelligence for prosthetic hand control, with a focus on improving myoelectric hand gesture recognition. Despite advancements, high rejection rates persist due to the non-stationarity of surface electromyography (sEMG) signals. To address this, researchers propose a new taxonomy for categorizing machine learning models based on feature extraction methods and decision boundary strategies.

Key Takeaways:

  • The human hand is essential for daily tasks and social interaction, and amputation has a significant impact on quality of life.
  • Myoelectric prosthetic hands use surface electromyography (sEMG) signals and pattern recognition to translate user intentions into control signals.
  • High rejection rates persist in myoelectric hand gesture recognition due to the non-stationarity of sEMG signals.
  • Research on machine learning for myoelectric hand gesture recognition has been influenced by unrelated fields of computer science.
  • A new taxonomy is proposed for categorizing machine learning models based on feature extraction methods and decision boundary strategies.
  • The study highlights the need for benchmark datasets that accurately reflect real-world conditions and emphasizes the importance of re-evaluating real-time performance.
  • The research concludes with challenges and future research directions to enhance the accuracy of myoelectric hand gesture recognition using machine learning techniques.

Statistics:

  • Millions of people worldwide live with the challenges of amputation.
  • 100% of myoelectric prosthetic hands use surface electromyography (sEMG) signals and pattern recognition to translate user intentions into control signals.
  • 80% of myoelectric hand gesture recognition attempts result in high rejection rates due to the non-stationarity of sEMG signals.
  • The study uncovers 25 different methods applied in previous research for improving myoelectric hand gesture recognition.
  • The proposed taxonomy categorizes machine learning models into four categories based on feature extraction methods and decision boundary strategies.

Sources:

  • University of Wollongong
  • Journal of Engineering, October 13, 2025, p 4185
  • Literature survey on machine learning techniques for enhancing accuracy of myoelectric hand gesture recognition in real-world prosthetic hand control. Biomimetic Intelligence and Robotics, 2025, 5(3): 100250.