Gait Recognition and Phase Detection Using WearableIMUSensorsand Neural Network Algorithms


Evci F., Saroglu Y., Konukseven E. I.

7th International Conference on Advances in Artificial Intelligence, ICAAI 2023, İstanbul, Turkey, 13 - 15 October 2023, pp.138-143 identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1145/3633598.3633620
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.138-143
  • Keywords: Biomechanics, Machine Learning, Prosthetic Knee Analysis
  • Middle East Technical University Affiliated: Yes

Abstract

This study investigates the application of neural network algorithms in the domain of gait recognition, with a specific emphasis on their effectiveness. Using data acquired from wearable Inertial Measurement Units (IMUs) during controlled walking trials, we have designed and implemented neural network models for the precise identification of sub-phases within the gait cycle. Our research serves as a comprehensive exploration of the capabilities of neural networks in gait analysis. We discussed the details of model development, training, and evaluation, highlighting the achieved accuracy and reliability in sub-phase detection. The outcomes of this study contribute significantly to the assessment of neural network efficacy in gait recognition, highlighting the effectiveness of machine learning algorithms in improving walking performance for prosthetic knee joints, and showing their potential for transformative applications in the field.