Methodology based on machine learning through neck motion and POF-based pressure sensors for wheelchair operation

A.X. Gonzalez-Cely,Blanco-D´ıaz Cristian Felipe,D. Delisle-Rodriguez,Camilo A.R. Diaz,T.F. Bastos-Filho,S. Krishnan,Cristian Felipe Blanco-Diaz

DOI: https://doi.org/10.1016/j.sna.2024.115111

IF: 4.291

2024-02-20

Sensors and Actuators A Physical

Abstract:Polymer Optical Fiber (POF)-based sensors have gained recognition in recent years for biomedical applications because of their low cost, physical properties, and feasibility. A novel methodology is proposed here for classifying neck movements using POF- based pressure sensors and machine learning algorithms. To address this, signal pre-processing, feature extraction, and selection methods are implemented, considering variance, root mean square, and Hjorth parameters. Linear Discriminant Analysis, Support Vector Machine, k-Nearest Neighbors (kNN), and Decision Tree (DT) were used for classification. A maximum accuracy of 0.91 was obtained with kNN and DT for recognizing four neck movements by using the best discriminant nine features. These findings indicate that the proposed methodology is suitable for neck-motion classification using POF-based pressure sensors. Future work will focus on the implementation of this strategy for the design of intelligent Human Machine Interfaces based on electric-powered wheelchairs, which would allow for more independence for people with upper- and lower-limb disabilities.

engineering, electrical & electronic,instruments & instrumentation