A Wearable Sensor Network With Embedded Machine Learning for Real-Time Motion Analysis and Complex Posture Detection

• Published in: IEEE sensors journal 2022-04, Vol.22 (8), p.7868-7876
• Main Authors: Mascret, Q., Gagnon-Turcotte, G., Bielmann, M., Fall, C. L., Bouyer, L. J., Gosselin, B.
• Format: Article
• Language: English
• Subjects: Activity recognition; Algorithms; Classification; Embedded systems; Human activity recognition; Kernel functions; Machine learning; Motion perception; Motion sensors; Network latency; Preconditioning; Radial basis function; Real time; Real-time systems; sensor fusion; Sensor phenomena and characterization; sensor systems and applications; Sensors; signal processing algorithms; supervised learning; Support vector machines; Wearable sensors; Wearable technology; Wireless communication; Wireless sensor networks
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2021.3139588

This work presents an embedded system driven by a wearable sensor network and machine learning to perform complex posture detection and high-precision human activity recognition (HAR) in real time. The presented prototype performs real-time HAR using raw data collected from three wireless wearable motion sensor nodes in parallel. The sensors communicate the measured inertial data to a Raspberry Pi 3 running a pre-trained classifier, which performs motion detection and classification in real-time. Our approach based on raw data and machine learning provides more efficiency and simplicity by decreasing the computation cost and the latency. Our detection and classification algorithm utilizes a new custom preconditioning method called Multi-Mapping Spherical Normalization (MMSN) , in combination with a Support Vector Machine with Radial Basis Function Kernel (RBF-SVM). This new preconditioning algorithm allows to sparse the raw inertial data to increase successful classification results without adding any computational burden. The presented approach achieves a motion classification accuracy of 98.28% for 12 body motions, while allowing for real-time prediction with low latency output (< 20 ms which is 50% less than some studies) for preconditioning and processing thanks to the new MMSN preconditioning method and the use of raw data. We tested our approach with 10 able-bodied subjects.