Motion Classification Associated with Smart Textile Sensors in Sports Training Garments

Abstract

The integration of smart textile sensors into sports training garments represents a significant advancement in athletic performance monitoring and injury prevention. While laboratory-based studies have demonstrated the high accuracy of these wearable systems in controlled environments, there remains a critical gap in understanding their reliability and motion classification capabilities under unconstrained, real-world field conditions. This paper addresses this gap by presenting a comprehensive field evaluation of motion classification using piezoresistive smart textile sensors embedded seamlessly into athletic apparel. The study captures continuous movement data from athletes engaging in dynamic, sport-specific activities, transitioning away from the traditional constraints of treadmill or strictly regimented laboratory protocols. Through the application of machine learning classification pipelines, encompassing both time-domain and frequency-domain feature extraction methodologies, this research rigorously assesses the robustness of the sensor data against environmental noise, mechanical artifacts introduced by intense physical exertion, and sweat-induced signal degradation. The findings indicate that while smart textile sensors offer unparalleled comfort and unobtrusiveness, their classification accuracy in field settings relies heavily on advanced signal preprocessing and context-aware algorithmic adaptation. By systematically evaluating these variables, this paper provides critical evidence and methodological frameworks necessary for bridging the divide between theoretical sensor capabilities and practical utility in professional sports training environments.