Development of a Sensor Suit for Gait Environment Detection Using Noncontact Sensors and Integrated Model

This study introduces a novel noncontact wearable sensor suit designed to classify various walking environments, integrating capacitive stretch sensors, and pneumatic mechanomyography (pMMG) sensors. Unlike traditional surface electromyography (sEMG), which is prone to noise and electrode displacement, the proposed sensor suit, designed in pants, ensures consistent signal acquisition without requiring direct skin contact. Experiments with eight participants in five walking conditions—flat ground, ramps (ascent/descent), and stairs (ascent/descent)—demonstrated superior performance using long- and short-term memory (LSTM) models. In intrasubject experiments, where individual models were trained and tested on data from a single subject, the sensor suit achieved a classification accuracy of 96.1%, outperforming the 94.0% accuracy of traditional sEMG-based systems. In intersubject experiments, where a general model was trained on multiple subjects and tested on unseen individuals, the suit achieved a classification accuracy of 91.0%, significantly surpassing the 76.4% accuracy of sEMG. Statistical analysis further validated the suit's ability to generalize between users and environments, with significant p-values (p <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$&lt; $</tex-math></inline-formula> 0.01) observed under multiple conditions. These findings emphasize the potential of the sensor suit for practical applications in wearable robotics and gait assist systems. Future research will focus on improving the system's adaptability and extending its capabilities beyond classifying walking trials to detecting pathological patterns, thereby significantly enhancing its clinical relevance.