This letter presents a shoe-integrated fall detection system that is based on seven time-of-flight (ToF) range sensors to identify falls on different terrains. The ToF range sensors, which measure the real-time distance between the sensors and the ground surface, can be used to detect simultaneously the variation of the gait cycle and the geometry of the terrain. Using the statistical features of the ToF range signals and an error-correcting output codes-based model, 17 experimental activities, executed by nine participants, are classified into three classes - activities of daily life, near falls, and falls. The experimental F1-scores for these three classes are 100, 98.17, and 99.07%, respectively. The arrangement of ToF range sensors is simplified to develop a compact shoe-integrated system architecture for fall detection.
All Science Journal Classification (ASJC) codes
- Electrical and Electronic Engineering