관성 센서 기반 보행자 식별에 관한 연구

초록

This paper proposes a pedestrian identification method using a multi-task learning approach on IMU-based acceleration data. Gait data were collected from six adult male subjects under both normal and muscle-fatigued conditions. The proposed method achieved high accuracies of 100% for the personal identification task and 96.65% for the gait-state classification task. Future work will be conducted with a larger sample size to validate the generalizability of these results.