Title: Behaviors Recognition and Step Detection for Pedestrian Navigation via a Foot-mounted Inertial Measurement Unit
Author(s): Zebo Zhou, Shanhui Mo, Shuang Du, Jianghui Geng
Published in: Proceedings of the 30th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2017)
September 25 - 29, 2017
Oregon Convention Center
Portland, Oregon
Pages: 357 - 367
Cite this article: Zhou, Zebo, Mo, Shanhui, Du, Shuang, Geng, Jianghui, "Behaviors Recognition and Step Detection for Pedestrian Navigation via a Foot-mounted Inertial Measurement Unit," Proceedings of the 30th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2017), Portland, Oregon, September 2017, pp. 357-367.
Full Paper: ION Members/Non-Members: 1 Download Credit
Sign In
Abstract: The inertial measurements imply the characteristics of pedestrian gait cycle and it can be used as a step indicator for the dead reckoning navigation. The parameters in conventional step detection methods need to be specified for different persons. However, it cannot well adapt to the general pedestrian applications. In this contribution, a foot-mounted pedestrian navigation system prototype is designed and developed with the emphasis on the behavior recognition and step detection. The hardware platform consists of four modules, i.e. a STM32 micro-processer, an IMU, a Bluetooth and a battery. In order to accurately and reliably count steps, we propose an adaptive time- and frequency-domain joint step detection method by utilizing the means of activity classification. The proposed method includes four stages: finite impulse response (FIR) filtering, adaptive window length determination, extrema points searching and cross-points identification. It adaptively determines both step frequency and threshold in a flexible window thus improving the step counting success rate. Finally, the real experiments are carried out to evaluate the performance of our developed foot-mounted pedestrian navigation system prototype. The result indicates that the pedestrian navigation platform performs well in real time and significantly improves the step detection accuracy compared with the conventional methods.