Step Detection Robust against the Dynamics of Smartphones

Sensors

Hwan-hee Lee²  Suji Choi¹  Myeong-jin Lee² 

[1] Mechatronics R&D Center, Samsung Electronics, 1-1 Samsungjeonja-ro, Hwaseong, Gyeonggi 445-330, Korea; E-Mail:;School of Electronics and Information Engineering, Korea Aerospace University, 76 Hanggongdaehang-ro Deogyang-gu, Goyang, Gyeonggi 412-791, Korea; E-Mail:

关键词: step detection;    accelerometer;    step average;    adaptive magnitude threshold;    adaptive temporal threshold;    peak-valley relationship;    step mode;    device pose;   

DOI  :  10.3390/s151027230

来源: mdpi

【 摘 要 】

A novel algorithm is proposed for robust step detection irrespective of step mode and device pose in smartphone usage environments. The dynamics of smartphones are decoupled into a peak-valley relationship with adaptive magnitude and temporal thresholds. For extracted peaks and valleys in the magnitude of acceleration, a step is defined as consisting of a peak and its adjacent valley. Adaptive magnitude thresholds consisting of step average and step deviation are applied to suppress pseudo peaks or valleys that mostly occur during the transition among step modes or device poses. Adaptive temporal thresholds are applied to time intervals between peaks or valleys to consider the time-varying pace of human walking or running for the correct selection of peaks or valleys. From the experimental results, it can be seen that the proposed step detection algorithm shows more than 98.6% average accuracy for any combination of step mode and device pose and outperforms state-of-the-art algorithms.

【 授权许可】

CC BY   
© 2015 by the authors; licensee MDPI, Basel, Switzerland.