A zero-velocity detection method with transformation on generalized likelihood ratio statistical curve. (October 2018)

• Record Type: Journal Article
• Authors: Liu, Xixiang; Wang, Songbing; Zhang, Tongwei; Huang, Rong; Wang, Qiming
• Is Part Of: Measurement. Volume 127(2018)
• Page Start: 463
• Page End: 471
• Publication Date: 2018-10
• Subjects: ZUPT -- Zero-velocity detection -- Generalized likelihood ratio statistical curve -- Geometric transformation; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2018.05.113
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: A new zero-velocity detection method for pedestrian navigation systems is given. The relationship between the pedestrian gait and the GLRS curve is given. The valley bottoms of the GLRS curve are obtained by geometric transformation. The slope of the GLRS curve is used to distinguish different type valley bottoms. Abstract: MEMS-SINS (Micro-Electro-Mechanical System – Strapdown Inertial navigation System) is always used for pedestrian navigation. But MEMS-SINS cannot fulfill navigation task independently for its low precision gyros and accelerometers, thus ZUPT (Zero-Velocity-Updating) should be introduced. ZUPT uses zeros-velocity when the foot contacting the ground as observation information aiding MEMS-SINS. Zero-velocity detection is a key and difficult technology. Currently, zero-velocity detection methods use the output of accelerometers or (and) gyros from MEMS-IMU (Inertial Measurement Unit) as measurement and compare those measurement with given threshold to judge zero-velocity constraints occur or not. Due to the diversity of pedestrian gaits and low signal-to-noise ratio of MEMS-IMU data, zero-velocity detection methods based on threshold detection own inherent shortcomings such as poor environmental adaptability and inconvenient usage. To overcome these, in this paper ideas using threshold detection are discarded, characteristics of pedestrians gaits is detailedly analyzed and two characters – zero-velocity interval must be located at the valley bottom of the generalized likelihood ratio statistical curves but the bottom of the curve is not necessarily zero-velocity interval – are concluded, thus a novel detecting method is designed. With proposed method, transformations including translation and symmetrical flip on generalized likelihood ratio statistical curves of MEMS-IMU are data are used to detect true zero-velocity interval and discard interfering curve bottoms. Theoretical analysis and experimental results show that the proposed method can detect zero-velocity occurring effectively without any given threshold and detecting rate on different groups of people and different velocity is 100%.