Stance Phase Detection for ZUPT-Aided INS Using Knee-Mounted IMU in Crawling Scenarios

Abstract

This letter presents a study on the navigational accuracy of crawling using heel and knee-mounted inertial measurement units (IMU). The navigation method is based on a stance phase detection in combination with the zero-velocity UPdaTe algorithm and strapdown inertial navigation systems (INS). A parameter search considering window size, detector type, and statistical threshold was performed to determine the parameters that minimized the circular error probable (CEP). Under the experimental conditions, the knee-mounted IMU position rendered improved navigation accuracy compared with the heel-mounted IMU. Based on the IMU used in this experiment, accelerometer-based zero-velocity detectors appeared to have improved performance for the knee-mounted IMU compared with the angular rate energy and stance phase hypothesis detectors. Under the optimal parameters, the CEP/corresponding root-mean-square error (RMSE) for heel-mounted and knee-mounted IMUs was 0.95/1.05 m and 0.22/0.43 m, respectively, for ten crawling trials over a 42.6 m straight trajectory marking a 77.3% CEP and 59.1% RMSE improvement when using a knee-mounted versus a heel-mounted IMU.