Interval heading differential correction with motion recognition to enhance the foot-mounted INS performance

The zero velocity update (ZUPT) and heading correction serve as two critical modules in foot-mounted inertial navigation systems (FM-INS). In FM-INS that rely solely on ZUPT, the heading angle and the gyroscope bias are unobservable. The unobservability leads to the accumulation of heading errors, which degrades the accuracy of position estimation. To reduce the divergence of heading angle error, we propose a novel swing interval heading differential correction method (SIHDC) within the classical framework of the extended Kalman filter and ZUPT (INS-EKF-ZUPT, IEZ) in FM-INS. This method applies corrections during the swing interval from foot lift to footfall. When the heading angle difference between adjacent moments falls below a preset threshold, the heading difference measurement is constructed to improve the state estimates. Compared to traditional heading angle correction methods, heading angles and gyroscope biases can be estimated more effectively during non-zero velocity intervals with the SIHDC method. Experiments were conducted to verify the feasibility of the proposed algorithm, and comparisons were made with the traditional IEZ framework and the IEZ with zero integrated heading rate (ZIHR). Experimental results indicate that, for a total travel distance of 80 m and 425 m in two experiments, the trajectory closure errors of our proposed method are only 0.54% and 0.66%, respectively. The positioning accuracy of the experiments, compared to the classical IEZ with the ZIHR method, is improved by 92.14% and 78.77%, respectively.