An Improved Bayesian Zero-Velocity Detection Algorithm for Pedestrian Navigation Based on MIMU

In the pedestrian navigation the zero-velocity detection is the prerequisite for zero velocity update, and its accuracy will greatly affect the navigation accuracy. Aiming at the problem that the Bayesian zero-velocity detection algorithm (SHOE) can't provide the accurate zero-velocity interval and that it may judge a zero-velocity interval as multiple zero-velocity intervals, this paper proposes an improved Bayesian zero-velocity detection algorithm with adaptive threshold(ISHOE). In the original algorithm, the threshold is reset when a zero-velocity interval begins. But in the improved algorithm, the threshold won't be reset untill the zero-velocity ends. Considering that the attenuation threshold will cause the acquired zero-velocity interval to last too long, this paper takes the threshold at the time when the non-zero-velocity interval enters the zero-speed interval as a fixed threshold, and collects the zero-velocity interval to obtain a more reasonable zero-velocity interval. The algorithm proposed in this paper can obtain zero-velocity intervals, and effectively avoids judging a zero-velocity interval as multiple zero-velocity intervals. In order to test the detection accuracy of the improved Bayesian zero-velocity detection algorithm with adaptive threshold, this paper compares and evaluates the new algorithm with the original algorithm (SHOE), acceleration variance detection algorithm (MV), acceleration amplitude detection algorithm (MAG) and angular velocity energy detection algorithm (ARE) under two different pace conditions. The results show that, without adjusting the threshold of the improved Bayesian detection algorithm with adaptive threshold, the algorithm can still achieve high accuracy and is suitable for applications in multiple sports modes.