Research on Height Constraint Algorithm Based on Hidden Markov Model

Indoor pedestrian navigation has recently become an important field of interest in satellite-denied scenarios. Zero Velocity Update prevents an accumulated error growth caused by the noise of MEMS-IMU. However, the height error is still an issue and accumulates over time. We propose a height constraint algorithm based on Hidden Markov Model and Recursive Viterbi algorithm without any other sensors besides shoe-mounted IMU. The presented algorithm addresses the issue of setting the height reference threshold because of the unfixed height change value of pedestrian when climbing stairs. And we propose a simple method to fetch height state without complex gait phase detection. For the assessment of the performance of the proposed height constraint, we compare the height error estimated with and without the proposed algorithm. The experimental results show that the height constraint algorithm can reduce height error within 0.1 meters with preferable stability and robustness at the same time.