A Reliable Indoor Pedestrian Positioning Method Considering the Topological Map Constraints

Abstract

Indoor inertial pedestrian navigation system has a wide range of applications in the fields of rescue and security. A mainstream approach is to estimate the pedestrian's motion by inertial navigation mechanism and to correct the accumulated errors of the inertial navigation system by using the human motion model information i.e. zero-velocity gait. To further improve the inertial positioning accuracy, indoor two-dimensional map information is introduced to compensate for accumulated errors of the inertial navigation system based on the Zero-velocity Update. This work introduced an inertial pedestrian navigation method using improved resampling-based particle filtering. The map topology constraints were included in the particle resampling process, and the effective particles selected by resampling were used to correct the global pose in real-time. The experimental results showed that our improved resampling algorithm enabled the particle filtering to better adapt to different topological areas, and effectively improved the accuracy of the pedestrian positioning method in complex indoor environments.