Map aided Low cost MEMS Based Pedestrian Navigation Applications
Date
2018-08-07
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Nowadays, indoor pedestrian location system has a big market requirement, more than 25000 developers in the world are focusing on this market. Various kinds of techniques, such as map based method, inertial navigation based method, Wi-Fi based positioning, Bluetooth technique, vison based technique, could be used to obtain the pedestrian’s position in indoor environment. However, each method has its own drawbacks, therefore, numerous methods have been proposed and integrated for pedestrian navigation by researchers. To date, the major challenges for an indoor pedestrian navigation system is to reduce the cost of the system, including the time-cost and the economic-cost, without decreasing the accuracy of the system. Considering that the MEMS sensor-based inertial sensors are low-cost, convenient, and self-independent, and the global IMU embedded smartphone adoption rate keeping increasing year by year. Therefore, inertial navigation based method is applied in this research to obtain a primary navigation solutions. However, the estimated solution of INS grows with time. Moreover, the accuracy of most smartphone embedded MEMS sensors is not as good as traditional inertial sensors. Specifically, MEMS gyro errors can cause heading errors and position errors; MEMS accelerometer error affect steps detection of Zero Velocity updates. Therefore, aiding constrains, such as Non-Holonomic constraints and Zero Velocity updates are used to correct the inertial navigation errors. In smart cities, the coverage rate of Wi-Fi keeps increasing, and the widespread distribution of Wi-Fi makes Wi-Fi suitable for indoor positioning. Take advantage of the pre-existing Wi-Fi access points, the Wi-Fi fingerprinting estimated positioning can be used to integrate with inertial navigation solutions. However, Wi-Fi signal is not accessible anytime and anywhere. So, Wi-Fi position is only an alternative aiding information for the proposed indoor position system. Map-based navigation is a traditional way to locate a pedestrian, and it is a low-cost method, which does not need any aiding infrastructures. Currently, most of the public building can provide indoor digital maps to users. Therefore, indoor map information can be added to inertial navigation system to improve the inertial navigation solutions.
Map Matching and Map Aiding algorithm are novelty integrated in this research to effectively use the free map information. Map Aiding is accommodating and does not need any assumptions about the path of the user. Map Matching is used for fixed trajectory part, such as corridors in buildings. Two methods are used in this research to complement each other, Map Matching will be added on the map-aided INS solution. A cascade connected Extended Kalman filter and Auxiliary particle filter integration algorithm which comprised a double-deck architecture is presented in this research to fuse all the above information. This structure can take advantage of merits of Extended Kalman filter and Auxiliary particle filter to estimate the navigation solution. The underlying Extended Kalman filter uses Zero Velocity and Non-Holonomic constraints as inputs of Extended Kalman filter to improve the preliminary INS navigation results.
To verify the proposed methods, experiments in different scenarios are conducted in different scenarios. The test results clearly indicate that the cascade structure algorithm can reduce the computational burden of the system. Also, through the proposed methodologies, integrating indoor map information, smartphone embedded sensors, and the pre-existing Wi-Fi, the indoor position system could provide continuous, accurate, and low-cost positions for pedestrians in indoor environments.
Description
Keywords
Citation
Yu, C. (2018). Map aided Low cost MEMS Based Pedestrian Navigation Applications (Doctoral thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/32796