Tightly-Coupled Integration of Inertial and Magneto-Inductive Sensors for Large-Scale Indoor Localization
Changhao Chen,Andrew Markham,Niki Trigoni
2018-01-01
Abstract:Indoor localization and navigation problem has long received much attention from both academics and industries, due to its importance in many applications, such as location-based services, logistics, human-robot interaction and virtual reality. This problem used to be hard to solve, because some traditional radio-based localization methods such as GNSS, WiFi, UWB, suffer from serious attenuation, multipath effect and distortion in indoor areas, especially in the presence of moving people and a great amount of obstacles and concrete reinforcement walls. However, the appearance of magneto-inductive device (MI) using low-frequency, quasistatic magnetic fields, provides another possibility to handle localization problem in challenging areas by being less influenced by those distortion factors. Moreover, the widespread implementation and increasing accuracy of small-size MEMS inertial sensors (IMU) expects a promising future to give odometry without the need for external references in a period of time. In this paper, we exploit the sensor information from IMU and MI to achieve long-term large-scale indoor localization by making following contributions: we presented a highly-coupled integration method between IMU and MI sensors, which could mitigate high system error drifts of IMU, and low data-rate problem of MI; a novel Kalman filter framework, named Range-Constrained Kalman Filter (RCKF) was proposed to combine strapdown inertial navigation system (SINS) with motion characteristics and range measurement; a robust indoor localization system architecture was designed and tested in real experiments. On low-cost IMU and MI platform, the results from our proposed integration method show a significant improvement than that without range constraints. 90-percentile error of our localization systems is within 5-6 meter in our experiment with a total walking distance of 730 meter.