Indoor Localization Using Mind Evolutionary Algorithm-Based Geomagnetic Positioning and Smartphone IMU Sensors

Abstract

With the pervasiveness and ubiquitous distribution of the magnetic field in indoor environments, indoor localization using magnetic positioning (MP) has attracted considerable attention. This work concentrates on the MP and pedestrian dead reckoning (PDR) method, and constructs a fusion system for smartphones using MP and PDR based on the extended Kalman filter (EKF). The mind evolutionary algorithm (MEA) is introduced to search for the optimal magnetic position based on a heuristic searching strategy, which uses the similartaxis and dissimilation for the evolutionary operation. In the PDR module, the acceleration characteristics of different walking patterns are analyzed and the corresponding features are extracted. The enhanced genetic algorithm-based extreme learning machine (EGA-ELM) is adopted to train these features and address the gait recognition problem of different walking patterns. Finally, to obtain a lightweight and high-precision fusion method, MEA-based MP is integrated with PDR based on the EKF. Extensive experiments are conducted to evaluate the proposed methods. The testing results showed that MEA-based MP can obtain a location error within 2.3 m and steps can be recognized with a mean accuracy of 95% when different users participate in testing. The positioning results after fusion with PDR reveal that the mean location error and root-mean-square error (RMSE) are 1.25 m and 1.53 m respectively, which outperforms the MP, PDR, MP and PDR fusion methods using improved particle filter (IPF) and genetic particle filter (GPF).