Research On Indoor Location Algorithm Based On Pedestrian Dead Reckoning

With the rapid development of mobile communication technology and Internet technology,the demand for indoor services based on location has gradually increased,and the requirements for indoor positioning technology are becoming more and more demanding.Pedestrian Dead Reckoning(PDR),as an indoor positioning technology which only requires intelligent terminal equipment to locate pedestrians,has attracted more and more attention from researchers due to its convenience and scalability.However,due to low-cost sensors and complex human behavior,traditional PDR technology still faces many defects.This paper has conducted in-depth research on the two problems that traditional PDR technology cannot adapt to different walking patterns of pedestrians and has the cumulative error.The main work of this paper is as follows:1)Aiming at the problem that PDR cannot adapt to different walking patterns,a walking pattern adaptive PDR algorithm,WPA-PDR,is proposed.First,WPA-PDR provides a walking pattern adaptive step detection algorithm,which combines the zero-crossing detection method and the peak prediction method based on Dynamic Time Warping(DTW)to accurately detect the start moment and end moment of each step of a pedestrian;Then,WPA-PDR extracts relevant features based on the results of step detection,and recognizes different walking patterns of pedestrians through a random forest classification;Finally,WPA-PDR provides walking pattern adaptive stride length estimation,which selects the corresponding stride length parameter according to the recognition result of the walking pattern to improve the accuracy of stride length estimation.The simulation results show that the average step detection accuracy of WPA-PDR in the three walking patterns(normal walk,fast walk,and march in place)is 97.9%,and the average recognition rate of the three walking patterns is 99.4%.In addition,we evaluate the comprehensive performance of WPA-PDR by estimating the walking distance of the actual scene in this paper.The results show that the average walking distance error of WPA-PDR is 2.9%,which is much smaller than the comparison algorithm.2)Aiming at the problem of cumulative error and low precision of sensor output data in PDR,a map matching algorithm based on Hidden Markov Model and Particle Filtering,HMM-PF-MM,is proposed.HMM-PF-MM consists of two modules: HMM-based activity sequence matching and PF-based PDR position correction.The former regards some key indoor locations(stairs,elevators,turns,etc.)as indoor landmarks,which can locate pedestrians to specific indoor locations when they pass the indoor landmarks,thus solving the problem of accumulated errors.The latter is used to correct the PDR position estimation result on the path between the special points in the room.The latter takes the output of PDR as the state of particle filter,and the stride length and heading as the observed value,then corrects the estimation result of PDR position on the path between indoor landmarks,so as to reduce the influence of the output noise of inertial sensor on the positioning accuracy of PDR.The requirements of HMM-PF-MM for indoor map are simple.Adopting the simple pseudo-indoor map constructed in this paper can obtain a good positioning accuracy.The simulation results show that the detection accuracy of HMM-PF-MM is 99.3%,and the average positioning error of HMM-MM is 1.3m,which is 50.1% higher than that of the comparison algorithm,while the average positioning error of the complete HMM-PF-MM is 1 m,which is 57.9% higher than that of the comparison algorithm.HMM-PF-MM not only solves the cumulative error problem of PDR,but also improves the positioning accuracy of PDR.