| With the rapid development of wireless sensor network technology and smart mobile terminals,location-based services with positioning technology as the core are increasingly receiving widespread attention.Global Navigation Satellite System has already had mature applications outdoors.However,indoors,due to the deployment cost of positioning equipment,and the complexity of the indoor environment,the widespread availability of indoor location services has been restricted.The indoor positioning method based on the Pedestrian Dead Reckoning algorithm and the portable intelligent mobile terminal is independent of external environment at all,and there is no need to arrange positioning equipment in the positioning area in advance.It is a low-cost and convenient indoor positioning method.This paper focuses on the three key technologies of step detection,step length estimation and heading estimation in PDR algorithm to start research,proposes a step length estimation model of coefficient self-determined based on peak-valley step detection.The collected yaw angle is optimized according to the characteristics of modern building structure and the gait characteristics of pedestrians during walking.Through experimental test and algorithm performance comparisons,the advantages and feasibility of the solution in this paper are verified.The main work and innovations of the paper are as follows:(1)By analyzing the accelerometer data collected by pedestrians holding i OS devices under different step length and step frequency,a peak-valley step detection method is proposed.In consideration of steps over-counting in traditional peak detection,this paper proposes four features of acceleration data,including dynamic threshold,minimum sampling interval,neighboring difference,and peak-valley sequence,which can eliminate abnormal peaks and valleys in motion data and improve the accuracy of step detection.The experimental results show that walking 500 step freely in the same environment,the average detection accuracy of this method is improved by 1.17%,compared with the peak detection,and the method of peak-valley step detection is suitable for different walking speeds.(2)Based on fuzzy logic algorithm,a step length estimation model of coefficient self-determined is proposed.There is one or more personality parameter in the existing step length estimation models,which needs to be manually adjusted before using by different users.This paper collects 490 sets of data from 10 volunteers walking under different step length and step frequency,and designs a fuzzy controller to output adaptively the parameter in the Weinberg step length estimation model.Through experimental verification and comparison with other step length estimation models,it is proved that the method proposed in this paper is more universal and suitable for different shapes of people walking at various step frequency and step length.The average estimation error of the test volunteers walking100 m in a straight line is less than 4m(3)Based on the idea of heuristic heading deviation reduction,the detected yaw angle is corrected according to the rectangular feature of the building structure and the gait characteristic of the inverted pendulum.In view of the rectangular structural characteristics of modern building structure and building internal layout,straight-line walking of pedestrians is normal.In addition,when pedestrians walking in a straight line,the yaw angle swings within a certain range.Therefore,this paper proposes to judge whether the pedestrian is walking in a straight line according to a swing threshold of yaw angle,and then performs an average filtering process on the data of straight walking.Four special turning angles of 0°,90°,-90°,180°/-180° are corrected.Combining the above step detection and adaptive step length estimation model,the walking trajectory estimated by using the corrected heading angle is closer to the preset route,which verify the feasibility of the method. |
PDF Full Text Request