Research On Multiple Gait And Handheld Indoor Positioning System Step Length Estimation Algorithm Based On Dead Reckoning

With the popularity of smart phones and the increasing demand for indoor positioning,indoor positioning systems based on internal inertial sensors of mobile phones,also known as handheld indoor positioning systems,have become a hot topic of research.At present,the research of indoor positioning system focuses on a single motion mode.The accuracy of the hand-held indoor positioning system for multi-gait is not ideal,and there is a lack of a general indoor positioning algorithm under multiple attitude movements.This paper proposes a multi-gait handheld indoor positioning step length algorithm based on dead reckoning,which can accurately estimate the number of steps and steps in multiple motion poses.There is a lack of a reliable and superior performance general algorithm for step detection and step size estimation under different motion states.In this paper,a new zero-speed detection algorithm based on adaptive threshold is designed.According to the inertial data characteristics under different motion states,the motion state is discriminated and the corresponding algorithm is selected,so that the system can accurately detect pedestrians in various gait scenarios.Based on the K-means algorithm,the gait discrimination is first performed,and then the corresponding step detection algorithm is selected according to the gait discrimination result.Finally,the extended Kalman filter is triggered to perform zero-speed update.In the research of multi-step step estimation algorithm,this paper designs a hybrid model step size estimation algorithm based on dynamic model and inverted pendulum model.The weighted heavy factor is used to weight the two models,and the step size estimation model parameters is fitted through the training data.And finally estimate the adaptive step size results.The experimental results show that the average step error of the multi-step step detection algorithm in the normal walking mode is 0.80%,which is 4.2%lower than the average step error of the traditional step detection algorithm.In the slow walking mode,The average step error of the step detection algorithm in this paper reaches 0.40%,which is 3.6%lower than the average step error of the traditional algorithm.In the fast walking mode,the average error of the step detection algorithm proposed in this paper is reduced by 0.80%.Compared with the traditional step detection algorithm,it is reduced by 28%,which better solves the problem that the traditional step detection algorithm is not suitable for the fast walking mode.The experimental results show that the average positioning error of the system in normal walking mode is 0.53%,the average positioning error in slow walking mode is 2.07%,and the average positioning error in fast walking mode is 2.36%.The multi-step adaptive gait detection algorithm and hybrid model step size estimation algorithm proposed in this paper effectively improve the robustness and positioning accuracy of the system,and have certain practical application significance in disaster relief,medical care,shopping malls,etc.Has a good application prospects.