Study Of Indoor Pedestrian Positioning Technique Based On Inertial Navigation

In the indoor positioning technology based on inertial navigation,the data of inertial measurement unit of and integral calculation will bring some error to the system,and the error accumulated over time,resulting in serious false location.The zero velocity update algorithm can be used to estimate the error and error correction in the static range,so the performance of the stance detection algorithm determines the precision and accuracy of the inertial navigation system.The traditional detection algorithms such as: Amplitude detection,acceleration Moving Variance(MV)and Angular Rate Energy(ARE)detection are useful for slow motion detection results,but for the larger speed range,the detection accuracy is poor.Therefore this paper proposes the stacne detection algorithm based on Hidden Markov model which use the information of the speed of the body.The speed trend of different motion state is basically the same,and there is only a little difference in the duration of the velocity and the rest interval.In a whole walk cycle,the speed first accelerates and then slows down,and finally the velocity tends to zero.Therefore,the complete walk cycle can be divided into three states: state 1(acceleration),2(deceleration),3(stance).In inertial based self-contained pedestrian positioning systems,because the drifts of the gyroscopes grow with time,it relies on the earth magnetic field to suppress the heading errors.However,the earth magnetic field suffers from severe interference in indoor scenarios,and the magnetometer itself has measurement errors,the above reasons have dramatically limited the performance of the magnetometer-aided heading error calibration.This paper proposes a magnetic-aided heading error calibration approach.On this basis,a novel Quasi-Static magnetic field(QSF)detection approach is proposed to extract the usable magnetic information fed into Zero velocity Update-aided Extended Kalman Filter(ZUPT-aided EKF)algorithm to conduct the heading calibration.The experiment results show that the position error is less than 0.5% of the total walking distance.