Key Technology Research Of Pedestrian Navigation System Based On Self-Contained Sensors

The personal navigation system(PNS)is widely used in areas such as individual solider combat,emergency rescue and personnel monitoring as it can provide the individual solider indoor and outdoor with real-time and accurate location information.This paper chose an autonomous PNS in the non-satellite signal environment as the research object,and then a PNS prototype was built on the basis of proposing the overall PNS design scheme.Supported by this platform,the key technologies including the zero velocity interval(ZVI)detection technology,the initial alignment technology,and the PNS algorithm based on the body features/environment characteristics constraints of the PNS are studied.The research contents are as follows:Aiming at the problem of the ZVI detection in human gait,a ZVI detection algorithm based on multi-constrained conditions is proposed.The algorithm can extract the ZVI in human gait by setting constraints for the sensors output,which realizes the accurate detection of ZVI while the pedestrian walks steadily.But put it into consideration that the fixed threshold may cuse false or missing detection when the pedestrian walks at different speeds,an adaptive ZVI detection algorithm based on the SPWVD-RMFI is put forward.By extracting the gait frequency of the pedestrian and carrying out the calibration experiment for ZVI detection threshold at different gait frequencies,the function relationships between the detection thresholds of ZVI and the human gait frequencies are established to realize the adaptive detection of ZVI while people walk at different speeds.For the problem of MEMS gyro output signal with large random drift noise,an optimized wavelet threshold method used to reduce the noise is put forward.The method can effectively overcome the disadvantages of traditional threshold de-noising method by constructing a new wavelet threshold function to calculate the wavelet coefficient.The optimized method can effectively eliminate the random noise in the MEMS gyro signal and the signal is reconstructed with higher SNR and smaller MSE,which indicates that the optimized method has better performance in de-noising.To solve the problem that the low-cost inertial sensors in PNS fail to do self-alignment,this paper studies the initial alignment method of PNS assisted by magnetic sensors under the stationary base,and then deduces the error state equation as well as the observation equation of the system.Finally,through the feedback correction,the initial alignment of the PNS is achieved.Since the ferromagnetic interference in the real environment leads to the measurement errors of the magnetic sensors,this paper analyze and build the error model for the background magnetic field of PNS.Meanwhile,a field error calibration scheme is designed to compensate the background magnetic field error,which further provides accurate magnetic measurement data for the initial alignment and navigation calculation.Without adding external sensors,effective external observations are extracted by making full use of human movement features while walking and the surrounding environment characteristics.On this basis,a PNS algorithm based on the body features/environment characteristics constraints is put forward in this paper,in which the optimized extraction algorithm for the heading angle and the height information in ZVI are designed.Moreover,the error state vectors of the system are estimated and the navigation parameter errors are corrected by using the Kalman filter.The results of testing and contrast experiments indicate that the PNS designed in this paper has higher positioning performance.