Research Of Pedestrian Intertial Navigation And Positioning Based On Smartphone

With the development of Internet of Things technology,it is important to label the time and place of anything.Human needs for indoor positioning increasingly urgent.Common indoor positioning technology Bluetooth positioning technology,Wi-Fi positioning technology,radio frequency tag location,base station positioning,track estimation technology and so on.The advantages of trajectory estimation technology with or without external signal source,without a lot of infrastructure,low sensor costs,but there are also drawbacks such as sensor bias,random noise and uncertain relative position of pedestrian-carrying terminals.These problems limit the development of technology.Focusing on the above problems of sensor error and the relative position change of portable terminal,this paper focuses on the calculation of pedestrian trajectory under the four kinds of gestures of handheld pedestrian,reading,receiving call,rejection arm and running when walking normally.In the time domain,the sensor data is analyzed for error calibration.In the frequency domain,the Butterworth filter is used to filter out the high frequency noise.For the magnetic field distortion caused by external interference,Kalman filter is applied to the magnetic field data to reduce the influence of the magnetic field distortion on the heading angle influences.Finally,the correction of the step size calculation model under the four kinds of gestures is completed,and the joint pedometer method and the pedestrian course calculation are proposed.This paper completed the following three aspects of work.1.This paper analyzes the principle of the sensor and the sensor error.There are constant sensor error,random error.Magnetometer data are subject to environmental interference will cause the heading angle is not accurate or sudden change.In this paper,the correction of constant error calibration is done by pretreatment of data,and Kalman filter processing is applied to the magnetic field data.The gyroscope data is obtained by using the corrected three-axis magnetic field data and acceleration data to solve the pose settlement and error calibration Kalman fusion filtering can effectively restrain the sudden change of heading angle and make the heading angle reach the correct value quickly.2.In this paper,SVM is chosen as a classifier,which uses accelerometer data and gyroscope data to carry out multi-feature hand gesture recognition.In the four hand-held mobile phone posture feature analysis,the completion of the data feature extraction.The standard deviation of angular velocity,the value of co-acceleration,the data of Y-axis and Z-axis are used as the recognition algorithm of eigenvector.The correct recognition rate is 93%.3.The pedestrian steps,the number of steps and the heading angle are calculated under the four kinds of gestures.It is proposed to analyze the acceleration data by determining the main axis and the forward axis so as to avoid the multi-step-counting problem in the mode of not stepping,shaking the mobile phone up and down,shaking left and right,performing pendulum motion in static state and the like without real stride.The accuracy of step-counting can reach 98%,which is more accurate than the acceleration-calculating step-counting algorithm.The method of correcting the peak-valley value calculation step,the average error of the step is about 2 centimeters,and uses the fusion filter algorithm to obtain the pedestrian's heading.