A Research On Pedestrian Location Technology Based On Wearable IMU

Indoor positioning technology is increasingly vital.Wearable IMU can be seen every-where and IMU-based positioning is not only cheap,but also less affected by the external environment.The thesis mainly discusses a wearable-IMU-based positioning system:the algorithms,the hardware,and an Android-based mobile phone positioning app.First,the basic theory of IMU positioning is discussed:the main sensors.And then,thesis explains in detail the two main inertial navigation positioning algorithm frame-works:inertial navigation system and step-heading system.After analyzed,thesis will use the algorithm framework based on the step-heading system.Then,thesis elaborated on the algorithm.For step detection,thesis uses Butterworth digital low-pass filter and peak detection algorithm to achieve.For heading estimation,a nine-axis data fusion algorithm based on CF is used.For step length estimation,a step-length regression model based on BP neural network is discussed.Next,the wearable device based on a STM32 MCU is studied.First,it analyzes the functions of wearable devices,selects hardware modules and designs important circuits,and then adjusts the functionality of some algorithms.On the other hand,in view of the characteristics of hardware devices,thesis optimizes the algorithm functionality to bet-ter adapt to hardware devices,and implements a matching Android system-based mobile phone positioning App,so that the algorithms,hardware and the App work well.Finally,the experiments are conducted.Resultly,the step count of the step detection algorithm differs from the actual number by no more than 3 steps,and as the trajectory dis-tance increases,the accuracy rate will also be improved accordingly.In terms of heading estimation,the closing error obtained(1.09%)is lower than the old algorithm(5.14%).In terms of step length estimation,the mean absolute error of the paper's algorithm(3.29cm)is lower than that of a CNN algorithm(5.47cm).In the end,the thesis selected two tra-jectories of a straight line and a closed rectangle in the actual indoor environment,and conducted joint experiments with the hardware device and the App.In each experiment,the App can accurately draw the tester's trajectory in real time and the distance error does not exceed 4.3%,and the closing error does not exceed 2%.