Pedestrian Navigation System Algorithm Research Based On Pedestrian Dead-reckoning

Driven by the rapid development of navigation technology and the widely application on mobile devices,the need for obtaining position information timely and accurately is more and more urgent.However,even mature GNSS positioning technology can only be used outdoor and under the condition that the signal would not be sheltered,so the navigation and positioning system based on pedestrian himself or herself is necessary where GNSS signal is absent like urban canyon,field,underground and indoor place.In this paper,we design a pedestrian navigation system(PNS)algorithm based on Pedestrian Dead Reckoning(PDR)to obtain pedestrian timely position.For the purpose of getting good results when apply the PDR algorithm on PNS,we analyze the way to install PNS device on pedestrian's body,especially comparing the motion data from different body parts of pedestrian and the data handling way with different algorithms.Finally we choose MTi-G-710 placing on the back waist and obtain motion data through MIMU.Considering Step Detection(SD),we deeply analyze pedestrian step characteristics and different phases in one step cycle,also the corresponding walking inertia data,then design three step detection algorithms including Peaking SD,Adaptive Peaking SD and Zero-Crossing SD.Experiments proves that SD accuracy has been improved through these algorithms.Considering Step Length Estimation(SLE),we deeply study the existing SLE algorithms and classify them,then choose the linear and nonlinear SLE model to design the SLE algorithm.Finally we propose three SLE algorithms including Linear SLE algorithm,Nonlinear SLE algorithm with Fixed Step Length Coefficient and Modified Nonlinear SLE algorithm.Experiments proves the SLE accuracy has been improved through these algorithms and already meets the walking distance accuracy requirement.When designing the overall PDR plan,we firstly choose Zero-Crossing and Nonlinear SLE algorithm as the SD and SLE combination algorithm which performs high stability and accuracy through the algorithm combination experiments,then compare the positioning accuracy using strap-down heading and compass heading in PNS PDR algorithm.Finally we propose that using magnetometer to get the original heading angle,then using the fusion heading of strap-down and compass heading in our PDR algorithm.