Algorithm Research On Personal Navigation System Based On Particle Filter Aided By Indoor Map

With the development of science and technology,as well as the national strong support for smart cities,the big data and Internet of Things get rapid development.Location-based services are able to provide a higher quality of life for the masses and better realize automatic and intelligent construction for all walks of life.As MEMS inertial device becomes more and more miniaturized and integrated,the personal navigation system has been widely concerned.However,because of the complexity of the indoor environment,the kind of indoor positioning method can not be widely used as GPS.Taking MTI-G-700 produced by Dutch company as the main research object,we study the personal navigation system based on particle filter aided by indoor map.In order to overcome the positioning error caused by inertial device drift for a long period of time,thresholds algorithm of zero velocity detection is proposed,which combines the angular velocity on y-axis,the magnitude of the gyro,the magnitude of acceleration on z-axis with the magnitude of acceleration.To eliminate the error detections,we set a time threshold according to the duration time of foot contacting the ground during one step when pedestrian has a normal walking.When talking about the fire rescue or some of the unexpected events in the tall buildings,vertical accuracy may seem to be as important as horizontal accuracy,while the height calculated by pure inertial navigation algorithm is erratic.At the same time,there has been little discussion about height error correction.Buildings based on prior knowledge,that is,the height of each floor are fixed,we correction the height of the strapdown inertial navigation solution.The results show that this algorithm has strong robustness.In order to overcome the cumulative error of inertial navigation system,we build particle filter state equation and measurement equation based on dead reckoning movement model.In this state equation,course angle and horizontal coordinates are viewed as state variables.In the measurement equation,step length and course angle variation are viewed as observations.At the same time,the indoor map is taken account into the particle filter in order to detect the particle crossed the wall and update particle weights.Meanwhile,in order to solve the question that inaccurate heading angle error model in particle filter measurement equation model particles lead to the severe divergence,we propose the virtual sign matching algorithm and set the signposts at the turning.When particles are detected to diverge at the corner,we correct the current position to the corresponding signpost.This method is equivalent to initialize the initial position of inertial navigation system.It can eliminate the accumulated error at location level.Finally,the MEMS inertial sensor MTI-G-700 was bundled on the upper surface and tested on the campus building.The validity of the navigation algorithm was verified by a large number of experiments.