Visual-aided Indoor Mapping And Pedestrian Navigation Research

The lacking of indoor maps,location fingerprints and indoor landmark is the essen-tial bottleneck for current indoor navigation systems.Nowadays,with the popularity of smartphone and its powerful performance,we can calculate the location information and infer human' s activity from the inertial sensors of smartphone.However,lacking of the direct observation from external environment,will cause the accumulated loca-tion errors and the human activity perception errors.The visual information collected from smartphone as one of the observation directly to external environment,which can be used as the correction for the accumulated location errors and the human activity perception errors.In this thesis,we take the inertial data and the video data collected by pedestrians during indoor activities as input.Firstly,detect the spatial structure of indoor environment and then restore the walking trajectory of pedestrian,which realize the work of calibrating the fingerprints automatically.For the problem of pedestrian getting lost in navigation process,we proposed a route planning algorithm based on landmark visibility,and a structure landmark based location verify algorithm.The main contents of this thesis are as follows:1)This study proposes a smartphone-based indoor navigation map construction algorithm.Indoor map is the basic data for indoor navigation.Constructing indoor map need a large amounts of human-labor and time.Traditional methods for constructing indoor map depend on manual processing and professional equipment.The state-of-art indoor map construction method uses the smartphone,which can realize automatically constructing indoor map without human intervention.However,the results of these methods lack the structure information and landmark information,which is no help with pedestrians understanding and using.With the aid of the collected visual data,we can restore the walking trajectory precisely.The Bayes classification method was used to classify the structure type of navigation node in indoor environment.The text landmark can be extracted from the collecting video data by OCR(Optical Character Recognition)method.Using all above methods,we can construct the pedestrian indoor navigation map.The experimental results showed that this method could achieve satisfactory performance,which provide the basis for efficient establishment of indoor map.2)This study proposes a Wi-Fi fingerprints automatic calibration based indoor localization method.Traditional fingerprinting-based indoor localization technologies are the extensive and time-consuming that is required to construct and update the radio map.Take advantage of the walking trajectory for calibrating fingerprints and constructing radio map is one of the effect way to solve this problem.However,most methods need the initial location of trajectory and the indoor map as a prior,which decrease its robustness and practicableness.In this thesis,we propose a system that can be used to collect fingerprints and construct the radio map by exploiting the tra-jectories of smartphone users.By integrating multisource data from the smartphone sensors(e.g.,camera,accelerometer,and gyroscope),this system can accurately recon-struct the geometry of trajectories.An algorithm is also proposed to spatially estimate trajectories into the indoor reference coordinate system and construct the radio map,without user collaboration or prior knowledge(e.g.,indoor floorplans or initial locations of trajectories).The experimental results show that the average location error of the fingerprints is 0.67 m.A weighted k-nearest neighbor method(without any optimiza-tion)is also used to evaluate the constructed radio map.The average localization error is about 3.2 m,indicating that the quality of the constructed radio map is at the same level as those constructed by site surveying.3)This study proposes a route planning algorithm based on landmark visibility and a structure landmark based location verify algorithm.Considering the extent and visibility of pedestrians'use of landmark in indoor environment,a multi-object route planning algorithm was proposed to obtain the optimize route with shortest distance and highest landmark utilization.Defining the spatial structure as structure landmark,on the basis of automatic identification of structure landmark,an improved Hidden Markov Model was implemented for online localization and offline localization.Without any help of external auxiliary equipment,this method can achieve online localization performance more than 80%and offline localization performance more than 70%.It is helpful to reduce the risk of get lost and pedestrian navigation cognitive burden.