The Study Of Indoor Pedestrian Trajectory Estimation And Localization Algorithm Based On Inertial Sensor

In recent years,with the rapid development of Location Based Service(LBS)applications and the continual improvement of Micro-Electro-Mechanical System(MEMS)technology,pedestrian inertial navigation and trajectory tracking technology based on inertial sensors has got more and more researchers' attention and favor because of the advantages of its widely applicability,antiinterference ability and no user obstruction.However,the cumulative error problem caused by the continuous quadratic integral calculation seriously limits the positioning accuracy of inertial navigation algorithm.The correction process of the cumulative error becomes the key of algorithm performance improvement.In addition,with the upgrade of the intelligent terminal hardware,its built-in embedded microcontroller sensor unit makes it possible to be the implementation platform of inertial navigation and trajectory tracking algorithm.Nowadays,the study of indoor pedestrian trajectory tracking and localization algorithm on the emerging intelligent terminal has gradually become a hot topic.This thesis mainly realizes the improvement and innovation of pedestrian inertial navigation and trajectory tracking algorithm on wearable sensor node platform and intelligent terminal platform respectively.The specific research achievement are as follows:(1)On the wearable sensor node platform,a compartmental trajectory calculation algorithm is proposed with the cycle of pedestrian's gait to solve the continuous cumulative error problem in traditional strapdown inertial navigation which caused by the quadratic integral calculation.The proposed algorithm firstly implements state division and recognition on pedestrian's gait,then it calculates the rotation angle,horizontal displacement and heading angle of pedestrian's foot separately during different states in each step cycle.Finally,it updates the pedestrian's position coordinates and recovers the trajectory step by step.The experimental results show that the mean error of destination and total distance of the proposed algorithm is 0.74 m and 1.41 m,which has significantly improvement compared to the traditional strapdown inertial navigation algorithm.(2)A pedestrian dead reckoning algorithm based on activity recognition in multi-floor indoor environment is proposed on the intelligent terminal platform.The algorithm mainly includes two parts: Firstly,it combines the acceleration data with the received signal strength indication data to distinguish pedestrian's activity states.The experiment results show that the average misjudgment rate of activity states is 3.74%,which has significantly improvement in recognition accuracy rate compared to the traditional algorithm based on the mean and variance characteristics.Secondly,based on the recognition result,it realizes optimization and innovation on the step detection,step length estimation and heading estimation of the traditional PDR algorithm.It also realizes the expansion of the traditional PDR algorithm from two-dimensional plane to three dimensional space.The experimental results show that the ratio of step points with coordinate error within 1.5m in the estimated trajectory is higher than 85%,which indicates high accuracy of the proposed algorithm.(3)For the proposed pedestrian trajectory estimation algorithms in this thesis,a pedestrian inertial navigation system based on Shimmer sensor node and a pedestrian localization system in multi-floor indoor environment based on Android smartphone are set up for algorithm verification.The system implementation process is mainly composed of sensor node calibration,sensor data communication,smartphone data collection and location display on the map webpage.The verification results show that both system has superior performance in terms of accuracy and stability.