Research On Indoor Positioning System Based On Inertial Navigation And Multi-sensor Data Fusion

Services based on indoor location have become more important in today's society,such as anti-terrorism rescue,indoor navigation and other scenarios.However,indoor positioning relying on pure inertial navigation technology can lead to rapid failure of positioning due to its rapid error accumulation.Unable to meet the actual needs,the thesis is based on inertial navigation technology,combined with multi-sensor data fusion to reduce cumulative error,to achieve reliable and effective indoor positioning.The thesis describes the basic theory of inertial navigation system in detail,and introduces the functional principle and error model of the sensor components used for positioning.Based on the Kalman filter algorithm,the thesis summarizes the principle,derivation process and algorithm execution flow of Kalman filter,and introduces the extended Kalman filter suitable for the nonlinear system.Then the thesis introduces several kinds of fixed threshold zero-velocity detection methods for the zero-velocity update method used in the estimation of pedestrian trajectory,and proposes dynamic threshold zero-velocity detection for its drawbacks,which helps to achieve accurate zero-velocity detection under multi-step state,and then improve the accuracy of the trajectory estimation.In addition,for the trajectory glitch caused by inaccurate error estimation during the zero-velocity period,the thesis adopts a step-period smoothing algorithm to slow down the trajectory burr,making the reconstruction trajectory smoother and more realistic.Aiming at the problem of the height and heading long-term drift in the zero-velocity updating method,the thesis then proposes a multi-sensor data-assisted positioning,which improves the positioning system based on the air pressure data,geomagnetic data,linear walking and building direction identification.Based on the air pressure data to identify the action of the going upstairs and downstairs,identify the current floor,and then estimate the height error to reduce the height drift;use the geomagnetic data to solve the heading as the reference heading,combined with the magnetic malformation discriminant algorithm proposed in the thesis to filter out the disturbed geomagnetic data,and obtain an effective reference heading.The heading is used to estimate the current heading error of the system and reduce the heading drift.The pseudo-sensing data of straight line walking and building direction identification is used to help obtain the reference heading,and combined with the geomagnetic data to solve the long-term drift problem of the heading.Then,through multi-scenario experiments,the indoor positioning system of this thesis achieves effective high-precision indoor positioning,and the closed path threedimensional positioning error percentage reaches 0.39%.Finally,based on the Android platform and existing algorithms,the thesis designs and implements an indoor positioning monitoring platform,and gives the design and implementation of the functions of each part of the platform and its UI interface,realizing the mapping of pedestrian positioning in real time.