Research On Indoor Human Navigation Technology Based On Muti-Sensor

With the development technology,people has the grate requires of position and navigation,especially in indoor environment.Because the singles will be blocked by obstacle indoor,we can’t use GPS which depends on satellite technology.So indoor position and navigation technology have gradually become the focus of current research,which are based on various wireless network or inertial sensors.In this paper,we study the problem of trajectory estimation based on inertial sensor,and implement the system based on STM32.Generally speaking,the MEMS sensor has some disadvantages,such as low precision,random error,and so on.To solve these shortcomings,this paper established the error model,and eliminates the influence of zero drift error.Magnetometer is more easily to be influenced in indoor environment.This paper realized a fast field error calibration compensation method of three-axis magnetic sensor in STM32 embedded systems.Attitude algorithm and pedometer are the most important parts in the system.In this paper,Quaternion is the tool to update the attitude.Firstly,the initial attitude angle can be calculated through the information of accelerometers and magnetometers.Then we can fuse sensor value to update the attitude angle by the extended kalman filter in shape of Quaternion.As to step estimation,a detecting algorithm based on acrossing the middle threshold is used to analyze the vertical acceleration value and reckon step information.In the trajectory estimation,Heuristic Drift Elimination(HDE)algorithm has some disadvantages such as the inaccurate heading angle and poor robustness,the AHDE algorithm was proposed which implementing Heuristic Drift Elimination based on heading angle with Extended Kalman Filter.The heading angle is renewed by Quaternion which is achieved from the integration of gyroscope data and accelerometer data.After updating the heading angle with the algorithm of Angle Heuristic Drift Elimination,the angle can be used to reckon the pedestrian trajectory combined with step number and stride length.The experimental results show that,when walking in typical structured indoor environments,the error of the proposed algorithm is less than 2 meters in 250 meters distance,and the error of the HDE algorithm is 4 meters roughly.On the other hand,when the sampling data rate of the sensors is 100 Hz,the convergence range of feedback coefficient of the AHDE system can be extended to [0.005,0.23],however the range of the HDE system is [0.001,0.028].That is to say the AHDE system is more robust than the HDE system.