MEMS Sensor Array Calibration And Research On Key Algorithm Of Foot-mounted Pedestrian Navigation

Indoor and outdoor seamless positioning has become a research hotspot of current Location Based Services.Traditional satellite navigation technology cannot be used indoors because the signals will be severely weakened when passing through buildings.On the other hand,in many special scenarios,such as mines,underground surveys,fire rescues,etc.,where it is difficult to provide infrastructure support,how to construct a passive navigation system with high stability and high accuracy is the current research difficulty.As an autonomous navigation and positioning technology,the Inertial Navigation System(INS)does not need to pre-build hardware facilities,and there is no problem of signal reception.It can obtain the position,velocity and attitude information of the carrier in real time.The most primitive inertial navigation system uses highprecision inertial devices,which are often expensive and bulky,and are difficult to use in people's daily lives.However,with the development of Micro-Electro-Mechanical System(MEMS)manufacturing technology,MEMS-based Inertial Measurement Units are getting lower and lower in cost,smaller in size,and lower in power consumption.It has been widely used in the field of consumer electronics,providing a reliable hardware foundation for pedestrian inertial navigation.However,due to the limited precision of consumer-grade MEMS inertial sensors,the output data contains a lot of noise,which will cause large errors when performing strap-down inertial navigation.In this paper,the Kalman filter model combining Zero-velocity Update(ZUPT)and INS is designed by using the IMU array composed of consumer-grade MEMS inertial sensors.The ZUPT is used to correct the error of the strap-down inertial navigation mechanical layout,and suppress the divergence of integral error,so finally higherprecision pedestrian navigation is achieved.The main contents include:(1)The MEMS sensor array is calibrated.In this paper,the accelerometer error of a single MEMS IMU in the array is calibrated by multi-position method.The bias,scale factor error and sensitivity axis non-orthogonality of the accelerometer in a single IMU,and also the installation misalignment error between the IMUs in the array,are estimated.The experimental test verifies it is necessary for the calibration of low-cost MEMS sensor arrays;(2)The paper introduces the zero-speed detection problem.This paper tested four commonly used zero-velocity detection methods and analyzed the relationship between them.The experimental test analysis compares the advantages and disadvantages of the four detection methods,and finally determines the conditions included both acceleration and angular velocity to detect the zero-velocity;(3)The differential equation of inertial navigation is simplified.The Kalman filter model fused INS and ZUPT based on position error,velocity error and attitude error is designed.The zero-velocity detection algorithm detects the zero-speed phase and at the stationary state,the velocity estimation value is used as the observation to perform the system state error estimation,and the estimated error will be used to correct the parameter of INS navigation solution.At the same time,the observability analysis of the above Kalman filter model is carried out.It is found that the algorithm can suppress the error divergence of velocity,roll angle and pitch angle,and improve the position error drift,but there is no improvement for the heading angle error.The experiment verifies the observability analysis of the filtering model.The test shows that the estimated path is roughly consistent with the actual trajectory,and the one-minute plane error is about 1m.