Research On Dual Foot-Mounted IMU Pedestrian Navigation With Equality Constraint

In recent years,pedestrian navigation system based on foot-mounted inertial sensor has received extensive attention and research.Because the errors of inertial sensor will accumulate over time,ZUPT technology is usually used to reduce the errors.When a pedestrian's foot is stationary on the ground,the foot speed is considered to be zero.The zero velocity is fused to the extended Kalman Filter(EKF).However,in the inertial pedestrian navigation system aided ZUPT,the heading error still accumulates over time,resulting in the position deviating from the true trajectory.In this thesis,a dual IMU pedestrian navigation system based on equality constraints is proposed to reduce the heading error of ZUPT aided inertial pedestrian navigation system.Dual-foot system based on two single foot ZUPT aided navigation systems,installed a pair of ultrasonic sensors to measure the distance between feet and,in turn,according to the position between feet to establish a equality constraint.Two independent ZUPT subsystems have fused together,and it provides a new observation for the correct heading error.Furthermore,the earth ellipsoid constraint model is established to suppress the height error of the dual foot system.Under the condition that the height variation of two adjacent stance phase feet does not exceed a certain threshold,the foot position is constrained to the same earth ellipsoid,and the constraint conditions are integrated into EKF to suppress the height error.In this thesis,we analyze the global observability of ZUPT and dual IMU system based on equality constraints.Previous linear observability analysis showed that the ZUPT measurements gave the EKF the correct velocity,accelerometer bias,pitch,roll,and pitch and roll direction gyroscopic bias estimates.However,other important conditions such as position,heading,and gyroscopic bias in the direction of gravity cannot be observed from the ZUPT.The global observability analysis shows that the bias of the gyroscope gravity direction in ZUPT algorithm is weakly observable.In this thesis,a pedestrian foot IMU data simulation model is proposed based on the global observability analysis,and the global observability of foot ranging is verified by combining the simulation data.In the simulation of planar square trajectory,the proposed method reduces the distance ratio of plane closure difference of feet from 0.13%/0.15%to 0.03%/0.07%,and the height error from 6.96m/7.3m to 0.09m/0.31 m.Finally,a dual IMU data acquisition system combined with foot ranging is implemented with low-cost devices in this thesis,and three experiments of indoor plane,indoor upstairs and outdoor scene are carried out with this device to verify the effectiveness of the algorithm.Compared with 8.16% of ZUPT only,the error mileage ratio of long-distance outdoor experiment is reduced to 0.51%,and the mean square error is reduced from 59.53 m to 5.52 m,greatly reducing the influence of course error.