Research On Pedestrian Trajectory Estimation Algorithms Based On MEMS Inertial Navigation

Geographical location information of the firefighters cannot be provided in time when they work in forest safety in highly confined forests.It is easy to lose satellite signals owing to the shelter of trees.Inertial navigation system can provide the function of autonomous positioning,but it can hardly achieve the expected requirements during the positioning due to the cumulative error prone to occur during the integral process in practical applications.Therefore,the inertial navigation system was combined with the magnetometer information and pedestrian gait information based on the sensor fusion technology.Moreover,a foot-mounted hardware platform was designed to realize the calculation of pedestrian trajectory in this work.Firstly,the operating principle of MEMS inertial navigation system was explained in this study.The error sources of accelerometer,gyroscope and magnetometer were analyzed,and then the error models of the MEMS sensors were constructed.In addition,accelerometer calibration algorithm based on a least square method,magnetometer calibration algorithm based on ellipsoid fitting and gyroscope zero bias removal algorithm were designed.The calibration and compensation of the sensor errors were well realized.Secondly,the pedestrian trajectory estimation algorithm based on MEMS inertial navigation was studied in this work.The quaternion-based pose update algorithm,initial alignment algorithm,speed and position update algorithm were elucidated.The typical gait of pedestrians is analyzed and divided into two stages:swing and stationary.In the former stage,inertial navigation is used to calculate the step size,and zero-velocity updating is used to correct sensor errors in the latter stage.Two zero-velocity detection algorithms are designed based on angular velocity and acceleration,and the perfornances of the two algorithms are tested and compared.The Kalman filtering algorithm was introduced based on error state.The system state and observation equations were established.Inertial navigation and sensor errors were corrected based on the zero velocity correction method and Kalman filter.Finally,a foot-binding experimental platform is built according to the system requirements,and the numerous experiments such as zero-speed detection,single-step and continuous-step were performed to verify the proposed methods.The experimental results show that the zero-speed detection algorithm presenled in this study has a relative error of 0.52%only in the case of fast walking,which is much less than the threshold method.Using zero-speed correction algorithm based on Kalman filter,the relative error of horizontal square experiment is less than 4.3%,and the relative error range of indoor floor experiment is-5.8%?9.7%.These data indicate that it can effectively reduce the long-time drift of inertial navigation,and the positioning effect is relatively stable.It can meet the positioning requirements of forest fire fighters.