Previous Abstract Return to Session A3 Next Abstract

Session A3: Inertial Measurement Units

Research on Dual-MIMU Trajectory Tracking Based on Support Vector Machine Constraint
Wang Qiuying, Cheng Ming, Cui Xufei, Guo Zheng, College of Information and Communication Engineering, Harbin Engineering University, China
Location: Big Sur

Abstract: Inertial navigation is extensively used in many fields especially in military areas. Along with the progress in MEMS technology, the MEMS inertial sensors are more stable and accurate. Installing MEMS inertial sensors on the feet to track trajectory is one of the main methods of indoor independent pedestrian navigation system. Because of its strong autonomy and high hidden characteristics, what’s more?It does not require additional signals and base stations and can be applied in real time. it is widely used in the fields of man combat and personnel rescued. However, due to its algorithm, the inertial error increases as time goes?the error is about to rise to 10000m after 5 min’s walk with the gyroscope whose accuracy is 10 ° / h. In this case, the zero velocity update(ZUPT) method is usually used to correct the information of the carrier .As an error compensation technique, the velocity update(ZUPT) takes the velocity error as the observation, apply the observability analysis to amend other information of the carrier, but not all the errors can be observed. Since the zero velocity update (ZUPT) can only observe velocity information and two horizontal angle information, position information can’t be obtained .Aiming at this problem, put forward a point of view of Dual-MIMU Trajectory Tracking Based on Support Vector Machine(SVM) Constraint. Fix two micro inertial measurement units (MIMU) on the feet of pedestrians. According to the relative position relation of dual MIMU system in force space and the constraint of the maximum step size, the inequality equation is constructed and a Kalman filter algorithm under inequality constraint is designed. The process is described in detail in the paper. According to the paper of Walking Pattern Recognition Based on Inertial Sensing published by Wang Shaochu and Human Walking Analysis and Displacement Calculation Based on Inertial Sensing System wrote by Ren Kaitian from Tianjin University of China. we can see the law of human movement, and the horizontal angle is the highest when the toes are moving from the ground. This paper adopts the theory of support vector machine (SVM) to collect a large number of the angular velocity of the foot movement, then establish the database and divide the process of people’s movement, it improves the number of observations, through observability analysis, improve the accuracy of zero speed correction. The MTI-G710 sensor manufactured by XSENS Company in the Netherlands was used in the experiment to build a real pedestrian test system. The experimenter walk in the teaching building at a normal pace with two feet were fixed with two MTI-G710 sensors. Use the sensor to collect the triaxle angular velocity and triaxle acceleration during pedestrian travel. Result can be obtained through the test after 5min’s walk, then process the data of dual foot-mounted MIMU. The average position error decreases from 10m to 1m, It is fully proves the accuracy of dual foot-mounted MIMU Trajectory Tracking Based on Support Vector Machine Constraint is better compared to single MIMU pedestrian positioning methods, it can provide a higher positioning accuracy and meet the requirements of pedestrian positioning.



Previous Abstract Return to Session A3 Next Abstract