단축 가속도계의 Pseudo 바이어스 모델을 이용한 필터 기반 차량 속력 추정 알고리즘

Abstract

This study presents a filter-based vehicle longitudinal speed estimation algorithm with improved accuracy by estimating the accelerometer bias using only a single-axis accelerometer in the section where the odometer reliability is low. Most vehicles use odometers to obtain speed information because odometers can measure speed with high accuracy under normal driving conditions. However, since the typical odometer has a characteristic that depends on the measured revolutions per minute of the vehicle wheel, if a slip situation occurs in which the wheel slides owing to a sudden start and sudden stop, then the odometer outputs an inaccurate speed measurement. To solve this problem, we redefine the sensor error model of the accelerometer attached to the vehicle and propose a method to estimate the attitude and self-bias component, which are error components that have large effects on the accelerometer, through an adaptive lowpass filter. In addition, we propose an approach to apply the zero-velocity update method to improve the accuracy of the estimated speed information by integrating the error-compensated acceleration value based on the filter. To evaluate the performance of the proposed algorithm, an actual vehicle experiment was performed, and the speed estimation results of the odometer disabled area were analyzed. The results confirmed that the accuracy of speed estimation was greatly improved.