카메라센서를 이용한 장애물 차량위치 추정의 보정

Abstract

This paper presents Kalman filter design and experimental validation of the obstacle position estimation with stereo vision sensor. In detecting front vehicles by using stereo vision sensor, there are limitations that as measurement distance becomes larger, variance of measurement error becomes larger as well. So, using static measurement noise covariance does not guarantee accurate position estimation in Kalman filter for total distance range. This paper proposes dynamic measurement noise covariance scheme to achieve more accurate position estimation in Kalman filter. The proposed algorithm was evaluated by MATLAB/Simulink and validated by experiment with a vehicle equipped with laser scanner and stereo vision sensor.