자율 주차 시스템의 직각 주차를 위한 DGPS와 초음파센서 융합

This paper describes a position estimation algorithm by DGPS and ultrasonic sensor fusion for perpendicular parking. The proposed estimation algorithm consists of local Kalman filters for GPS and ultrasonic sensors and a master Kalman filter which integrates local filters. These Kalman filters make up a decentralized Kalman filter. Since a combination system of GPS and IMU could have a bias and accumulated error, ultrasonic sensors are used for correction. The GPS system offers estimated position information and covariance of positions as a local filter. The local Kalman filter for ultrasonic sensors receives distance measurement of an obstacle and calculates estimated obstacle position and covariance of obstacle position. The master filter makes a decision from which information is accurate and gives a weighting to an accurate sensor. This position estimation algorithm can offer accurate target position for parking and handle uncertainties at parking space. The performance of proposed control algorithm has been tested by computer simulations and verified by vehicle test.