부산대학교 도서관

GNSS(Global Navigation Satellite System) sensors are commonly used for the localization of autonomous vehicles or mobile robots in outdoor environments. However, there is a limitation that the uncertainty of the GNSS sensors becomes too large to be used for the localization in the urban environment called ‘street canyon’ because the signal from satellites is interfered by many obstacles such as buildings around the GNSS sensor. In this paper, it is proposed to measure the magnitude of the uncertainty of the GNSS sensor in urban environments experimentally and localize the autonomous driving robot based on the fusion of GNSS sensor data and odometry data using an EKF(Extended Kalman Filter). Through the proposed method, the GNSS sensor model that reflects the characteristics of errors of the GNSS sensor can be constructed. The proposed method proved that the uncertainty of the GNSS sensor can be modeled accurately in urban environments where the uncertainty of the GNSS sensor is large compared with the NMEA 0183 standard information provided by the GNSS sensor. Therefore, the proposed method proved that it is possible to use the GNSS sensor for the localization of the autonomous driving robot in urban environments.