수중로봇의 실시간 위치추정을 위한 영상소나기반 SLAM

Abstract

This paper presents experimental results of a realtime sonar-based SLAM (simultaneous localization and mapping) technique using the probability-based landmark-recognition method. Sonar based SLAM is used for the navigation of unmanned underwater vehicle (UUVs). Inertial sensors such as inertial measurement units (IMUs) and doppler velocisy logs (DVLs), and external information obtained from sonar are combined using the extended kalman filter (EKF) technique to obtain the navigation information. We estimate the vehicle location using inertial sensor data, and it is corrected using sonar data, which provides the relative position between the vehicle and a landmark placed on the bottom of basin. To verify the suitability of the proposed method, we performed experiments in a basin environment using the UUV, “yShark”.rwater vehicle (UUVs). Inertial sensors such as inertial measurement units (IMUs) and doppler velocisy logs (DVLs), and external information obtained from sonar are combined using the extended kalman filter (EKF) technique to obtain the navigation information. We estimate the vehicle location using inertial sensor data, and it is corrected using sonar data, which provides the relative position between the vehicle and a landmark placed on the bottom of basin. To verify the suitability of the proposed method, we performed experiments in a basin environment using the UUV, “yShark”.