라이다 센서를 이용한 접/이안 모니터링 시스템의 대형 선박 추적 방법

Abstract

One of the port operations, the berthing of vessels, is done by a pilot and tug boats at the port because of the low maneuverability of large vessels near the port. However, as vessels approach the port, many accidents occur because the vessel’s surrounding environment or the port facilities cannot be accurately identified by the pilot. To resolve this problem, a LiDAR sensor-based monitoring system has been developed. As LiDAR technologies advances, the sensor provides long-range measurements, but with increased sparseness. Conventional point registration methods generally focus on how to handle uncertainties from sensor measurement noise and/or sparseness from the occlusion of the sensor field of view. In this paper, we propose frame-to-frame sparse point cloud registration by finding the best point-to-plane correspondences from multiple-plane hypotheses. The proposed method is verified with its own dataset and applied to the berthing monitoring system in the real port as a record of extremely rare points.