Obstacle recognition method for ship based on 3D lidar

Abstract

With the rapid development of intelligent ships, the perception of the navigation environment, which is the basis of intelligent navigation, has attracted more and more attention. At the same time, the lidar with high-ranking accuracy and strong anti-interference ability has also received more attention in the ship perception field. Aiming at the automatic detection and identification of typical obstacles for intelligent ships during navigation, an algorithm is proposed for the obstacle recognition of waterways based on 3D lidar. At first, the original point cloud obtained by 3D lidar is preprocessed to filter out noise signals. Then, the water surface and the waterway line are segmented by analyzing the point cloud features. Finally, a new DBSCAN clustering model is proposed and the obstacle features are summarized to complete the classification. A series of field experiments have been carried out. The experimental results show that the algorithm performs high accuracy and reliability for obstacle recognition.