Autonomous Berthing System For Ships

With the rapid development of artificial intelligence and big data technology,the research of autonomous ships has attracted considerable attention in shipping industry.According to “Guidelines for Autonomous Cargo Ships” released by China Classification Society,autonomous berthing is an important capability of autonomous cargo ships,which include situation-awareness capability and berthing operation capability.This paper proposes a multi-sensor fusion algorithm,and further develops an autonomous berthing system.In the Guidance-NavigationControl framework,the Navigation sub-system consists of three modules,namely multi-sensor fusion,target detection,and target tracking.The module of multi-sensor fusion combines the point cloud data and image data with rasterization and semantic segmentation to form a valid data format based on a spatial geometric model.The module of target detection employs the grid clustering technique based on DBSCAN and cluster classification technique based on multi-feature matching to discover berthing targets with specific characteristics,and further determines the identification and positioning of targets.The module of target tracking employs the technique of data nearest neighbor association to correlate the target and the detection value,and further estimates the berthing target’s state.The Guidance subsystem plans the berthing path according to the berthing target.The Control sub-system employs the LOS algorithm and PID controller to complete the berthing operation.The simulation tests and on-water tests of autonomous berthing of a trimaran model validate the effectiveness and reliability of the proposed algorithm.It has significant value for autonomous berthing application.