ABSTRACT This project aims at developing an efficient simulation environment for the vessel navigation system. The design of the overall navigation system is described along with its application for the case study of an autonomous vessel. The UNITY and Robot Operating System were utilised as the virtual environment and the main control framework, respectively. By treating the autonomous vessel as a single robot, each part of the system was formulated to enhance the efficiency and visualisation in the development progresses for autonomous system. The virtual environment using UNITY overwhelms the space-time constraints in the testing stage. The navigation system combined with conventional navigation algorithms and a computer vision algorithm were implemented in the Robot Operating System. Hence, the navigation system was enhanced by an assistant object detection algorithm to be more active on fusing environment information. The excellence of the proposed object detection model was demonstrated with reliable performance with 94% mean Average Precision, which renders the model exhibiting visual sensibility in navigation. Overall, it is believed that this study can contribute to offering some insights into developing autonomous marine vessels.
Read full abstract