Development of Onboard Decision Supporting System for Ship Docking Operations

Abstract

Maritime operations inevitably influenced by the wind, wave, sea currents, or other perturbations at sea. Providing decision support for these operations based on historical and real-time data of ship status is thus of great concern in terms of ship safety. However, it is challenging for collecting and analysis large quantities of ship data in real operations. Moreover, the development of an onboard decision support system (DSS) will be a gradual and iterative process subject to extensive testing and simulation. Consequently, the paper presents an integrated simulation framework which provides testing and simulation environment for the DSS development. The system enables navigation data transmission from a well-designed simulator and automatic determining of the safe maneuver of a ship within the framework. The development of DSS is divided into three steps. Firstly, we collect the ship maneuvering data from the simulator and classify them; Then we implement an imitation learning (IL) algorithm to learn an initial policy from the data; Finally, based on the policy, the reinforcement learning (RL) algorithm is used to determine the safe decision for operations. In this way, it could speed up the learning efficiency by extracting more information from available experience. To verify the effectiveness of the proposed integrated simulation framework, in this study, we implemented the proposed DSS in ship docking operation under various environmental disturbances. It is interacted with the simulator to obtain data. By processing these data, it provides the shipmaster with the information about the consequences of the ship maneuvering decisions. The simulation results demonstrate that the proposed DSS could assist the shipmaster in deciding policies and increase the efficiency of decision making.