Despite receiving much attention from researchers in the field of naval architecture and marine engineering since the early stages of modern shipbuilding, the berthing phase is still one of the biggest challenges in ship maneuvering due to the potential risks involved. Many algorithms have been proposed to solve this problem. This paper proposes a new approach with a path-planning algorithm for automatic berthing tasks using deep reinforcement learning (RL) based on a maneuvering simulation. Unlike the conventional path-planning algorithm using the control theory or an advanced algorithm using deep learning, a state-of-the-art path-planning algorithm based on reinforcement learning automatically learns, explores, and optimizes the path for berthing performance through trial and error. The results of performing the twin delayed deep deterministic policy gradient (TD3) combined with the maneuvering simulation show that the approach can be used to propose a feasible and safe path for high-performing automatic berthing tasks.

It is necessary to predict hydrodynamic derivatives when assessing the maneuverability of a submarine. The force and moment acting on the vehicle may affect its motion in various modes. Conventionally, the derivatives are determined by performing captive model tests in a towing tank or applying a system identification method to the free running model test. However, a computational fluid dynamics (CFD) method has also become a possible tool to predict the hydrodynamics. In this study, virtual captive model tests for a full-scale submarine were conducted by utilizing a Reynolds-averaged Navier-Stokes solver in ANSYS FLUENT version 18.2. The simulations were carried out at design speed for various modes of motion such as straight forward, drift, angle of attack, deflection of the rudder, circular, and combined motion. The hydrodynamic force and moment acting on the submarine appended rudders and stern stabilizers were then obtained. Finally, hydrodynamic derivatives were determined, and these could be used for evaluating the maneuvering characteristics of the submarine in a further study.

This study presents a pressure variation model (PVM) derived from the regression analysis of dynamic pressure computed through numerical analysis to estimate the velocity of underwater vehicles. Furthermore, the drift angle estimation algorithm was developed using predicted velocities from PVM and pressure sensor differences. This approach estimates the single-motion states of underwater vehicles, such as straight, turning, and gliding. Furthermore, it confirms the viability of state estimation even in multiple motions involving turning and gliding motion with a drift angle and spiral motion. The comparison with numerical analysis results validated prediction accuracy within 15%.

This study proposes a computational fluid dynamics (CFD)-modified potential (CMP) model. This hybrid model uses linear potential theory with a corrected damping ratio obtained from CFD simulations to analyze the seakeeping performance of a small vessel. According to the analysis procedure of the proposed model, a motion analysis, including the prediction of the roll and pitch damping ratio of a small barge, was conducted; to verify reliability; the results were compared to those of an experiment performed in a physical tank. The relative errors in the experiment for peak amplitude in the roll motion response amplitude operators (RAOs) using the CMP model were relatively small, whereas those obtained from only the potential analysis were large errors in all three conventionally used roll-damping ratios. In addition, the computational time consumed by the CMP model was longer than that consumed by the potential theory but faster than the full CFD simulation for all wave conditions. Subsequently, based on the motion analysis results, the seakeeping performance was evaluated in a real sea environment, and the results on the single significant amplitude (SSA) were discussed through comparison with the results of the potential analysis and experiment.