Monte-Carlo simulation method for ship collision avoidance performance considering various encounter situations in port congestion zones

As interest in autonomous maritime technology continues to grow, various collision avoidance algorithms for autonomous vessels have been developed. However, evaluating and comparing the performance of these algorithms presents challenges due to the significant influence of factors such as the number of obstacles, specific encounter scenarios and obstacle arrangements. To address these challenges, the present study employs a Monte Carlo simulation technique to quantitatively evaluate the performance of ship collision avoidance algorithms in port congestion zones. To accurately reflect the conditions of real congested harbor areas with high ship traffic, four major encounter scenarios were defined, with the positions, velocities, and heading angles of obstacles randomly generated within predefined ranges to incorporate randomness into the simulation environment. Using the developed Monte Carlo simulation technique, the performance of the Worst Case Velocity Obstacle (WVO) algorithm and a hybrid algorithm combining WVO with modified Artificial Potential Field algorithm(APF) were qualitatively evaluated. The simulation results revealed the limitations of the WVO algorithm, particularly in scenarios involving crossing and converging encounters, due to frequent heading changes and passive rudder actions. In contrast, the hybrid algorithm, which incorporate WVO algorithm with modified APF method, demonstrated improved collision avoidance performances, including maintaining greater safe distance and reducing collision occurrence through actively using rudder angles.