Application of stochastic optimization to collision avoidance

This paper applies simulation-based optimization to the problem of vessel traffic management in a high vessel density environment. Specifically, a Monte Carlo simulation has been developed that models a relatively small craft operating in a high vessel density environment under poor visibility conditions. In this simulation, the small vessel maneuvers to keep all other vessels outside some acceptable range (one of the objectives of the optimization) subject to the requirements on the other vessels to obey pre-established traffic management rules. Several stochastic optimization algorithms- blind random search, simultaneous perturbation stochastic approximation, and simulated annealing are applied to this problem ´with conclusions drawn regarding their relative applicability and performance, as well as the practical implications of the results.