Multi-USV Task Assignment Based on NSGA II-MC

With the development of artificial intelligence technology, the level of autonomy of unmanned platforms is continuously improving. They are capable of completing certain specific tasks based on external information inputs and can achieve battlefield situational awareness. However, when USVs (Unmanned Surface Vehicles) search for underwater targets, the problem of target motion dispersion is particularly prominent. To address the multi-USV task allocation optimization problem, an allocation optimization model for USV clusters has been constructed, which aims to maximize the probability of target detection and minimize the collaborative time of USV clusters. An improved task allocation optimization algorithm, NSGA II-MC (Non-dominated Sorting Genetic Algorithm II-Monte Carlo), has been proposed. Firstly, a more realistic detection model was selected. Secondly, the encoding strategy and crossover mutation operations of the algorithm were adjusted. An adaptive crossover and mutation probability mechanism was introduced, and the fitness was calculated using the Monte Carlo method. Experimental results show that compared with the traditional NSGA-II algorithm, the proposed algorithm generated two more solutions in the model solving process and exhibited better convergence. The research results enrich the methods of multi-objective optimization algorithms and multi-USV task allocation, providing strong support for more efficient underwater target search.