Path Planning of an Underwater Vehicle by CFD Numerical Simulation Combined with a Migration-Based Genetic Algorithm

This paper proposes a physics-informed global path planning framework for underwater vehicles integrating CFD simulation and the genetic algorithm. The CFD simulation models the flow field along the planned path of the underwater vehicle. The current velocity data are incorporated into the following path planning that is based on an improved genetic algorithm (GA), which uses migration operators to share the information about feasible solutions or paths, improving the fitness of the whole population. In the three steps of the GA procedure, an elite selection strategy is adopted to avoid losing excellent solutions. A segmented crossover strategy is adopted to avoid low-quality crossover. An adaptive mutation strategy is used to enhance the ability to escape a local optimal solution. Using the improved GA, single-target and multi-target underwater path planning are investigated. In multi-target path planning, a combined algorithm is proposed to solve the optimal traversal order of target points and plan a feasible path between target points. The simulation results show that the proposed algorithm has good planning ability for both simple and complex underwater scenarios. Compared with the conventional GA and an improved GA, the number of average iterations decreases by 45.3% and 29.9%, respectively, for 2D multi-target path planning. The number of average inflection points decreases by 50.3% and 44.2%, respectively, for 2D multi-target path planning.