Constrained path planning of autonomous underwater vehicle using selectively-hybridized particle swarm optimization algorithms

This paper presents an autonomous underwater vehicle (AUV) path planning scenario as an optimization problem constrained by the combination of hard constraints and soft constraints. The path planner aims to generate the optimum path that safely guides an AUV through an ocean environment with priori known obstacles and non-uniform currents in both 2D and 3D. The path planner uses 2 variants of particle swarm optimization (PSO) algorithms, which are the selectively Differential Evolution (DE)-hybridized Quantum PSO (SDEQPSO) and Adaptive PSO (SDEAPSO). The performances of the path planners using different constraints are analyzed in a series of extensive Monte Carlo simulations and ANOVA (analysis of variance) procedures based on their respective solution qualities, stabilities and computational efficiencies. Based on the simulation results, the SDEQPSO path planner with the setting of hard constraint for boundary condition and soft constraint for obstacle avoidance was found to be able to generate smooth and feasible AUV path with higher efficiency than other algorithms, as indicated by its relatively low computational requirement and excellent solution quality.