Path planning of multiple autonomous marine vehicles for adaptive sampling using Voronoi-based ant colony optimization

Abstract

In this paper, a hybrid Voronoi-based ant colony optimization (V-ACO) technique for multiple autonomous marine vehicles (AMVs) is proposed to solve adaptive ocean sampling problem. The Voronoi-based scheme utilizes Voronoi partition with tournament selection method that enables more Voronoi edges lie in higher scientific interest regions. This scheme is then combined with ant colony optimization (ACO) using modified heuristic function, to find collision-free optimal trajectories for multiple AMVs to collect ocean measurements. For comparison, conventional ACO, rapidly-exploring random tree star (RRT*) and Dijkstra’s algorithm are also applied and tested for adaptive ocean sampling. Results of simulation tests specifically highlight the effectiveness and robustness of the proposed V-ACO path planner in generating trajectories of multi-AMVs that maximize data collection for adaptive ocean sampling in high scientific interest areas while considering specified mission time, inter-vehicle and obstacles avoidance constraints. Furthermore, field experiments validate the capability of the proposed V-ACO path planner in finding optimal solutions for adaptive ocean sampling.