Iterated local search for coverage path planning with imperfect extended detection

Abstract

Naval mines, many of which are remnants of past conflicts, still threaten ships, wildlife, and the environment. To detect these underwater mines, autonomous minesweepers can be used, e.g. a REMUS 100 autonomous underwater vehicle (AUV) equipped with sidescan sonars. We frame the minesweeping task as a combinatorial optimization problem called coverage path planning with imperfect extended detection (CPPIED). An efficient solution to the CPPIED minimizes both the path length and the number of turns (in that order), while satisfying stochastic coverage constraints in a grid environment. The current state of the art is a hybrid heuristic called DpSweeper. We show that, although this algorithm rapidly produces feasible solutions, the path quality is limited by its greedy nature. Since local search algorithms are known for their capacity to improve feasible solutions to large-scale combinatorial problems, such as vehicle routing problems with spatial coverage, we propose an iterated local search heuristic to enhance the solutions generated by DpSweeper. Our heuristic reduces the path length of solutions provided by DpSweeper by 9% to 28%, and the number of turns by 19% to 42%, hence improving the planning methods for minesweeping operations and bridging vehicle routing and coverage path planning problems.