Two-Level Evolutionary Approach to Persistent Surveillance for Multiple Underwater Vehicles with Energy Constraints

Abstract

Currently, the coordinated use of autonomous underwater vehicles groups seems to be the most promising and ambitious technology to provide a solution to the whole range of oceanographic problems. Complex and large-scale underwater operations usually involve long stay activities of robotic groups under the limited vehicle’s battery capacity. In this context, available charging station within the operational area is required for long-term mission implementation. In order to ensure a high level of group performance capability, two following problems have to be handled simultaneously and accurately – to allocate all tasks between vehicles in the group and to determine the recharging order over the extended period of time. While doing this, it should be taken into account, that the real world underwater vehicle systems are partially self-contained and could be subjected to any malfunctions and unforeseen events. The article is devoted to the suggested two-level dynamic mission planner based on the rendezvous point selection scheme. The idea is to divide a mission on a series of time-limited operating periods with the whole group rendezvous at the end of each period. The high-level planner’s objective here is to construct the recharging schedule for all vehicles in the group ensuring well-timed energy replenishment while preventing the simultaneous charging of a plenitude of robots. Based on this schedule, mission is decomposed to assign group rendezvous to each regrouping event (robot leaving the group for recharging or joining the group after recharging). This scheme of periodic rendezvous allows group to keep up its status regularly and to re-plan current strategy, if needed, almost on-the-fly. Low-level planner, in return, performs detailed group routing on the graph-like terrain for each operating period under vehicle’s technical restrictions and task’s spatiotemporal requirements. In this paper, we propose the evolutionary approach to decentralized implementation of both path planners using specialized heuristics, solution improvement techniques, and original chromosome-coding scheme. Both algorithm options for group mission planner are analyzed in the paper; the results of computational experiments are given.