Load diversity-driven local task reallocation for unmanned surface vehicle swarm in maritime search and rescue

Objectives A load diversity-driven local reallocation algorithm is proposed to address the difficulty of achieving optimal resource allocation in post-triggered task reallocation under the dual constraints of endurance and task carrying capacity for unmanned surface vehicles (USVs) in offshore rescue scenarios. Methods Firstly, an incremental local reallocation strategy is designed for unsaturated-load USVs within the performance impact framework to overcome the inadequate adaptability of global reallocation when pre-allocation algorithms are directly applied in the reallocation stage. This strategy performs local reallocation on newly added tasks based on the pre-allocation sequence and task execution status to achieve rapid response to new tasks. Then, a neighborhood replacement local reallocation strategy is further developed to solve the limitations of the aforementioned strategy, which only applies to unsaturated-load USVs and leads to insufficient utilization of rescue resources. This strategy replaces assigned tasks with new tasks and ensures the reassignment of replaced tasks to other USVs. It maximizes the number of allocated tasks under time constraints and maintains system stability simultaneously. Results Experimental results demonstrate that, compared with the Performance Impact algorithm, the proposed algorithm improves the task completion rate by 5.1% and reduces the single decision time by 48.4% while satisfying task time constraints, effectively enhancing the rescue resource utilization efficiency of unmanned surface vehicles. Conclusions The proposed load diversity-driven local reallocation algorithm effectively addresses the task reallocation problem under resource-constrained conditions, providing technical support for task scheduling of unmanned surface vehicles in offshore rescue operations.