多机器人任务调度的簇引导自适应麻雀算法

Cluster-Guided Adaptive Sparrow Algorithm for Multi-Robot Task Scheduling

摘要:

云计算环境下的多机器人任务需在就绪时间、截止时间、执行时长、优先级与CPU异构等多重约束下实现高效调度。为提升任务分配的可行性与资源利用效率，构建了一种簇引导的自适应麻雀搜索算法，通过引入任务特征聚类、选择性反向学习与分级混合变异机制实现全局探索与局部精化的协同优化。在多规模数据集上的实验表明，该算法在收敛速度、任务完成率、按时完成率与完工时间等指标上均优于多种经典优化方法，并在大规模场景中表现出稳定性与可扩展性。结果验证了所设计方法能够有效降低跨簇任务分散导致的冲突代价，适用于复杂、异构与大规模的多机器人任务调度应用。

Abstract:

Efficient task allocation in cloud-based multi-robot systems requires handling multiple constraints, including release times, deadlines, execution durations, priorities, and heterogeneous CPU capacities. To address these challenges, a cluster-guided adaptive Sparrow Search Algorithm (CG-SSA) is developed, which integrates task feature clustering, selective opposition-based initialization, and a multi-level hybrid mutation strategy to balance global exploration and local refinement. Experiments conducted on datasets of various scales show that the proposed method achieves faster convergence, higher task completion and on-time rates, and shorter makespan compared with several classical optimization algorithms. The results demonstrate that the method effectively reduces conflict costs caused by cross-cluster task dispersion and exhibits strong stability and scalability in large-scale heterogeneous task scheduling scenarios.