Scholarlink

<p>This paper proposes an innovative approach to solving the multi-robot task allocation (MRTA) problem in complex environments. The proposed methodology integrates k-means clustering for initial task grouping with a multi-objective variable neighborhood search (MOVNS) algorithm enhanced with adaptive iterative neighborhood search (AINS) to optimize the following metrics: task execution time, travel distance, and load balancing across robots. In this proposed method, a Pareto front optimization is employed to balance conflicting objectives, enhancing the scalability and adaptability to large-scale and dynamic task sets. Comparative evaluations demonstrate that our proposed approach achieved reductions of over 30% in travel distance and 40% in task execution time compared to traditional greedy methods, while improving load balancing metrics by 50% in scenarios with 50 tasks and 5 robots. Furthermore, it consistently achieves superior results compared to established state-of-the-art approaches like non-dominated sorting genetic algorithm II (NSGA-II) and strengths-Pareto evolutionary algorithm (SPEA) across all evaluated metrics. These results highlight the scalability and robustness of applying the proposed framework in large-scale task scenarios.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Clusterized Multi-objective Approach With Variable and Adaptive Neighborhood Search for Multi-robot Task Allocation

  • 期刊论文

  • 发表日期：

    2025年10月16日

查看全文
  • Tatiana M. B. dos Santos,
  • Milena F. Pinto,
  • Gabrielle O. Timotheo,
  • Iago Z. Biundini,
  • Celso M. de L. Junior,
  • Leonardo M. Honório,
  • Esteban W. G. Clua

摘要

This paper proposes an innovative approach to solving the multi-robot task allocation (MRTA) problem in complex environments. The proposed methodology integrates k-means clustering for initial task grouping with a multi-objective variable neighborhood search (MOVNS) algorithm enhanced with adaptive iterative neighborhood search (AINS) to optimize the following metrics: task execution time, travel distance, and load balancing across robots. In this proposed method, a Pareto front optimization is employed to balance conflicting objectives, enhancing the scalability and adaptability to large-scale and dynamic task sets. Comparative evaluations demonstrate that our proposed approach achieved reductions of over 30% in travel distance and 40% in task execution time compared to traditional greedy methods, while improving load balancing metrics by 50% in scenarios with 50 tasks and 5 robots. Furthermore, it consistently achieves superior results compared to established state-of-the-art approaches like non-dominated sorting genetic algorithm II (NSGA-II) and strengths-Pareto evolutionary algorithm (SPEA) across all evaluated metrics. These results highlight the scalability and robustness of applying the proposed framework in large-scale task scenarios.