MRTA-Benchmark dataset: 250K Optimal Multi-Robot Task Allocation Instances with Heterogeneous Robots, Precedence Constraints & Dynamic Coalitions

This dataset contains 250,000 optimally solved Multi-Robot Task Assignment (MRTA) instances designed for benchmarking and training task allocation methods. Each instance features heterogeneous robots with varying skill sets, multi-skill tasks requiring tight/dynamic coalition formation, spatially distributed robot and task locations, and precedence constraints between tasks. The scenarios are generated with randomized configurations and solved optimally on the DelftBlue supercomputer using an exact MILP formulation.本数据集包含25万个经最优求解的多机器人任务分配（Multi-Robot Task Assignment, MRTA）实例，专为任务分配方法的基准测试与训练而设计。每个实例均涵盖具备差异化技能集的异构机器人、需进行紧密/动态联盟构建的多技能任务、空间分布的机器人与任务点位，以及任务间的先后约束条件。该数据集的场景通过随机配置生成，并依托代尔夫特蓝（DelftBlue）超级计算机，采用精确混合整数线性规划（Mixed Integer Linear Programming, MILP）模型完成最优求解。

搜集汇总

背景与挑战

背景概述

该数据集包含25万个最优多机器人任务分配（MRTA）实例，专为基准测试和任务分配方法训练设计。其特点包括异构机器人技能、任务间优先级约束、动态联盟形成以及空间分布的机器人任务位置，所有实例均通过精确MILP公式在超级计算机上求解生成。