COHERENT Benchmark

该数据集是一个大规模的现实场景搭载基准，专为不同类型的机器人协作设计，包括四旋翼飞行器、机器狗和机械臂，涵盖了各种真实世界的场景。该基准分为单一类型（32项任务）、双类型（30项任务）和三类型（38项任务）的任务，每种任务都具有不同的复杂性和真实步骤。该数据集包含了100项复杂的长远规划任务，旨在研究异构多机器人任务规划。

This dataset is a large-scale real-world deployment benchmark designed for collaboration among diverse robot types, including quadrotors, quadruped robots, and robotic arms, covering a wide range of real-world scenarios. This benchmark is categorized into three groups: single-type robot tasks (32 tasks), dual-type robot tasks (30 tasks), and triple-type robot tasks (38 tasks), with each task exhibiting distinct complexity levels and realistic execution procedures. This dataset contains 100 complex long-horizon planning tasks, which is intended to facilitate research on heterogeneous multi-robot task planning.