MARS-8, MARS-Loop, MARS-NoLoop

本数据集由上海科技大学自动化与机器人中心的MARS Lab创建，包含三个子数据集：MARS-8、MARS-Loop和MARS-NoLoop，用于评估和比较SLAM算法在不同环境下的性能。数据集通过高级硬件同步的映射机器人收集，结合外部跟踪系统和专业3D扫描仪获取地面实况数据。MARS-8用于基本SLAM评估，MARS-Loop和MARS-NoLoop则分别用于评估闭环性能和无闭环情况下的性能。数据集创建过程中，确保了数据的精确同步和高质量，适用于研究SLAM算法的定位和映射精度，特别是在视觉SLAM领域的应用。

This dataset was developed by the MARS Lab at the Center for Automation and Robotics, ShanghaiTech University. It comprises three sub-datasets: MARS-8, MARS-Loop, and MARS-NoLoop, which are tailored for evaluating and comparing the performance of SLAM (Simultaneous Localization and Mapping) algorithms across various environments. The dataset was collected via a mapping robot synchronized with high-grade hardware, with ground truth data acquired using an external tracking system and professional 3D scanners. Specifically, MARS-8 is designed for basic SLAM assessment, while MARS-Loop and MARS-NoLoop are respectively used to evaluate loop closure performance and performance in loop closure-free scenarios. During its creation, strict measures were taken to ensure precise synchronization and high data quality, rendering the dataset suitable for researching the localization and mapping accuracy of SLAM algorithms, especially for visual SLAM applications.