COSMO-Bench
收藏资源简介:
COSMO-Bench是一个用于分布式C-SLAM后端评估的基准数据集系列,包含24个数据集,来源于先进的C-SLAM前端和真实世界的LiDAR数据。这些数据集旨在反映多机器人团队在实际部署中观察到的数据,包括非参数/非平稳噪声的测量、包含异常值的测量等,以测试分布式C-SLAM方法在现实世界条件下的操作能力。COSMO-Bench数据集提供了丰富的闭环,支持长时间的轨迹和准确的参考解决方案,以及真实的噪声模型和测量分类,为研究人员测试和评估新型C-SLAM后端提供了便利。
COSMO-Bench is a benchmark dataset series for evaluating distributed C-SLAM backends. It includes 24 datasets sourced from state-of-the-art C-SLAM frontends and real-world LiDAR data. These datasets are designed to reflect the data encountered by multi-robot teams during actual deployments, including non-parametric/non-stationary noise measurements and measurements with outliers, to test the operational capabilities of distributed C-SLAM methods under real-world conditions. The COSMO-Bench dataset series offers rich loop closures, supports long-duration trajectories and accurate reference solutions, as well as realistic noise models and measurement classifications, facilitating researchers to test and evaluate novel C-SLAM backends.




