Kimera-Multi Dataset

We conducted live experiments on the MIT campus with a fleet of Clearpath Jackal rovers equipped with a Realsense D455 RGB-D Camera and an Ouster or Velodyne 3D LiDAR. During the experiment, there were often humans and vehicles moving around the robots, and the robots traverse through a variety of both indoor and outdoor environments. We compiled three real-world datasets based on the data recorded. The first dataset is the Campus-Outdoor dataset, which consists of six Jackal rovers traversing a total of 6044 meters over a duration of around 20 minutes on the MIT campus. The second dataset is the Campus-Tunnels dataset, which consists of eight Jackal rovers traversing a total of 6753 meters over a duration of almost 30 minutes in the tunnels underneath the university campus. The trajectories are entirely indoors and consist mostly of long homogeneous tunnels and corridors. The last dataset is the Campus-Hybrid dataset, which consists of eight Jackal rovers traversing a total of 7785 meters over a duration of almost 30 minutes both on and below the university campus. This dataset is also multi-level, with some trajectories crossing each other on different levels. For more in depth description, see (Tian et al., 2023).

Dataset Repo

• MIT-SPARK/Kimera-Multi-Data

References

1. Tian, Y., Chang, Y., Quang, L., Schang, A., Nieto-Granda, C., How, J. P., & Carlone, L. (2023). Resilient and Distributed Multi-Robot Visual SLAM: Datasets, Experiments, and Lessons Learned. IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS).
   Bibtex

   @inproceedings{Tian23iros-KimeraMultiExperiments,
     author = {Tian, Y. and Chang, Y. and Quang, L. and Schang, A. and Nieto-Granda, C. and How, J.P. and Carlone, L.},
     title = {Resilient and Distributed Multi-Robot Visual {SLAM}: Datasets, Experiments, and Lessons Learned},
     booktitle = {IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems (IROS)},
     note = {\linkToPdf{https://arxiv.org/pdf/2304.04362.pdf},\linkToVideo{https://youtu.be/7yYMRNMdKjY},\linkToCode{https://github.com/MIT-SPARK/Kimera-Multi},\linkToWeb{https://github.com/MIT-SPARK/Kimera-Multi-Data}\award{, Open Data prize honorable mention}},
     pdf = {https://arxiv.org/pdf/2304.04362.pdf},
     funding = {DCIST, MathWorks},
     year = {2023}
   }