CARLA-Loc: Synthetic SLAM Dataset with Full-stack Sensor Setup in Challenging Weather and Dynamic Environments

Yuhang Han, Zhengtao Liu, Shuo Sun, Dongen Li, Jiawei Sun, Ziye Hong, Marcelo H. Ang Jr.
National University of Singapore
2023
Code arXiv

Quick preview of all maps (x10 play speed, Clear Noon and dynamic presets.)

Abstract

Simultaneous Localization and Mapping plays a crucial rule in the reliability of autonomous driving and acts as the prerequisite for subsequent tasks like motion prediction, path planning, and high-precision map construction. Toward in-depth analysis on the effect for overall SLAM performance in challenging environmental conditions and providing a more comprehensive benchmark, we propose CARLA-Loc, a synthetic dataset of challenging and dynamic environments built on CARLA simulator. We integrate multiple sensors into the dataset with strict time synchronization. 8 scenarios are posed in our dataset with different dynamic levels and weather conditions.

BibTeX

@article{han2023carla,
        title={CARLA-Loc: Synthetic SLAM Dataset with Full-stack Sensor Setup in Challenging Weather and Dynamic Environments},
        author={Han, Yuhang and Liu, Zhengtao and Sun, Shuo and Li, Dongen and Sun, Jiawei and Hong, Ziye and Ang Jr, Marcelo H},
        journal={arXiv preprint arXiv:2309.08909},
        year={2023}
      }