ROLiD
收藏资源简介:
ROLiD数据集是由福建省智慧城市感知与计算重点实验室和厦门大学信息学院创建的,主要用于研究LiDAR点云数据中的随机物体干扰。该数据集包含水雾和烟雾的点云数据,共计超过128,000帧。数据通过32线LiDAR在开放工厂环境中采集,分别在车辆的前、后、左、右四个方向喷洒水雾和释放烟雾,并记录点云数据。数据集的创建旨在模拟真实世界中LiDAR传感器在复杂环境下的感知干扰,特别是在自动驾驶系统中的应用,以提高系统的安全性和鲁棒性。
The ROLiD dataset was developed by the Fujian Provincial Key Laboratory of Smart City Perception and Computing and the School of Information Science, Xiamen University, and is primarily intended for research on random object interference in LiDAR point cloud data. This dataset includes point cloud data of water mist and smoke, with a total of over 128,000 frames. The data was collected using a 32-channel LiDAR in an open factory environment, where water mist was sprayed and smoke was released from the front, rear, left, and right directions of a test vehicle respectively, while the corresponding point cloud data was recorded synchronously. The dataset is designed to simulate perceptual interference of LiDAR sensors in complex real-world scenarios, particularly for applications in autonomous driving systems, so as to improve the safety and robustness of such systems.

- 1A New Adversarial Perspective for LiDAR-based 3D Object Detection福建省智慧城市感知与计算重点实验室,信息学院,厦门大学,中国 · 2024年



