WOMD-LiDAR
收藏资源简介:
WOMD-LiDAR是由Waymo LLC创建的大规模、高质量、多样化的LiDAR数据集,专门用于自动驾驶中的运动预测任务。该数据集包含超过100,000个场景,每个场景持续20秒,涵盖了城市和郊区的多种地理环境。与Waymo Open Dataset相比,WOMD-LiDAR数据集场景数量增加了100倍。数据集的创建过程中,采用了先进的LiDAR压缩技术,将原始数据压缩了约8倍,使得最终数据集大小约为2.3TB。WOMD-LiDAR的应用领域主要集中在自动驾驶系统的运动预测模型开发,旨在通过提供丰富的原始传感器数据,提高模型的预测准确性和鲁棒性。
WOMD-LiDAR is a large-scale, high-quality, and diverse LiDAR dataset developed by Waymo LLC, exclusively tailored for motion prediction tasks in autonomous driving. This dataset comprises over 100,000 scenarios, each with a duration of 20 seconds, covering a wide range of urban and suburban geographic environments. Compared to the Waymo Open Dataset, the WOMD-LiDAR dataset boasts a 100-fold increase in the number of scenarios. During the dataset curation process, advanced LiDAR compression technologies were employed to compress the raw sensor data by a factor of approximately 8, yielding a final dataset size of around 2.3 TB. The primary application scope of WOMD-LiDAR lies in the development of motion prediction models for autonomous driving systems, where the goal is to enhance the prediction accuracy and robustness of such models by providing rich raw sensor data.

- 1WOMD-LiDAR: Raw Sensor Dataset Benchmark for Motion ForecastingWaymo LLC · 2024年



