One Thousand and One Hours: Self-driving Motion Prediction Dataset
收藏资源简介:
本数据集名为‘一千零一小时:自动驾驶运动预测数据集’,由Lyft Level 5团队创建,旨在为自动驾驶车辆的运动预测和规划提供大规模详细数据。数据集包含170,000个场景,每个场景25秒,覆盖20辆自动驾驶车辆在帕洛阿尔托的一段6.8英里路线上的1,000小时记录。数据集不仅包含车辆、行人及自行车等交通参与者的精确位置和运动数据,还包含高清晰度语义地图和航空视图,用于辅助预测。此数据集适用于开发自动驾驶机器学习任务,如运动预测、运动规划和模拟,旨在解决自动驾驶中的关键问题,如提高预测准确性和规划效率。
This dataset is named "One Thousand and One Hours: Autonomous Driving Motion Prediction Dataset", created by the Lyft Level 5 team. It is designed to provide large-scale, detailed data for motion prediction and planning of autonomous vehicles. The dataset includes 170,000 scenarios, each with a duration of 25 seconds, covering 1,000 hours of driving records collected from 20 autonomous vehicles traveling along a 6.8-mile route in Palo Alto. Besides precise position and motion data of traffic participants such as vehicles, pedestrians and bicycles, the dataset also provides high-definition semantic maps and aerial views to assist prediction-related tasks. This dataset is suitable for developing machine learning tasks for autonomous driving, including motion prediction, motion planning and simulation, and aims to solve key challenges in autonomous driving such as enhancing prediction accuracy and improving planning efficiency.



