比亚迪唐五车型采集自动驾驶场景风险数据集

数据集主要面向基于路端感知的自动驾驶等级提升方法研究，涵盖实车采集数据，信息包括交通参与者的坐标、速度等。总计16个实际场景片段，共1732张图像、1732个数据记录文件。内含1个数据集说明文件。课题组以比亚迪唐五车型（插电式混合动力SUV）为车辆平台，并配备了一套感知系统，用于收集道路交通参与者的数据。该数据集以实车采集的多维度路况信息为基础，涵盖了城市道路、复杂路口等多样化场景，且针对不同场景中可能出现的突发状况（如行人横穿、车辆加塞等）进行了精准的风险等级标注。数据量大约3.95GB。

This dataset is primarily intended for research on methods to enhance the level of road-side perception-based autonomous driving. It includes real-vehicle collected data containing information such as coordinates and speeds of traffic participants. In total, there are 16 real-world scene segments, consisting of 1732 images and 1732 data record files, along with 1 dataset description document. The research team adopted the BYD Tang 5 plug-in hybrid SUV as the experimental vehicle platform, which was equipped with a set of perception systems to collect data on road traffic participants. Based on multi-dimensional road condition information collected by real vehicles, this dataset covers diverse scenarios including urban roads and complex intersections. Additionally, accurate risk level annotations are provided for potential emergencies in various scenarios, such as pedestrians crossing the road and vehicles cutting in traffic. The total data volume of the dataset is approximately 3.95 GB.

搜集汇总

背景与挑战

背景概述

该数据集面向基于路端感知的自动驾驶等级提升研究，包含实车采集的交通参与者坐标和速度等多维度信息，总计16个实际场景片段、1732张图像及对应数据文件。数据集以比亚迪唐五车型为平台，覆盖城市道路和复杂路口等多样化场景，并针对突发状况进行了精准的风险等级标注，数据量约为3.95GB。