常见自动驾驶高精地图数据集对比分析
Comparative Analysis of Common Autonomous Driving High-Precision Map Datasets

随着自动驾驶研究的蓬勃发展，传统电子地图无法满足自动驾驶领域需求，从而面向于计算机的自动驾驶系统的高精地图应运而生，而作为高精地图基础——高精地图数据集显得尤为重要，但目前高精地图数据集领域发展刚刚起步，对于常见自动驾驶高精地图数据集对比尚不完善，本文针对上述问题，收集目前主流和新兴的7类数据集，简要介绍其主要内容，并从识别目标、识别环境、数据规模和应用领域方面进行量化分析，同时对主流数据集主要内容从驾驶场景、天气条件、识别目标、数据体积、传感器类别等多方面进行归纳分析得出各大数据集的适用场景以及存在的不足。
With the vigorous development of autonomous driving research, traditional electronic maps cannot meet the needs of the field of autonomous driving, so high-precision maps for computer-oriented autonomous driving systems have emerged as the times require. As the basis of high-precision maps, high-precision map datasets are particularly. It is important, but the development of high-precision map datasets has just started, and the comparison of common high-precision map datasets for autonomous driving is not complete. In view of the above problems, this paper collects the current mainstream and emerging 7 types of datasets, and briefly introduces their main contents. Quantitative analysis is carried out from the aspects of target recognition, recognition environment, data scale and application field. At the same time, the main contents of mainstream data sets are summarized and analyzed from the aspects of driving scenes, weather conditions, target recognition, data volume, sensor types, etc., applicable scenarios and shortcomings of the dataset.