车辆超车轨迹数据集

车辆超车轨迹数据集中人工驾驶数据基于公开轨迹数据集HighD提取而来，以10Hz采样频率对4K无人机视角视频中的车辆轨迹数据进行提取与标定，典型定位误差小于10cm。本数据集以车辆横向位移连续变化的起止点作为超车车辆两次变道起止点，查询超车车辆及其交互车辆的id并获取轨迹数据，基于此提取并整理高速公路路段环境下，约16.5小时交通流数据下的超车事件共20组。此外，使用Matlab对超车场景进行重建与仿真，将超车车辆替换成自动驾驶车辆并输出仿真车辆的轨迹信息，共得到自动驾驶场景下超车案例20组。

The human-driving data in the vehicle overtaking trajectory dataset is extracted from the public HighD trajectory dataset. The vehicle trajectory data in 4K unmanned aerial vehicle (UAV) perspective videos were extracted and calibrated at a sampling frequency of 10 Hz, with a typical positioning error of less than 10 cm. This dataset uses the start and end points of continuous lateral displacement changes of vehicles as the start and end points of the two lane-change maneuvers of the overtaking vehicle, queries the IDs of both the overtaking vehicle and its interacting vehicles, and acquires their corresponding trajectory data. Based on this, 20 groups of overtaking events were extracted and organized from approximately 16.5 hours of traffic flow data collected in highway segment environments. Additionally, the overtaking scenarios were reconstructed and simulated using Matlab, where the human-driven overtaking vehicles were replaced with autonomous vehicles, and the trajectory information of the simulated vehicles was output, yielding a total of 20 groups of overtaking cases in autonomous driving scenarios.

搜集汇总

背景与挑战

背景概述

车辆超车轨迹数据集包含人工驾驶和自动驾驶两种场景下的超车轨迹数据，来源于HighD数据集并经过高精度标定，涵盖高速公路环境下20组超车事件。数据集还提供了Matlab仿真重建的自动驾驶超车案例，适用于自动驾驶交互和人机混驾研究。