德清自动驾驶仿真场景库数据

为自动驾驶企业提供仿真测试服务基于德清路侧采集真实道路交通数据,同时将政府开放数据与社会数据进行融合利用，通过对高精路网的解析完成到OpenDrive静态路网的转化，并将交通灯的位置挂载到OpenDrive路网中。然后对车路协同的数据进行交通流的还原（视数据质量决定直接用结构化数据还是用感知数据进行算法还原），按事件类型对事件中涉及的人、车、物、路、环境等要素进行提取还原，从而实现对场景动态交通流OpenScenario的转化。最后通过接入ROS、OSI等插件/接口实现对V2X消息的接入，完成对自动驾驶仿真场景的构建。
可以基于已有的数据，结合交通流仿真算法能力对场景做几方面的泛化：
道路结构泛化 – 更改道路结构（如增减车道）
道路施工泛化 – 封闭部分路段的车道或半幅道路
信号灯配时泛化 – 更改信号灯的配时方案（或接入动态配时数据）
V2X泛化 – 增减场景中的V2X预警信息
也支持通过ROS或OSI等接口接入激光雷达、摄像头、毫米波雷达等传感器数据，以实现对传感器的仿真。

This dataset provides simulation testing services for autonomous driving enterprises. It is built upon real road traffic data collected by roadside facilities in Deqing, and integrates government open data and social data for joint utilization. First, it parses the high-precision road network and converts it into the OpenDrive static road network format, while mounting the positions of traffic lights onto the OpenDrive road network.

Then, it restores the traffic flow from vehicle-road collaborative data: the specific restoration method, i.e., directly using structured data or performing algorithmic restoration via perception data, depends on the quality of the collected data. It extracts and restores elements including people, vehicles, objects, roads and environments involved in traffic events according to event types, thereby realizing the conversion of dynamic traffic flow scenarios into the OpenScenario format.

Finally, it accesses V2X messages by connecting plugins or interfaces such as ROS and OSI, and completes the construction of autonomous driving simulation scenarios.

Based on the existing data and combined with traffic flow simulation algorithm capabilities, the scenarios can be generalized in the following aspects:
- Road structure generalization: Modify the road structure (e.g., add or remove lanes)
- Road construction generalization: Close partial road lanes or half of the road section
- Signal light timing generalization: Modify the signal light timing scheme (or access dynamic timing data)
- V2X generalization: Add or remove V2X warning information in the scenario

It also supports accessing sensor data such as LiDAR, cameras and millimeter-wave radars via interfaces like ROS or OSI to realize sensor simulation.

搜集汇总

数据集介绍

特点

德清自动驾驶仿真场景库数据是一个由德清县车网智联产业发展有限公司提供的企业数据集，包含101条数据，用于自动驾驶仿真测试。数据集通过融合多种数据源和技术，支持构建复杂的自动驾驶仿真场景，并能够接入多种传感器数据以增强仿真的真实性和准确性。