Abstract and Concrete Scenarios from CARLA Simulator

该数据集包含了432个由144个抽象场景根据自动驾驶系统验证的覆盖标准，使用CARLA模拟器生成的具体驾驶场景。这些场景旨在挑战 ego 代理的感知和控制系统，其中非 ego 代理的初始位置多样化，并具备在未来评估中改变天气条件的能力。具体场景规模为从144个抽象场景派生出的432个场景，任务是对自动驾驶系统进行验证的场景生成与模拟。

This dataset contains 432 concrete driving scenarios generated using the CARLA simulator. These scenarios are derived from 144 abstract scenarios in line with the coverage criteria for autonomous driving system validation, and are designed to challenge the perception and control systems of the ego agent. The initial positions of non-ego agents in these scenarios are diversified, and the scenarios support modifying weather conditions for future evaluations. The dataset consists of exactly 432 concrete scenarios derived from the 144 abstract ones, with its core task being scenario generation and simulation for autonomous driving system validation.