"Dataset for NmS-TCP"

"This database provides a comprehensive collection of high-fidelity simulation datasets designed for the evaluation of Automated Driving Systems (ADS) and the advancement of Test Case Prioritization (TCP) methodologies. The repository comprises three distinct datasets generated across widely adopted physics simulators\u2014LGSVL, BeamNG.tech, and CARLA. These datasets encompass a broad spectrum of driving conditions and failure modes, ranging from complex urban traffic interactions and edge-case lane-keeping challenges to highly constrained interactive scenario configurations. Providing high-dimensional feature vectors and corresponding safety labels, this database serves as a robust empirical benchmark for research in continuous integration (CI) regression testing, search-based software testing (SBST), and representation learning for early fault discovery in autonomous driving. "

本数据库提供一套高保真综合仿真数据集，旨在支撑自动驾驶系统（Automated Driving Systems, ADS）的性能评估与测试用例优先级排序（Test Case Prioritization, TCP）方法的研究演进。该数据集仓库包含三类独立数据集，均生成自当前主流的三款物理模拟器——LGSVL、BeamNG.tech及CARLA。这些数据集覆盖了丰富的驾驶场景与失效模式，涵盖复杂城市交通交互、极端工况车道保持难题，以及高度约束的交互式场景配置等多种场景类型。本数据库提供高维特征向量及其对应的安全标签，可作为自动驾驶领域中面向持续集成（continuous integration, CI）回归测试、基于搜索的软件测试（search-based software testing, SBST），以及用于早期故障发现的表征学习等方向研究的可靠实证基准。