CARLA Real Traffic Scenarios (CRTS)
收藏资源简介:
CARLA Real Traffic Scenarios (CRTS)是由华沙大学等机构创建的一个包含超过60,000个模拟交通场景的数据集,专注于短期战术驾驶操作。该数据集基于真实世界的交通数据,特别是从NGSIM和openDD数据集中提取的场景,旨在为自动驾驶系统提供训练和测试环境。CRTS通过开放源代码和定制地图,支持使用强化学习算法训练代理。数据集的应用领域包括自动驾驶系统的算法测试和概念验证,以及研究不同输入模式和奖励结构对策略性能的影响。
CARLA Real Traffic Scenarios (CRTS) is a dataset containing over 60,000 simulated traffic scenarios, created by institutions including the University of Warsaw, focusing on short-term tactical driving maneuvers. Based on real-world traffic data, particularly scenarios extracted from the NGSIM and openDD datasets, it is designed to provide training and testing environments for autonomous driving systems. CRTS supports the training of agents with reinforcement learning algorithms through open-source code and custom maps. Its application areas include algorithm testing and proof-of-concept validation for autonomous driving systems, as well as research into the effects of different input modalities and reward structures on policy performance.




