IUPUI-CSRC Pedestrian Situated Intent (PSI)数据集

IUPUI-CSRC Pedestrian Situated Intent (PSI)数据集是由印第安纳大学-普渡大学印第安纳波利斯分校开发的，专注于行人行为预测的基准数据集。该数据集包含110个代表性的行人过街案例，支持车辆与行人之间的有效交互，特别是通过创新的动态意图变化标签和基于文本的驾驶员推理过程解释。数据集的创建过程涉及从自然驾驶研究中抽样，确保了数据的多维性和代表性。PSI数据集的应用领域主要集中在提高自动驾驶车辆的社会智能，特别是在预测行人行为和优化决策制定方面，以解决城市交通中的复杂交互问题。

The IUPUI-CSRC Pedestrian Situated Intent (PSI) Dataset is a benchmark dataset developed by Indiana University–Purdue University Indianapolis (IUPUI) for pedestrian behavior prediction. It contains 110 representative pedestrian crossing scenarios that enable effective vehicle-pedestrian interactions, particularly through innovative dynamic intent variation labels and text-based explanations of drivers' reasoning processes. The dataset was constructed via sampling from naturalistic driving studies, ensuring its multidimensionality and representativeness. Primary applications of the PSI Dataset center on enhancing the social intelligence of autonomous vehicles, specifically for pedestrian behavior prediction and decision-making optimization to resolve complex interaction issues in urban traffic.