Data for: Human-like pedestrian models for automated driving systems

This dataset contains the evaluation data supporting the study of a computational rationality based pedestrian model for automated-driving simulation. The pedestrian model is formulated as a partially observable Markov decision process and trained with reinforcement learning in the CARLA simulator. The uploaded Excel file includes data used to evaluate crossing outcomes, human-like crossing behaviors, map transfer, and adaptation to different traffic environments.本数据集包含支撑面向自动驾驶仿真的基于计算理性的行人模型研究的评测数据。该行人模型被形式化为部分可观察马尔可夫决策过程（partially observable Markov decision process），并在CARLA仿真器中通过强化学习完成训练。本次上传的Excel文件包含用于评测行人过街结果、类人过街行为、地图迁移能力以及对不同交通环境的适应能力的相关数据。