SocNav2

该数据集名为SocNav2，旨在学习和评估在社会导航规范中考虑机器人与人类速度和轨迹的估计函数。每个样本包含在4秒时间间隔内拍摄的35个“快照”，并对两个与社会导航相关的陈述进行评分。该数据集规模达到53,600个评分样本，其任务是在考虑人类意识的导航场景中建模机器人的干扰。

This dataset, named SocNav2, aims to learn and evaluate estimation functions that take into account robot and human velocities and trajectories in social navigation norms. Each sample includes 35 "snapshots" captured over a 4-second time interval, and assigns scores to two statements related to social navigation. The dataset consists of 53,600 scored samples in total, with the core task of modeling robot intrusiveness in human-aware navigation scenarios.