SocNav1

数据集SocNav1由阿斯顿大学计算机科学与工程学院创建，旨在评估机器人在社交导航中可能引起人类不适的程度。该数据集包含9280个样本，涵盖了多种社交导航场景，如机器人与人类的空间交互。数据集的创建过程涉及随机生成场景和多个人类标注者的参与，确保了数据的多样性和主观性。SocNav1适用于训练和评估非欧几里得机器学习算法，如图形神经网络，主要应用于机器人导航系统，以提高机器人在人类环境中的社交接受度。

The SocNav1 dataset was developed by the School of Computer Science and Engineering at Aston University, with the objective of assessing the extent to which robots may induce human discomfort during social navigation. This dataset comprises 9280 samples covering diverse social navigation scenarios, including spatial interactions between robots and humans. The creation of the dataset involves randomly generating scenarios and engaging multiple human annotators, thus ensuring the data's diversity and subjectivity. SocNav1 is applicable for training and evaluating non-Euclidean machine learning algorithms such as graph neural networks, and is primarily utilized in robotic navigation systems to enhance the social acceptability of robots in human-centric environments.