This review summarizes the research progress in socially aware navigation for social service robots, focusing on the establishment of human-robot interaction models, the application of subjective emotion state recognition in socially aware navigation, and the implementation of socially aware navigation under various learning frameworks. First, it explains the construction of navigation based on the Social Force Model and trajectory prediction models along with their respective applicable environments. Next, the paper introduces how different emotion recognition technologies accurately perceive human emotions to assist in the realization of socially aware navigation. Finally, it comprehensively analyzes the advantages and challenges of using supervised learning, reinforcement learning, and preference learning in achieving socially aware navigation.