The paper aims to comprehensively analyze the research situation in the field of service robot interaction in China and explore the research innovation directions of Human-Robot Interaction (HRI). Based on the bibliometric analysis of 309 articles (2001-2020) related to the theme of "Service Robot + Interaction" in the CNKI database, relevant knowledge maps are developed, such as CiteSpace Time Zone Evolution Map, VOSviewer Density Hotspot Map and BibExcel Strategic Coordinate Map, so that the state and trends of HRI research in China can be understood from three aspects:research evolution trend, research hotspots, and research quality. This study shows that, against the background of the trend of new role attributes and deep interaction needs, compared to HRI research and applications conducted at the level of "computational intelligence" such as positioning, planning and recognition, the current HRI research in China lacks the design of "Human-centered AI", such as the lack of attention to user experience, multimodal natural interaction, and social interaction, which has become a problematic point affecting service robot application innovation and industrial development. It is paramount to consider how HRI can and should be shaped by human-centered design.