Socially assistive robots (SARs) are increasingly recognized for their potential in helping older adults age in place. Effectively meeting the diverse needs of older adults requires a proper classification of SARs' functions. However, existing function categories are primarily proposed from the perspective of researchers, rarely from older adults themselves. This study addresses this gap by employing a user-centered design approach to explore how older adults classify SARs' functions. Using the open card sorting technique, we first identified 30 functions supporting the daily tasks of older adults. Subsequently, twenty-five older adults aged 60 to 86 with diverse educational backgrounds participated in an iterative function categorization. The initial sorting ranged from 2 to 10 groups but generally centered around 4 to 6. Through multidimensional scaling and hierarchical clustering, we revealed a hierarchical structure of SARs' functions. Which identified four main categories: "Home Management," "Smart Healthcare," "Entertainment and Communication," and "Monitoring and Maintenance," with their sub-categories and super-categories. These findings provide valuable insights for designing SARs that better align with older adults' needs.

Keywords: Socially assistive robots; function category; older adults; open card sorting; user-centered design.