Desire-Driven Reasoning for Personal Care Robots

Abstract

In this paper, in order to reduce the burden of caring for individuals who are bedridden, we present a method of reasoning about the objects and actions that could satisfy their physiological needs. First, we introduce a method of representing knowledge about everyday objects in terms of their properties and functions. Based on this representation, we then propose a desire-driven reasoning approach that bridges the gap between physiological desires and robot actions. This can also deal with issues caused by uncontrolled domains, including incomplete knowledge and dynamic environments. Finally, we evaluate the proposed method by applying our newly developed KUT-PCR personal care robot to real household scenarios.