Enhancing Open Conversations Using Visual Percepts From a Socially Assistive Robot Preliminary Assessment
摘要
Being able to converse with a Socially Assistive Robot can be a way to help older adults participate socially and stay included in the community. To make a robot more engaging, it can use perception of its surrounding for situational grounding of the conversations. The integration of percepts with a conversational agent is influenced by real-time processing capabilities of the robot’s sensory information, which also have an effect on the human-robot interaction experience. This paper presents the integration of OpenAI’s GPT-3.5 Turbo on T-Top, a tabletop robot with onboard visual and audio recognition capabilities. As a preliminary step before using T-Top in experimentations with older adults, characterization of this integration is conducted to assess the robot’s capabilities with reference to response time, conversation cueing and interaction modalities.