DishAgent: Enhancing Dining Experiences through LLM-Based Smart Dishes

Abstract

With the rapid advancement of smart technologies, there is an increasing demand to enhance everyday experiences, including dining. Recent Human-Computer Interaction (HCI) research has begun to emphasize the aesthetic, affective, sensual, and sociocultural qualities of directly interacting with food. However, these technologies are often constrained by the material properties of food, limiting their everyday applicability. This research introduces DishAgent, an innovative device equipped with a Large Language Model (LLM)-based smart dish and a swarm robotics system. DishAgent adapts to various dining scenarios by generating appropriate conversational contexts and coordinating the action commands of swarm robots, thereby enhancing the dining experience through real-time interaction. This paper explores the applications of DishAgent in intelligent dining guidance, dietary behavior intervention, food information query and social companionship, aiming to fill the critical gap in current technologies for simply and intuitively enhancing dining experiences.