AIKitchen: A Rapid Prototyping Framework for Exploring Dynamic In-context Behavior of Smart Robotic Environments

Abstract

Smart environments allow for rich, context-specific interactions. Additionally, the introduction of agentic AI gives UX designers new ways to create flexible, human-centered interactions. However, given the complexity of these environments, building them remains challenging. Classical methods for mitigating the complexity of prototyping behavior, e.g., Wizard-of-Oz prototyping, cannot adequately represent the flexibility and adaptivity necessary for such environments. We present AIKitchen, an open-source framework for prototyping user experiences in agentic, context-driven environments. This framework allows designers to rapidly create, evaluate, and iterate complex interaction scenarios. We demonstrate this approach in the context of a smart kitchen, which integrates environmental sensors, smart-home appliances, and robotic actuators. Through a series of examples, we show how designers can integrate various contextual sensors, actuators, and agents, orchestrate system behavior, and explore different interaction scenarios. This makes the framework valuable for designing context-aware, room-scale experiences and exploring what our future in smart environments might look like.