Hybrid Media

ISAM • 7th International Symposium on Academic Makerspaces

ShatterModel (2023)

Leveraging Proxemics to Design Human-AI Interactions in Adverse Environments

Description

Incorporating sensor-based AI interactions in evolving makerspaces presents challenges, including the need for AI models to adapt to model degradation, data shifts, target changes, and environmental variations. Traditional user interfaces like desktops and GUIs can constrain the natural movement and actions in creative spaces, hindering the realization of a smart makerspace. To achieve this vision, it's crucial to explore how human-AI interactions can be seamlessly integrated into dynamic physical creative environments. In this work, we emphasize the use of spatial analyses to facilitate natural, intuitive, and collaborative interactions. Specifically, in the context of glassblowing, we employ a proxemics-based approach. By studying a glassblower's movements and activity in space, we identify interaction zones. We demonstrate the integration of an AI model, the shatterModel, which predicts glass temperature, into the glassblowing process.