SafetyBuilder: An AR-based Framework for In-situ AI-assisted Creation of Child Safety Protection

Children often encounter safety hazards, such as sharp table corners or exposed electrical outlets, which are out of their radar. Childcare providers may lack awareness of these hazards and the technical expertise required to design effective protective solutions. To address this, we propose SafetyBuilder, an AR-based framework that enables childcare providers to detect safety hazards in the environment and create customized protective devices for 3D printing. The framework comprises three core components: real-time environmental hazard detection, AI-assisted suggestions for protective measures, and in-situ customization of 3D printable protective devices. We then evaluated SafetyBuilder via design workshops involving 10 participants and user testing of a proof-of-concept prototype system with nine participants. The results showed that the framework effectively supports users in identifying potential hazards, creating customized protective devices, and improving their confidence in managing child safety risks.