An Event-Driven Interactive Platform for Non-intrusive Thermal Comfort Personalization

The development of occupant-responsive and energy-efficient indoor environments relies on accurate, real-time measurement of thermal comfort. Conventional methods, such as single-point monitoring or wearable sensors, face limitations in scalability and user acceptance. This paper presents a non-intrusive interactive platform for thermal comfort management that integrates distributed environmental sensing, thermal imaging and an online visualization interface within an edge-cloud hybrid architecture. The system unobtrusively captures environmental and physiological data, especially skin temperature, without body-mounted devices. Unlike traditional thermostats, this system integrates explicit user feedback into the control loop. To address human interaction with indoor environments, a key barrier to smart building adoption, the platform uses a privacy-by-hardware design. Skin temperature is sensed only during user-initiated comfort voting events, limiting data collection to brief, intentional interactions. This links physiological state directly to subjective preference without continuous monitoring. A Raspberry Pi edge node performs local computation for low-latency response and privacy protection, while the cloud layer supports analytics and a web-based visualization dashboard reinforcing a human-in-the-loop design. A lab-scale deployment validated the platform’s technical calibration and performance, confirming that the proposed platform is deployable for personalized comfort modeling. The framework balances immediate responsiveness with scalable data management, while the visualization interface enhances interpretability. As a result, the platform supports occupant-centered building operation and lays the groundwork for predictive comfort modeling in future smart building applications.