Nonintrusive Behavioral Sensing and Analytics for Supporting Human-Centered Building Energy Efficiency

Occupant behavior is a significant factor affecting building energy use and occupant comfort. Capturing occupant behavior, therefore, holds great promise toward human-centered building energy efficiency. However, existing methods for behavioral sensing and analytics are mainly based on intrusive sensing techniques (e.g., visual and acoustic sensing), which are known for infringing occupant privacy and have limited applicability. As such, the authors propose a novel nonintrusive approach for behavioral sensing and analytics. It uses (1) environmental chemical sensing to detect air composition changes caused by occupant behaviors, and (2) machine learning to learn from the air data to extract behavior information (e.g., occupancy and behavior type). This paper focuses on presenting the proposed approach and its evaluation in extracting occupancy information. The preliminary experimental results show that the proposed approach achieved an accuracy of 64.59% in sensing and analyzing occupancy, indicating the potential of the nonintrusive approach in supporting human-centered energy efficiency.