A Non-Intrusive Ultrasound-Based Sensing Technique for Activity Detection: Proof of Concept Towards Optimized Personalized Comfort

Identifying the activity level is pivotal to take metabolic rate into account when assessing comfort in indoor environments. This work addresses two-fold aims. Firstly, the feasibility of employing multi-ultrasonic sensors on a multidomain monitoring platform for personalized comfort is investigated. Secondly, a data calibration and filtering methodology aimed at extracting features to detect office activities is proposed using temporal markers. A living environment for office use was considered and volunteer healthy subjects were monitored during 5 typical office activities. Results confirm the feasibility of integrating ultrasonic sensors in a monitoring platform to capture meaningful movement patterns to discern various office activities. In addition, results show that activity discrimination has an impact of 76% on the estimated Predictive Mean Vote (PMV) values. This information can be integrated in personal comfort models (PCMs) to optimize the occupants' well-being as well as thermoregulation of the built environment and, hence, the building energy consumption.