Occupancy Detection and Localization by Monitoring Nonlinear Energy Flow of a Shuttered Passive Infrared Sensor Academic Article uri icon

abstract

  • © 2001-2012 IEEE. Passive infrared (PIR) sensors are the current choice for individual presence detection in buildings. A major problem is that PIR sensors only detect moving individuals, which often provides false negative detections, resulting in uncomfortable lighting/temperature swings, short lifetime of the equipment, and waste of energy. In this paper, a rotating shutter is introduced to the rotationally shuttered PIR (Ro-PIR) sensor to explore its functionalities beyond motion sensing, including stationary individual localization, tracking, and facing direction detection. More specifically, by monitoring and analyzing polarity-phase of the nonlinear infrared energy flow induced by the rotating shutter, Ro-PIR is capable of zone-level localization for stationary individuals. By further analyzing the duty cycle of the output, occupancy facing direction can be predicted. Two theoretical models are created to identify facing directions (front/back or side) by analyzing three infrared radiation covering configurations, shaped by shutter movement. Combining sequence of covering configurations represents the induced polarity-phase output voltage, illustrating the occupancy localization information during one segmented scanning period. Parametric analysis, and empirical studies are performed to obtain the optimal setting of the shutter in terms of its width and shuttering period. Experimental results reveal 100% presence accuracy for stationary detection for up to 3 m, and moving detection for up to 8 m. Zone-level localization can reach 98% accuracy by applying machine learning classifier using two features extracted from polarity-phase signals. Experimental results also show less than 0.44 m root mean square error for tracking, and over 83% in detecting front/back or side facing direction.

author list (cited authors)

  • Wu, L., Wang, Y. a., & Liu, H.

citation count

  • 13

publication date

  • January 2018