A computer-vision-based HVAC control system

Computer vision is an approach that understands the object or the scene and its three-dimensional properties from one or a sequence of images of a moving or stationary object or a scene, taken by a monocular or polynocular moving or stationary observers. The application of computer vision in building systems is very new. For conventional heating, ventilating, and air-conditioning (HVAC) control, a well-tuned PID controller has been considered adequate. However, when transients, such as a sudden inrush or an evacuation of residents, etc., occur, the system usually takes minutes for it to be restored back to the steady state. By employing computer vision that can estimate the number of residents within an air-conditioned space, it is possible to identify such changes within seconds. This paper describes the hardware and software developments associated with the computer vision system and two modes of HVAC control, namely, step control and human tracking, which have been established to assist the well-tuned PID controller for optimal HVAC control. A computer simulation on an existing air-conditioned hall has been done to show the improvement in both the response rate and the total energy savings during a prolonged period of operation.