S. Imahara, K. Kubota, T. Kumazawa, and Y. Hondo (Japan)
Smart home, sensor network, motion sensor and particle filter.
We propose an algorithm to count the number of people in rooms of a smart home in which IR-based motion sensors are installed. Our algorithm is based on a particle filter that is one of the general state-space models. Our algorithm has two features. The number of people can be accurately detected when two sensors detect a person in their duplicated detection area. Response specification of IR-sensor and its observation noise are also considered. The proposed algorithm is evaluated using actual data collected in an experimental house and artificial data generated from actual data by adding noise. Experimental results show that our algorithm achieves 10% - 40% improvement in its accuracy compared with conventional methods.
Important Links:
Go Back
