State estimation based on a spiking neural network using ultrasonic oscillosensors

Recently, the social role of the nursing health facilities increases more and more as Japan becomes an aged society. There has been a problem on accidents such as “falling” and “moving away from facilities” in such facilities owing to the overload to the human monitoring and the privacy protection. This sensor can detect the state such as turning over in bed and getting up on the bed. The parameters should be updated automatically according to the monitoring results. In the ultrasonic oscillosensor, two thresholds are used for the state estimation of the human behavior on the bed. Therefore, these thresholds should be automatically updated according to some criteria. In this paper, we apply a fuzzy spiking neural network for the state estimation of the human behaviors on the bed, and a steady-state genetic algorithm for updating the thresholds based on the correctness of the state estimation. Next, we discuss the effectiveness of the proposed method through several numerical experiments.