Shape recognition for transparent objects using ultrasonic sensor array

Ultrasonic sensors are often utilized for perceiving the outside world. In this paper, we propose the construction of a new ultrasonic recognition system for identification of transparent object. The proposed recognition system consists of an ultrasonic transmitter, an ultrasonic receiver, and a recognition unit. The system´s receiver is made up of plural ultrasonic sensor array. The receiver outputs the peak values and the positions of ultrasonic sound pressure waves. Their values and positions are input in to recognition unit, which consists of the neural networks for object identification and the neural networks (NNs) for the measurement of the object´s position. We have designed and built a prototype system for identification and position measurement for polyethylene terephthalate bottle (Pet bottle). We tested this system by attempting to identify and measure the positions of seven kinds of Pet bottle. The experimental results showed perfect identification and an accuracy of about 5 mm.