Ultrasonic Robot Eye for Shape Recognition Employing a Genetic Algorithm

Abstract − This paper describes an ultrasonic robot eye for shape recognition using ultrasound pressure data and a Genetic Algorithm. This type of robot eye has commonly used a Neural Network (NN) for shape recognition. However, a NN perform poorly when lacking learned data. In order to overcome this problem when using a NN, we here attempt to replace the NN with a Genetic Algorithm (GA). Unlike a NN, the GA can recognize shapes without depending on learned data. Experimental results demonstrate that the recognition ratios of the proposed ultrasonic robot eye using the GA are higher than that of a conventional ultrasonic robot eye using a NN. Therefore, it is shown that our robot eye is effective for many robotic applications