Voice Recognition as a Mobile Robot Controller with the Adaptive Neuro-Fuzzy Inference System Method

Abstract

Voice recognition or speech recognition is a biometric technology that has very wide applications, one of which is for simple robot motion control. There are three stages in this research, namely data acquisition, feature extraction, and data classification. For feature extraction, the wavelet transform method is used which can analyze non-stationary and non-linear signals, while for data classification, the Adaptive Neuro-Fuzzy Inference System (Anfis) method is used. The result of data classification is 92.25% and 7.75% error. So, based on the results of the classification accuracy, the robot can be moved via voice commands and to anticipate the error value, the ultrasonic sensor feature is added to the robot as an alternative control.