Abstract
This paper presents a technique for a speech recognizer used to control the motion of an intelligent automated mobile robot. The aim is to interact with the mobile robot using natural and direct communication techniques. The voice is processed to get proper and safe movement of a mobile robot and satisfying high recognition rate. Features are extracted from speech signal using Mel Frequency Cepstral Coefficients (MFCC). To realize feature matching, an efficient Dynamic Time Warping (DTW)-based speech recognition system is presented which is applicable for isolated words of Arabic language. The tested words are compared to a trained database using this DTW algorithm. On the other side, the mobile robot is designed with two servo motors as driving actuators. These actuators are controlled by L298 motor driver circuit. The control algorithm is programmed and downloaded into a PIC18F45K22 microcontroller which is interfaced to a USB port of a 10" notebook computer. The robot proves a capability of understanding the full meaning of the five Arabic speech commands that steer it forward, backward, right, left, or stop.
