Voice Command Recognition for Movement Control of a 4-DoF Robot Arm

Abstract

Robots are widely used in industry. Robots generally have a control system or intelligence embedded in the processor. The robots consist of mobile mode, manipulator, and their combination. Mobile robots usually use wheels, and manipulator robots have limited degrees of freedom. Both have their respective advantages. Mobile robots are widely applied to environments with flat floor surfaces. The manipulator robots are applied to a static environment to produce, print, and cut material. In this study, the robot arm 4 Degree of Freedom (DoF) is integrated with a computer. The computer controls the whole system, where the operator can control the Robot based on voice commands. The operator's voice is one person only with different intonations. Voice command recognition uses the Mel-Frequency Cepstral Coefficients (MFCC) and Artificial Neural Networks (ANN) methods. The MFCC and ANN programs are processed in the computer, and the program output is sent to the Robot via serial communication. There are nine types of voice commands with different MFCC patterns. ANN training data for each command is 10 data, so the total becomes 90. In this experiment, the Robot can move according to voice commands given by the operator. Tests for each voice command are ten experiments, so the total experiment is 90 times with a success rate of 94%. There is only one operator, and experiments have not yet been carried out with the voices of several operators. The error occurred because there were several similar patterns during system testing.

Keywords

• artificial intelligence
• robotic arm
• voice command