Audio-visual information integration for humanoid robot gaze shift

A humanoid robot is being designed capable of integrating audio-video signal. To develop a model for sound source localization in humanoid robot. Two USB Sound Card microphones are used for sound acquisition and two cameras are used for object detection. A Raspberry-pi is used as a central unit for image processing and sound source localization. When the video data comes from cameras, it is displayed on the screen in frames, and its image processing is being done in parallel. When the required objects image processing and detection is done, then it came to the tracking process. To track the detected object the centre position is defined and in code the error is defined that the quantity of error is encountered by adjusting the positions of servos. A PWM model had adopted for precise and accurate results. The results shows that when the red object came in front of camera a green rectangle is formed around it and a vertical line a centre, when the object moves the green line moves along the object, due to the movement of green line away from the centre position the servos position is changed to make the green vertical line came to its origin.