Gesture based control for mitsubishi industrial robots

Hand gesture recognition is a study of recognizing different hand gestures shown by a person through video or image processing. Study of gesture recognition for human robot interaction is important in developing a natural method to interact with robots. This paper proposes a gesture recognition model to interact with industrial robots. The gesture recognition is done through an RGB camera and a graphical user interface created in windows form. Gesture recognition using RGB camera depends on skin detection. A good skin detection system allows the program to compare the image acquired with the gestures data. The main problem in skin detection process is the influence of light illumination. In this research, methods to overcome this problem for different skin colors are researched and proposed. Combinations of YCbCr, HSV color spaces filters and background subtractor for different skin colors are implemented in this paper. Experiment from this research gave a minimum recognition rate of 95% for six different static gesture on a simple background. The significance of this research is to allow implementation of gesture recognition on industrial robots without heavy investment and heavy computation.