An ANN Based Approach to Calculate Robotic Fingers Positions

In the certain applications, a human hand like robotic hand is needed to do the operations alike human can do. The fingers in the human hand have the capability to bend on different angles and they can apply the force or can hold an object. Current focus of our research is on controlling the robot fingers using vision based techniques, which have the joints in finger like human hand. This paper describes our approach of robotic fingers positions calculation using supervised Artificial Neural Network. User has to show a gesture to the system without any limitation or restriction. Hand cropping gives the region of interest and made the algorithm faster by processing smaller image. The gesture would be extracted from the input image and after detecting fingertips in the region of interest, fingers bending angles would be calculated using ANN.