The rapid advancement in computer vision and image processing has allowed machines to be more perceptive of their environments than ever before. Many industries are now looking to rely on vision-actuated robots for complex tasks that require high speed and precision. Subsequently, the goal of this project is to design a robotic arm that is able to react to its environment using computer vision. In order to prove the efficiency of this system, it will be applied on an athletic robotic arm that can play Ping-Pong against a human player.
The project comprises of various subsystems that can be summarized as follows: A stereo vision and ball detection algorithm implemented on an Xbox Kinect and ZED stereovision camera. The algorithm detects the 3D position of the ball and supplies it to a path prediction algorithm that uses aerodynamic and rebound models of the ball to predict the position of the ball at the other end of the table and, consequently, where the robot needs to be in order to hit the ball. A 4-DOF robotic arm “Dobot” is mounted on a lead screw mechanism actuated by an AC induction motor. The mechanism moves the robot to the required position where it successfully hits the ball.
This was a graduation project made by: Ahmed Tarek Fahmy Hesham Mohamed Fadl Hossam SamirArafat Moumen Yasser
Supervised By:
Dr. Mostafa Abdel Galil Associate Professor Head Of Electrical And Control Engineering Department, AAST
Dr. Mohamed Tamazin Assistant Professor Electronics and Communication Engineering Department, AAST