Development of Robotic Arm Control System Using Computational Vision

Abstract

Industrial robots are present in most of today's industries. Their benefits to industrial production make them the main research's target and constant development. However, are limited to sequential and repetitive movements without any decision making. Therefore, one of the current researches and development strands is the robot's integration with computer vision. Which seeks to enable them to see as humans, and thus, make them even more useful, efficient and independent. In this context, a PID control system was developed for a robotic arm. It is integrated to a computer vision system using Kinect\textregistered, in order to capture objects - within 6 possible positions - according to their color. For that, an Arduino MEGA 2560 was used for control system, and Software was developed in Processing programming language for integration, systems operation, and 6-position definition. The SimpleOpenNI and OpenCV for Processing libraries were used to work with Kinect\textregistered, and computer vision, respectively. Finally, the results were satisfactory, and the lowest percentage of catches made successfully among all positions was 83.33%.