Computer Vision-based Robotic Arm for Object Color, Shape, and Size Detection

Abstract

Various aspects of the human workplace have been influenced by robotics due to its precision and accessibility. Nowadays, industrial activities have become more automated, increasing efficiency while reducing the production time, human labor, and risks involved. With time, electronic technology has advanced, and the ultimate goal of such technological advances is to make robotic systems as human-like as possible. As a result of this blessing of technological advances, robots will perform jobs far more efficiently than humans in challenging situations. In this paper, an automatic computer vision-based robotic gripper has been built that can select and arrange objects to complete various tasks. This study utilizes the image processing methodology of the PixyCMU camera sensor to distinguish multiple objects according to their distinct colors (red, yellow, and green). Next, a preprogrammed command is generated in the robotic arm to pick the item employing Arduino Mega and four MG996R servo motors. Finally, the device releases the object according to its color behind the fixed positions of the robotic arm to a specific place. The proposed system can also detect objects' geometrical shapes (circle, triangle, square, rectangle, pentagon, and star) and sizes (large, medium, and small) by utilizing OpenCV image processing libraries in Python language. Empirical results demonstrate that the designed robotic arm detects colored objects with 80% accuracy. It performs an excellent size and shapes recognition precision in real-time with 100% accuracy.