Machine Vision for Industrial Robotic Manipulator using Raspberry Pi

This article discusses the development of a cylindrical object detection system that assists an industrial robotic arm using accessible technologies, for example, the embedded Raspberry Pi system, the PiCamera module and OpenCV, an open source library for development of Computer Vision System and Digital Image Processing. Distortions inherent in the digital camera lens were corrected and the homography between the robot plane and the image plane was established by applying the Direct Linear Transform (DLT) method. Experimental results indicate that a robot manipulator equipped with the proposed detection system has a maximum error in x coordinate of 2.187 mm and in y of 2.65 mm that allows the automatic capture of the target pieces.