Design, analysis, and model fabrication of an industrial manipulator for quality control purposes using computer vision as a sorting mechanism

Muthair Saeed; Haris Imran; Liaquat Ali Khan; Kamran Nazir

doi:10.22581/muet1982.2814

Muthair Saeed Mechanical Engineering Department, National University of Technology, I-12, Islamabad, Pakistan
Haris Imran Mechanical Engineering Department, National University of Technology, I-12, Islamabad, Pakistan
Liaquat Ali Khan Marine Engineering Department, Military Technological College Muscat, Oman
Kamran Nazir Marine Engineering Department, Military Technological College Muscat, Oman

DOI: https://doi.org/10.22581/muet1982.2814

Abstract

The industrial manipulator uses complex control systems and requires specialist equipment for object detection and sorting. These are expensive and complicated. Installing and maintaining these manipulators require professional personnel and equipment. The manipulator design in this project is fully modular, easier to automate, and inexpensive, which is designed to enable industrial automation in developing countries. Using Topology in conjunction with Fusion 360 and ANSYS material consumption is reduced by up to 30% while keeping the material strength and stress constant. 3D printing parts help in reducing manufacturing complexity, time, and material consumption by reducing the overall manufacturing cost. With a Workspace of two feet and an operating temperature range from 10-60 °C, it is designed to deliver 3 DOF. To prevent the system from overheating, an effective cooling system for the controller and driver, is also developed using CFD analysis. The manipulator is linked with several sensors and a camera, to provide full autonomous control. As a result, it can be utilized in various industrial applications for precise and quick output. In this project, a manipulator is designed to sort objects as they come off the conveyor and pick and place them at the design location. A conveyor is intended for item movement, both manipulator and conveyor are controlled by a microcontroller and a PC via Computer vision. Initially sorting of objects is performed through both color sensors. Then, using computer vision, an algorithm is trained to identify any defective item among the bunch using a camera and place it at the designated location. This computer vision algorithm can detect an object at a range of 3 feet and provides proper control with an accuracy of 90-95%, depending on the amount of information about the object collected and analyzed. This algorithm is also adaptable, as it can identify any object once it has been trained.