Implementering av maskinsyn for montering av fasadeplater

Abstract

The construction industry has seen a surge in efficiency and automation solutions in recent years. MacGregor, in cooperation with the University of Agder want to use crane and lifting technology to streamline construction processes. Facade panels are widely used due to ease of installation and require minimal maintenance. The process associated with installation has an automation potential. This thesis looks at the implementation of machine vision for installation of facade panels. Installation is done using a ABB IRB6600 industrial robot and a vacuum tool. The machine vision is based on OpenCV and C++. The main components of the control system are a Beckhoff CX2040 Embedded-PC, Basler acA2500 camera and a Techspec 4mm lens. The machine vision is developed using OpenCV algorithms that locate and estimate the position of the facade panels. The machine vision is also used to install the facade panels according other objects on the wall. The challenges were the calibration of lens and camera to remove geometric distortion. Repeated testing shows good performance in the center of the camera and less sufficient results elsewhere in the camera's field of view. The Beckhoff CX2040 is used as the main controller in the system. It controls input and output signals, system logic, machine vision and HMI. The controller turns out to handle simple machine vision tasks and worked excellent during the dissertation period.