Development of a real-time automated mobile robotic welding system in construction

Welding plays a pivotal role in the construction industry. However, the shortage of skilled welders has been an ongoing challenge. In addition, welding in construction requires exceptionally high welding quality in many cases since a minor defect can lead to serious safety issues. Therefore, automated robotic welding is necessary to replace skilled welders and improve welding quality. The primary objective of this study is to develop a real-time automated mobile robotic welding system. In addition, an image-based quality inspection that can be used for welding inspection is developed. Three studies are proposed to achieve the first objective: 1) development of an automatic system that detects a joint between two metal pieces to be welded using a camera and performs initial point alignment between the robot’s end-effector and the endpoint of the joint; 2) development of an automatic robotic welding system through integration of a welding instrument into the robotic arm for welding, which would require sophisticated control of the welding instrument and the robotic arm; and 3) development of an autonomous robotic system by integrating the automated robotic welding system with an unmanned ground vehicle (UGV). Since the proposed robot will be equipped with visual sensors (i.e., cameras and a laser scanner), a computer vision-based inspection system that uses these visual sensors would expedite weld quality assessment. Therefore, the second objective is to develop a computer vision-based quality inspection system. This system uses images from a camera as input, produces an as-built model, and allows users to take direct measurements from the model. These four studies are validated through multiple case studies. Finally, this thesis concludes with limitations and future directions for developing fully autonomous welding robots.