Extended reality-enabled smart manufacturing in multi-robot welding applications

Abstract

Enhancing the performance of robotic welding automation is crucial to achieve high quality and flexibility in high-variation, low-volume welding production, particularly for small and medium enterprises. However, there is limited scientific understanding of how to achieve optimal welding conditions using smart manufacturing technologies, such as extended reality (XR), in cyber-physical multi-robot welding systems. The objective of this thesis is to establish a smart multi-robot welding procedure which exploits XR models for virtual verification of optimal welding conditions. To achieve this, the thesis first characterises the decision-making inputs, possibilities and challenges of XR in multirobot welding. Numerical XR and digital twin simulations, along with empirical multirobot welding methods, are then used to develop a virtual verification approach for smart multi-robot assembly and welding procedures.

The results of this study highlight the importance of accurately performing pre-welding material-handling tasks, such as component positioning and root gap fit-up, which are critical for enabling high-quality welding conditions. This study demonstrates that component position and root gap can be accurately measured and controlled using smart multi-robot methods. However, contradictory results were observed in optical sensor measurements for root gaps smaller than 0.5 mm and component shapes parallel to the scanning direction, necessitating welding expert verification before proceeding to welding. The application of smart technologies like laser line scanning sensors, digital twins, and XR is suggested to enhance human-machine interaction between welding experts and multi-robot systems. This enables virtual verification and adjustment of root gaps, component positions, and pre-set angles. Further, the results show that angular welding distortions can be mitigated with smart multi-robot pre-setting in T-joints and corner joints, while distortions in single-V butt joints were insignificant.

The study shows that XR-enhanced virtual verification provides immersive quality assurance which can effectively enable near real-time inspection and adjustment in smart multi-robot welding, ultimately improving the setup and material handling of high-variation low-volume robotic welding process cycles.