| About this Abstract |
| Meeting |
2022 AWS Professional Program
|
| Symposium
|
2022 AWS Professional Program
|
| Presentation Title |
An AI-based Vision Methodology for Self-guided Seam Tracking in Gas Metal Arc Welding |
| Author(s) |
Mahyar Asadi, Amin Ghasemazar |
| On-Site Speaker (Planned) |
Mahyar Asadi |
| Abstract Scope |
Autonomy is fast-growing in welding, and artificial intelligence is integral to the recent automation trend. Smart vision systems are popular front lines of data collection for many intelligent welding architects, primarily if the process consists of arc welding. These systems are no longer explicitly programmed like robots to deliver a perfect task; instead, they are architectured to learn and improve over time. In other words, they can start with the current state of a welding process and continuously improve the process to a better state. Seam tracking is among the primary demand in the industry for welding control during fabrication. A self-guided machine can live-detect and optimize relative lateral and perpendicular distances between the torch and the seam centerline. This work focuses on a vision-based development of an AI model for real-time semi-automatic estimation of those critical distances and deviations for seam tracking and distance control during Gas Metal Arc Welding (GMAW) for fillet welds. This talk presents our journey from vision to smart eyes and eventually to universal mind lattice collecting tacit welding experience globally. The presentation starts with the intense market demands for smart welding systems to retain a uniform quality along the weld. We cover auto-learning from good and bad weld practice overtime to avoid repetitive mistakes and explore opportunities for a better weld. The talk shifts to an existing industry 4.0 platform that pervasively spreads out the learning domain to comprise information beyond a single welding machine. We present a platform that enriches the smarting level of our local system to dive deep into massive data collected around the globe and get actionable insights. We use real industrial examples from our project’s collection. |
| Proceedings Inclusion? |
Undecided |