A Data-Driven Framework for Automatic Robotic Welding Parametrization

December 12, 2025

Robotic welding has become a cornerstone of modern manufacturing, delivering significant gains in productivity, cost efficiency, and workplace safety. However, ensuring weld quality remains a significant challenge, as it must consistently meet or exceed the standards achieved through manual welding. A major difficulty lies in defining the process parameters such as current, voltage, wire feed speed, and gas flow rate, which are still predominantly set based on the empirical knowledge of experienced welders. This reliance on intuition often leads to inefficiencies, most notably: overwelding, where excessive material is deposited; suboptimal travel speed, which lowers productivity and raises costs; and lack of repeatability, which undermines process consistency and weld reliability. To address these issues, this thesis proposes a data-driven framework for the automatic parameterization of robotic welding. The approach relies on the construction of extensive datasets grounded in industrial expertise and the deployment of machine learning models capable of delivering consistent, auditable, and interpretable parameter recommendations. By not only predicting but also clarifying how variables influence outcomes, the framework enhances both performance and transparency. The results show that the proposed system can replicate and, in many cases, surpass the decisions of experienced technicians, effectively reducing overwelding and increasing process speed without compromising weld integrity. In doing so, this work contributes to more efficient, reliable, and scalable welding practices, supporting the transition toward increasingly automated and data-driven industrial environments.

Authors in the community:

• 

  G

  Gonçalo Nuno Marques Cardoso de Araújo

  ist1110822

Supervisors of this institution:

• 

  L

  Luís Manuel Marques Custódio

  ist13279

electrical-engineering-electronic-engineering-information-engineering - Electrical engineering, electronic engineering, information engineering

eng - English

Instituto Superior Técnico