Geometry Control in Wire Arc Additive Manufacturing: Challenges, Strategies, and Future Perspectives

Abstract The deployment of autonomous artificial intelligence (AI) within complex physical systems remains a grand challenge. While intelligent manufacturing aims to address this by infusing industrial processes with real-time adaptation, the prevailing approach often involves applying machine learning (ML) to isolated, well-defined sub-problems. This component-wise strategy, however, proves insufficient for processes like Wire Arc Additive Manufacturing (WAAM), where geometric fidelity is governed by deeply coupled, multi-scale thermo-fluid dynamics that defy simple predictive modelling. Here, using WAAM as a foundational case study of a complex cyber-physical system, we provide a systematic review of the intelligent strategies required for end-to-end geometry control. We deconstruct this challenge into a unified framework spanning the full process chain—pre-deposition, in-process, and post-deposition—and map the corresponding control methods onto the critical physical scales of the bead, layer, and part. Our synthesis reveals that the primary limitations are not a lack of algorithmic novelty, but fundamental issues of model generalisation from sparse, high-cost data and the integration of heterogeneous physical priors. This analysis points towards a necessary paradigm shift: moving from static, data-driven prediction towards dynamic, autonomous control architectures. We argue that realising the next generation of intelligent manufacturing requires a focus on integrated systems, such as physics-informed digital twins and multi-sensor fusion platforms, that can holistically manage process complexity in real-time.