Answers are grounded in this paper and cite their sources.
Ultrafast laser manufacturing is increasingly used for precision machining, surface preparation and functional surface engineering, but its nonlinear and material-dependent response makes process modelling and optimisation challenging. This review examines machine learning applications in ultrafast laser manufacturing, with particular emphasis on coatings and surface engineering. Existing approaches are compared across process modelling, parameter optimisation, monitoring and intelligent control, while distinguishing nominal process inputs from physically meaningful descriptors and hybrid physics–ML methods. The literature is most mature in offline prediction and surrogate-based optimisation of machining outcomes such as ablation depth, feature geometry, taper, roughness and material removal rate. In situ optical, acoustic and image-based monitoring has demonstrated process-state recognition, while event-specific closed-loop control and reinforcement-learning-based adaptive pulse positioning have been demonstrated for specific machining tasks. However, general multi-parameter adaptive control remains largely undeveloped. Key limitations include small process-specific datasets, limited independent validation and transferability, and frequent reliance on nominal machine settings. For coating applications, the principal opportunity is to extend machine learning beyond machining geometry toward prediction and control of surface state, coating/interface response and functional performance.
No takes yet. Share an insight, caveat, or question.
Kirk et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: