Applications of classical and quantum machine learning in manufacturing: predictive maintenance, scheduling and tribology

• Review of classical and quantum ML for predictive maintenance and tribology. • Identifies key data, reproducibility, and model generalization challenges. • Emphasizes hybrid, physics-informed ML for interpretable manufacturing AI. • Shows QML advantages in small-data and complex optimization tasks. • Proposes roadmap for scalable and trustworthy AI in manufacturing The rapid digitalization of manufacturing under Industry 4.0 has made machine learning (ML) and deep learning (DL) essential for predictive maintenance, defect detection, scheduling, and process optimization. Recently, quantum machine learning (QML) has emerged as a complementary paradigm to address small-data and high-complexity challenges beyond classical methods. This review summarizes advances in classical and quantum ML with applications in predictive maintenance, scheduling optimization, and tribology—where the multiscale nature of friction and wear reveals the limits of existing approaches. Persistent barriers include heterogeneous and imbalanced datasets, reproducibility gaps, limited generalization, and integration challenges involving latency, interpretability, and lifecycle governance. To overcome these, three priorities are identified: (1) establishing standardized, open datasets for reproducibility and benchmarking; (2) developing hybrid, physics-informed ML frameworks to enhance interpretability and transferability; and (3) applying QML pragmatically to niche, high-impact problems such as sparse-data wear prediction and combinatorial scheduling. Integrating robust classical ML with physics-based modeling and targeted quantum modules can accelerate progress toward scalable and trustworthy AI-driven manufacturing. This convergence of physics, data, and quantum intelligence outlines a pathway from proof-of-concept demonstrations to reproducible and deployable industrial AI ecosystems.