The industrial environment has gone through the fourth revolution, also called “Industry 4.0”, where the main aspect is digitalization. Each device employed in an industrial process is connected to a network called the industrial Internet of things (IIOT). With IIOT manufacturers being capable of tracking every device, it has become easier to prevent or quickly solve failures. Specifically, the large amount of available data has allowed the use of artificial intelligence (AI) algorithms to improve industrial applications in many ways (e.g., failure detection, process optimization, and abnormality detection). Although data are abundant, their access has raised problems due to privacy concerns of manufacturers. Censoring sensitive information is not a desired approach because it negatively impacts the AI performance. To increase trust, there is also the need to understand how AI algorithms make choices, i.e., to no longer regard them as black boxes. This paper focuses on recent advancements related to the challenges mentioned above, discusses the industrial impact of proposed solutions, and identifies challenges for future research. It also presents examples related to privacy-preserving and explainable AI solutions, and comments on the interaction between the identified challenges in the conclusions.

Industrial scheduling plays a central role in Industry 4.0, where efficiency, robustness, and adaptability are essential for competitiveness. This review surveys recent advances in reinforcement learning, digital twins, and hybrid artificial intelligence (AI)–operations research (OR) approaches, which are increasingly used to address the complexity of flexible job-shop and distributed scheduling problems. We focus on how these methods compare in terms of scalability, robustness under uncertainty, and integration with industrial IT systems. To move beyond an enumerative survey, the paper introduces a structured analysis in three domains: comparative strengths and limitations of different approaches, ready-made tools and integration capabilities, and representative industrial case studies. These cases, drawn from recent literature, quantify improvements such as reductions in makespan, tardiness, and cycle time variability, or increases in throughput and schedule stability. The review also discusses critical challenges, including data scarcity, computational cost, interoperability with Enterprise Resource Planning (ERP)/Manufacturing Execution System (MES) platforms, and the need for explainable and human-in-the-loop frameworks. By synthesizing methodological advances with industrial impact, the paper highlights both the potential and the limitations of current approaches and outlines key directions for future research in resilient, data-driven production scheduling.

Industrial scheduling remains a pivotal discipline for efficiency, resilience, and competitiveness in manufacturing and service operations. The digital transformation, driven by paradigms such as Industry 4.0, has fundamentally reshaped the scheduling landscape, introducing pervasive connectivity, real-time data flows, and cyber-physical integration. This review synthesizes advances across three enduring challenges: (i) scalability and computational complexity in large-scale, high-dimensional scheduling environments, (ii) robustness and adaptability to uncertainty and disruptions, and (iii) integration with digitalization through IIoT, digital twins, cloud–edge architectures, and interoperable, secure infrastructures. We highlight the surge of AI-enhanced and deep learning–driven methods that are redefining state-of-the-art practice. Deep reinforcement learning (DRL) now underpins policy learning for dynamic dispatching and rescheduling; graph neural networks (GNNs) and attention models enable generalization across diverse shop configurations; and digital twin–in-the-loop frameworks provide safe training and rapid adaptation under real-world volatility. At the same time, neural architectures are increasingly embedded within decomposition, metaheuristics, and multi-agent systems—forming hybrid stacks that combine the guarantees of operations research with the adaptability of learning-based approaches. The industrial impact of these developments is evident across semiconductor fabrication, flexible job shops, supply-chain–intensive production, and distributed, autonomous networks. Yet critical challenges persist in interpretability, trust, data quality, and legacy integration. To advance, future research must prioritize explainable and certifiable neural schedulers, standardized datasets and benchmarks, seamless AI–IoT–DT integration, federated and privacy-preserving collaboration, and human-in-the-loop frameworks. Collectively, these directions chart a path toward scheduling systems that are not only more efficient and scalable, but also transparent, secure, and resilient in the digital, interconnected era.

Supply chains are increasingly exposed to compounding disruptions, volatile demand, and sustainability constraints, which challenge optimization approaches designed for stable operating conditions. This review synthesizes recent advances in supply chain optimization with a focus on the integration of artificial intelligence and operations research in decision-making. The paper examines three major capability layers: prescriptive optimization for planning and resource allocation, predictive modeling for demand and risk anticipation, and digitalized execution through simulation and digital twin environments. Across these layers, the analysis shows that hybrid AI-OR architectures tend to outperform isolated methods in settings characterized by high demand volatility, multi-echelon complexity, and disruption exposure, by combining predictive adaptability with constraint-aware decision quality. The review also highlights a strategic shift from single-objective efficiency toward multi-objective performance that jointly manages cost, service, resilience, and environmental impact. From an implementation perspective, the evidence indicates that measurable industrial gains depend less on algorithm novelty alone and more on system-level integration, data governance, and cross-functional deployment. Key research gaps remain in benchmark standardization, explainability, uncertainty-aware optimization, and long-horizon validation under disruption. The paper concludes that the next generation of supply chain optimization will be defined by continuously learning, human-supervised decision ecosystems that remain robust under uncertainty while delivering operational and sustainability outcomes.