AUTHOR=Azeta Joseph , Omeche Theodore Tochukwu , Daniyan Ilesanmi , Abiola Johnson Opeyemi , Daniyan Lanre , Phuluwa Humbulani Simon , Muvunzi Rumbidzai 
  
TITLE=Artificial intelligence and robotics in predictive maintenance: a comprehensive review
  
JOURNAL=Frontiers in Mechanical Engineering
  
VOLUME=Volume 11 - 2025
  
YEAR=2026
  
URL=https://www.frontiersin.org/journals/mechanical-engineering/articles/10.3389/fmech.2025.1722114
  
DOI=10.3389/fmech.2025.1722114
  
ISSN=2297-3079
  
ABSTRACT=The integration of artificial intelligence (AI) and robotics into predictive maintenance (PdM) systems has brought about a fundamental change in the operations of the industries since it has left behind the previous method of reactive and scheduled maintenance models in favor of proactive and data-driven models. The current systematic review of literature (2015-2025) is aimed at the development of PdM, in which AI techniques, machine learning, sensor technology, and the incorporation of robotics contribute to more efficient systems and address the difficulties in their implementation and implications for the future of industries. The findings show that the support vector machines and neural networks with supervised learning algorithms are very accurate in fault classification and the remaining useful life prediction. On the other hand, the methods of unsupervised learning can be applied in the detection of anomalies in cases where a limited quantity of labelled data exists. Examples of deep learning architectures that are more effective in processing more complex sensor data, as well as time-series patterns, include convolutional neural networks (CNNs) and long short-term memory (LSTM) networks. Moreover, sensor systems that are already linked to the IoT provide the ability to monitor in real time, and this significantly improves fault detection. The AI-based PdM systems in combination are highly rewarded with reduced downtime, longer equipment life, and enhanced maintenance scheduling. There are still, however, concerns about data quality, computation loads, and implementation cost that remain a major barrier to common usage. The future of AI should be on explainable AI, hybrid modelling techniques, and enhanced sensor technology to render AI scalable, interpretable, and more industry-applicable.