AI Implementation in Manufacturing: Current Status, Challenges, and Future Research

This paper discusses the advantages of implementing artificial intelligence in manufacturing. Deep Learning, one subset of artificial intelligence, has gotten attention recently for its potential to help manufacturing troubleshooting and design more efficient manufacturing processes by reducing cost, cycle time, and increasing productivity. Deep learning can be used in process optimization, defect detection, supply chain optimization, tool lifetime prediction, and preventative maintenance applications. Although different deep learning models are being employed to solve above mentioned issues in manufacturing, the implementation of deep learning and reinforcement deep learning faces various challenges. These challenges include the quality of data, data governance, the available ways to generate synthetic data, and the lack of interpretability for better output explanations. This paper discusses the possible ways to generate quality training data for deep learning and reinforcement deep learning model development, and deployment. This paper considers only qualitative literature and case studies to understand the current trends and challenges, and considers these as the benchmark for creating a manufacturing AI checklist and designing future research trends to help the manufacturing industry in successful AI implementation. This paper also discusses the possibility of using a multi-platform, multi-sided business model to create a suggested timeline for AI implementation in the manufacturing industry.