Integrating AI/ML with Lean Manufacturing techniques can significantly enhance efficiency (doing things right) and effectiveness (doing the right things) in industrial processes. AI-powered solutions enable real-time detection, prediction, automated optimization, and dynamic decision- making, reducing errors, downtime, and other production wastes, and improving Overall Equipment Effectiveness (OEE). The objective of this project is to develop Lean automated solutions enhanced by a machine learning (ML) algorithm to optimize the efficiency and effectiveness of a beverage production line at Kassatly Chtaura. The innovation of this project lies in transforming a manual labor operated bottling line into an automated, self-optimizing production system that merges lean principles with machine learning, a novel approach still not common in industry and virtually inexistent in Lebanon. Instead of relying on static settings, periodic manual adjustments, and reactive quality control, the line becomes predictive, real-time, and data-driven. ML-enabled predictive systems detect micro-defects on the line, forecasting failures before they become scrap or rework. This shift from reactive manual work to intelligent manufacturing defines the core innovation of the project.

This is a two to three-year project than spans multiple disciplines including Lean manufacturing, quality management, mechanical design, artificial intelligence, and data analytics. The project aims to design and implement an AI/ML-enhanced Lean Manufacturing system that transforms a manually operated beverage bottling line into an intelligent, self-optimizing production environment. At their core, both Lean Manufacturing and AI/ML are based on data-driven decision- making, continuous improvement, and waste reduction. But while the Lean approach has traditionally relied on human expertise and direct implementation on the factory floor to optimize production processes, AI/ML takes these concepts further by automating data analysis, detecting patterns in real time, and making intelligent predictions beyond normal capabilities.

Bringing together the skills of undergraduate students from Industrial Engineering, Mechanical Engineering, Computer Science and Computer Engineering, the project will develop an integrated hardware-software prototype capable of real-time sensing, prediction, and autonomous decision- making. By embedding an AI/ML algorithm with Lean methods like Poka-Yoke (mistake-proofing) and implementing them into key production stages of a bottling line (such as rinsing, filling, de- palletizing, and packaging), the system will continuously detect micro-defects, forecast equipment failures, adjust process parameters, and eliminate production wastes, thereby enhancing both production flow and quality. The project will require data collection from the production line, ML model development and training, and implementation of an IoT sensor and camera so real-time data feeds a digital twin of the production line. The introduction of AI-driven Lean methods enables dynamic balancing of the Takt time (the timing of production required to meet customer demand), predictive analytics, automated root-cause analysis and decision making, automating production and removing the need for manual quality checks and reducing the need for human intervention to fix problems. The project also lays the foundation for an AI-driven lean manufacturing platform, which students can develop into a commercially viable solution.

Desired disciplines

Computer Science
Data Analytics
Engineering

Team Leader

Dr. Marc Haddad

Partner