Session: 10-10-02: Teaching Laboratories, Hands-on lab Experiences, Online Laboratory Teaching, Virtual Lab Simulation, Use of AI in Laboratory Experiments, Laboratory Equipment and Safety Practices, Technology-Aided Lecturing, Novel Manufacturing Processes II

Paper Number: 167195

AI-Driven Smart Manufacturing: Digital Twin and VR-Enabled Predictive Maintenance for Industrial Efficiency Optimization 

The advent of Industry 4.0 has driven the integration of artificial intelligence (AI), digital twin technology, and virtual reality (VR) into modern manufacturing ecosystems. This research introduces a comprehensive AI-enhanced digital twin framework for intelligent manufacturing monitoring, leveraging data from established repositories such as the NIST Manufacturing Data Repository and real-world industrial datasets, with a focus on predictive maintenance and process optimization in a manufacturing dataset collected from publicly available repositories and industrial case studies. The study employs a hybrid approach combining AI-driven predictive analytics, digital twin simulation, and immersive VR environments to provide real-time insights into key performance indicators (KPIs), including production efficiency, defect rates, energy consumption, predictive maintenance, downtime hours, worker productivity, and machine utilization.

The digital twin acts as a real-time cyber-physical representation of the factory floor, continuously synchronizing with sensor-driven data to mirror production states and identify potential inefficiencies. A predictive maintenance model powered by AI and deep learning algorithms forecasts potential equipment failures based on historical anomaly detection and operational telemetry, significantly reducing unscheduled downtime. Additionally, VR-based operator training modules simulate real-time scenarios to enhance workforce adaptability and refine system diagnostics. Time-series analysis, histogram distributions, and multi-variable correlation mapping enhance decision-making processes by dynamically adjusting production parameters in response to AI-driven insights. The system's ability to integrate real-time digital twin simulations with predictive maintenance enables proactive decision-making, optimizing overall equipment effectiveness (OEE) and ensuring seamless process automation. By leveraging deep learning models for real-time scene interpretation and reinforcement learning for adaptive motion planning, the arm setup autonomously identifies suitable installation points and executes accurate mounting procedures.

 The VR interface functions both as an operator control station and a digital twin, allowing users to visualize and interact with the system in a safe, simulated environment before committing real-world actions. Through a combination of motion capture, haptic feedback, and augmented perception, operators can supervise, refine trajectories, and intervene in complex scenarios as needed, thereby ensuring both safety and operational flexibility. This research demonstrates how AI-driven digital twins and VR-assisted monitoring can transform industrial automation, creating a resilient and adaptive manufacturing environment. The proposed system significantly improves operational efficiency, enhances workforce skill development, and enables predictive analytics to preemptively mitigate machine failures. The findings emphasize the feasibility of digital twin-augmented predictive maintenance as a cornerstone of next-generation smart manufacturing strategies, providing a scalable model applicable across diverse industrial sectors. Experimental results, conducted under controlled laboratory settings and in semi-structured field environments, demonstrate a marked improvement in sensor placement accuracy, task efficiency, and operator workload when compared to traditional teleoperated or purely manual approaches. This research underscores the potential of blending AI-powered autonomy with VR-enhanced teleoperation, paving the way for broader applications in industrial maintenance, inspection, and hazardous environment operations.

Presenting Author: Yunshun Chiou Drexel University

Presenting Author Biography: Dr. Richard Chiou is a Professor within the Engineering Technology Program in the Department of Engineering, Leadership, and Society at Drexel University, Philadelphia, USA. His educational background is in manufacturing with an emphasis on mechatronics. In addition to his years of industrial experience, he has taught many different engineering and technology courses at undergraduate and graduate levels. His tremendous research experience in manufacturing includes intelligent manufacturing, environmentally conscious manufacturing, Internet based robotics, and clean energy and energy efficiency. In the past years, he has been involved in smart and sustainable manufacturing for maximizing energy and material recovery while minimizing environmental impact. He has conducted many experimental studies on virtual reality robotics for human exposure and health hazard analysis in green manufacturing. He has also developed augmented reality/virtual reality (AR/VR) robotics and clean energy and energy efficiency laboratory recently. He has secured many research and education grants from the National Science Foundation, the US Department of Education, the Society of Manufacturing Engineers Education Foundation, and industries. Techniques developed from his research projects have been extended to the educational activities involving intelligent manufacturing with clean energy and energy efficiency.

Authors:

Yunshun Chiou Drexel University
Nijanthan Vasudevan Drexel University
Arjuna Karthikeyan Senthilvel Kavitha Drexel University
Tzu-Liang (Bill) Tseng The University of Texas at El Paso