Researchers at Rutgers University have developed artificial intelligence systems to address two key manufacturing challenges: making 3D printing more reliable in harsh environments and accelerating innovation in conventional manufacturing processes. The research, led by Associate Professor Rajiv Malhotra from the Department of Mechanical and Aerospace Engineering, resulted in two separate studies published in manufacturing journals.
The first study, published in The Journal of Manufacturing Processes, focuses on “expeditionary additive manufacturing” in extreme environments such as space, war zones, and disaster areas. The team developed a conditional reinforcement learning technique that uses cameras to monitor printing and automatically adjusts printer settings when defects are detected. “We trained the AI to expect the unexpected, rather than expect the expected,” Malhotra said. The system can reduce defects by 10 times or more without requiring the printer to stop or the software to be retrained.
The second study, published in The Journal of Intelligent Manufacturing, addresses the slow pace of manufacturing innovation. Traditional methods can take decades to develop and require extensive experimentation. Malhotra’s team created an AI system that reads scientific papers, extracts relevant information, and combines it with small amounts of experimental data to build predictive models. The approach achieved accurate results with just 30 samples instead of hundreds of experiments.
The expeditionary manufacturing research involved collaborators from the U.S. Army Armaments Graduate School, University of Michigan, and University of Connecticut. The manufacturing innovation study included researchers from the University of Connecticut and doctoral students from Rutgers. Both systems are designed to work in industries including aerospace, automotive, electronics, and defense applications.
Source: rutgers.edu
