Varsha Venkatesh, a robotics and autonomous vehicles mechanical engineering graduate student, learns how to program and use an industrial manipulator robot arm. (Image courtesy of Joseph Xu, Michigan Engineering.)
Modern manufacturing plants are comprised of hundreds of advanced components, both hardware and software, including robots, conveyor belts, sensors, control systems and communication networks. All these components work together in an intricately choreographed performance that results in the smooth and efficient operation of a factory.
But what happens when something goes wrong?
Machine or software failures, operator mistakes and cyberattacks can cause serious delays, or even halt production altogether. This leads to unscheduled factory downtime – which can be expensive – or to potentially dangerous situations that could result in injury or damage to plant personnel and machinery.
A group of engineering researchers from the University of Michigan have made it their goal to address these glitches. This will be done by detecting these glitches – in real time – and quickly and efficiently reconfiguring factory operations around them.
The aim of the engineering professors and students involved in the project is to develop a new methodology for controlling manufacturing systems in order to increase a factory’s productivity, and therefore its competitiveness.
They call their methodology “software-defined control.”
Central to this new approach will be a continuous, real-time simulation of an entire manufacturing plant. The team will produce this simulation, and develop software that compares a plant’s actual operation to what they should expect based on the simulation.
“The idea is you have the physical manufacturing plant and the simulated model of the plant, so if there's a difference between the two, you can detect a fault or a cyber-intrusion," said Dawn Tilbury, a professor of mechanical engineering at the U-M College of Engineering and the project’s principal investigator. "The goal is to develop control systems for manufacturing systems that are secure and reconfigurable automatically."
This system could then be used to reprogram the way parts flow through the plant, to avoid a faulty piece of equipment. This is especially valuable to factories that are seeing a rise in automation, which necessitates finding better ways to keep tabs on plant operations.
"Automation may increase efficiency and raise quality, but it brings with it vulnerabilities," Tilbury added.
Factory robots are increasingly networked, and the companies that produce these machines can often log in remotely to make repairs. While this access is legitimate and necessary, it can also be an exploitable weakness in a factory system’s cybersecurity. And as manufacturing systems become more digitally connected and complex, they will become increasingly susceptible to this type of disruption.
"Our work aims to develop the science and enabling technologies to transform manufacturing systems from the current paradigm of low efficiency and high susceptibility to system disruptions to a new era of system-level anomaly detection, classification and action," said Kira Barton, U-M assistant professor of mechanical engineering. "This will lead to less downtime, faster responses to disruptions and a more efficient manufacturing system."
The initial research project will focus on discrete parts manufacturing, but the research team believes that their software-defined control method will translate well to other manufacturing, such as for semiconductors or batch processes.
To learn more, visit the University of Michigan College of Engineering.