Gray Solutions is leaning heavily into artificial intelligence (AI) to aid vision solutions and digital-twin models. The focus is on systems and machines that have the data-collection capabilities required to meet demands of the newest technology. Greg Powers, vice president of cool stuff at Gray Solutions, discusses five technology trends impacting system integrators.
1. Digital twins and AI make equipment easier to deploy in the field. Gray Solutions uses digital twins for a variety of design and integration practices, such as process optimization, experimentation with what-if scenarios, material flows optimization, virtual testing/commissioning and proof of concepts (POCs).
“First, we use modeling/simulation to validate the model,” Powers says. “Once we have confirmed the concept model, we can then ingest the model with real data. From there, we can do some experiments to optimize the model. After running with real data then we have a digital twin of that process. This same model then can be used for commissioning. Gray Solutions uses modeling for most projects we work on from greenfield to brownfield projects. This model will be updated at every stage in the project all the way through commissioning.”
First and foremost, a digital-twin platform needs the necessary data to represent the physical model of the production process, Powers says. He also highlights the following important factors for selecting a digital-twin platform:
- the platform's capabilities for data ingestion, storage, processing and analysis
- system scalability and flexibility to change when business processes change
- integration of existing systems.
2. Artificial intelligence is already enhancing machine vision.
In addition to leveraging simulation/modeling software for the design process of a facility, Gray Solutions has been working with original equipment manufacturers (OEMs) to design machinery to be more easily deployed in a facility. “These tools can be used to merge both physical and virtual processes withing the facilities,” Powers says. “As we incorporate more data in our modeling software, then we can use AI to help assess data and provide some predictions. As for machine integration using AI, this is still in early adoption. It’s still difficult for customers to allow control of equipment based on AI. The exception to this would be in AI vision.”
Gray Solutions has focused largely on the use of AI with advanced vision systems for quality inspections. “Visual-inspection-automation (VIA) software goes beyond the capabilities of traditional machine vision in detecting anomalies and defects, even when products have natural variations. Using proven vision AI technology, manufacturers can scale production, reduce waste and adapt to workforce changes, while achieving even higher levels of quality control,” Powers says. “Machine-vision tasks now require highly complex analysis beyond the scope of traditional vision systems. As a result, the demand for high-quality inspection tasks in the food manufacturing space has grown exponentially. The increasing need to ensure the quality of manufactured food products has exceeded the scope of traditional vision-system techniques, which were widely incapable of performing human-like inspection tasks. Vison systems have made great strides with AI, and, with more companies supplying AI tools, the market has exploded.”
3. Generative AI changes the game for industrial control, but not immediately. Large language models (LLMs) are computational, deep-learning models that form the foundation of generative AI. Early models could predict text patterns, and now they incorporate image, video and audio input and output.
“To build these models, significant costs have been invested in training these models. More training the better the model is,” Powers says. “As we put more focus on smart equipment and smart processes, the expectations are that generative AI will be embedded in these designs. Currently, we are building AI processes that provide recommendations or predictions that get passed to operators. Eventually when the model is trained well enough then we will see more AI doing things like the controlling of the equipment.”
The customer market does not have fully trust in AI yet, Powers says. “Most important is the data this is being used to train the model,” he explains. “For customers to adopt AI, they must have clean and structured data to help train the model. If I were a customer looking at an AI solution in the future, I would put heavy emphasis of data collection, cleansing of data and data structures.
“Our focus along with equipment providers is to make sure systems/machines have capabilities in data collection required to meet these demands in AI. This means defining structured data, collection tools, data cleansing and integration. This will ensure that manufacturers are ready with the data required to train the model for AI,” Powers says.
Powers believes AI will have a huge impact on manufacturing in the future, maybe slower than other areas, but it will impact machine builders and system integrators. “I am optimistic that in the near future AI will not just be able to assist with running the equipment, but eventually AI will be running the equipment lines,” he says. “AI will be able to react very quickly to line changes—upstream and downstream events—through data collection where a machine operator might not be able to assemble/correlate all this data.”
4. Robotics is an important part of process automation. “Gray Solutions develop many robotic solutions in process automation,” says Powers. “This would include quality inspections, palletizing, machine tending and material movements using autonomous mobile robots (AMRs).”
5. Edge computing is necessary for manufacturing to make the most of AI and machine learning. “Having edge devices process and analyze data in real time at a local level vs. in the cloud has given AI early detection of problems. In addition, data can be fed directly into the model without having to aggregate the data,” Powers says.
