Artificial intelligence (AI) has made surprisingly deep inroads into the manufacturing realm over the past half-dozen years. According to management consulting firm McKinsey, only 20 percent of manufacturers reported using AI in 2017, but today that number hovers between 50 and 60 percent. With the tight margins and stiff competition most manufacturers are currently facing, AI adoption is not about hopping on the latest technological bandwagon. It’s about real value to your manufacturing operations: accelerating production cycles, speeding up in-process inspections, gaining greater control over materials and workflows, raising quality outcomes, and much more.
When Aligned Vision first released our LASERVISION system for in-process inspections and laser feedback, we quickly realized the value of AI in manufacturing inspection applications. From a weeks-long application development process in which we hand-engineered the inspection analysis algorithms, we transitioned to an AI-enabled application development process that took a day or two. What’s more, the AI-enabled algorithms achieved near-100 percent accuracy, right out of the gate.
What value might your enterprise gain from AI in manufacturing? Let’s take a look.
What can AI in manufacturing do for you?
Manufacturing devices and systems equipped with artificial intelligence are able to perceive their environment and determine the best course of action to achieve intended goals. In industrial settings, two sub-categories of AI predominate:
- Machine learning (ML) is the ability of a computer to use data to improve performance of a particular task without being explicitly programmed to do so
- Deep learning (DL) also uses data to improve performance but typically on larger data sets and more complex logic networks
AI in manufacturing offers two primary benefits: (1) it accelerates application development, and (2) it streamlines continuous improvement processes. Industries like welding and composites fabrication that rely on a combination of manual and automated processes will gain considerable value from AI in several operational areas.
Material and asset tracking – These systems, which are designed to help minimize downtime for production equipment and avoid costly material waste, are hubs for large quantities of data. They hold production equipment information like a machine’s age and its key operating parameters over time, the number of cycles it has performed, and maintenance and repair procedures performed on it. Regarding materials, they collect data on incoming material quality, material storage location and conditions, material life and expiration dates, and kit locations and life. A machine learning system might analyze this data, “learn” patterns within it, and detect anomalies that may prompt remedial action. A deep learning system might detect systemic patterns and uncover opportunities to improve supply chain efficiencies, minimize material backlogs, optimize equipment performance, better integrate material and production management, improve scheduling across the shop floor, and ultimately, reduce operational costs while better assuring the quality of the end product.
Process control – Manufacturing software solutions enable your engineering team to generate electronic work instructions for each of your floor operators. Process control systems also link design and engineering data with your shop floor automation, ensuring that your manufacturing operations carry out design intent accurately. ML lets you automate the the creation of work instructions or make improvements that raise production efficiencies. DL could analyze data generated over the course of multiple production cycles to uncover any bottlenecks or process points at which errors tend to be introduced, leading to continuous improvement opportunities.
Closed loop manufacturing – When you have fully implemented data delivery and connection systems across the factory, this data infrastructure creates a closed loop. Data flows seamlessly from design, simulation and analysis to manufacturing engineering production planning, then to process control and automatic inspection, then to documentation and back to design. ML could accelerate the completion of a closed loop in your factory. DL would employ data from the manufacturing floor to quickly improve design as well as process and product quality outcomes.
AI-enabled automatic inspection with LASERVISION
Inspection in production tends to lag behind other aspects of manufacturing operations in terms of speed and accuracy, especially in industries that still largely rely on manual inspection by trained human inspectors. As LASERVISION has entered these industries, it has proven to reduce inspection cycle time by 90 percent or more, while improving inspection accuracy to nearly 100 percent.
To implement LASERVISION for in-process inspection at your facility, you will need application-specific analysis algorithms. If you’re like most manufacturers, you cannot afford a long development cycle before you launch production of each new product. Artificial intelligence eliminates tedious, time-consuming hand engineering and gives you the application development speed you need.
While ML may be appropriate for simpler inspection applications, DL is the AI tool needed for more complex applications, where a multi-layer neural network is required to learn from a larger and more complex data set. In either case, AI makes it surprisingly fast and easy to implement automatic in-process inspection at your factory. We describe this implementation in a previous blog.
It wasn’t long ago that the Aligned Vision team thought AI in manufacturing was a futuristic concept someone blue-skied at a tech conference. No more. AI in manufacturing is a realistic strategy for today, one that just may be the key to your competitive edge in your manufacturing space.