May 26, 2020
Investing in AI and Mixed Reality will Improve Manufacturing Beyond COVID-19
By Julie Orlando, Chief Product Officer
To increase resilience in times of crisis, partially remote workforces and virtual factory control are not just futuristic ideals. They are a necessity.
Whether required to ramp up manufacturing to meet higher volume targets or pivot to combat the COVID-19 crisis with PCR tests, improved PPE, and lower-cost respiration devices, manufacturers are racing to meet unprecedented challenges.
The supply chain crisis within the COVID-19 crisis caused shortages at the same time as it lengthened lead-times for needed parts and materials, slowing manufacturing to a halt in some places and revealing previously overlooked obstacles in others.
Many ubiquitous parts needed to make medical devices and other needed supplies, such as basic plastic and rubber components, which were once manufactured inside the United States are now manufactured abroad, and are currently unreachable. Factories are pivoting on a dime to switch fabrication of these essential parts to in-house so they can continue building their products. Jumping over fabrication hurdles while facing reduced or staggered workforces requires updating outdated processes and integrating new tools.
Companies from GE to Intel are reevaluating their reliance on a complicated supply chain and considering distributed or vertical manufacturing. Many old and new guard manufacturers see artificial intelligence and other leading-edge tools as the path to achieve this new paradigm.
Safely steering through the shifting waters of our economic, health, and political landscapes requires building and sustaining a new battalion of industrial workers that can remotely control an entire production line. This effort will increasingly rely on a foundation built of cutting-edge technologies: Mixed Reality, decision-making AI, and full-to-partially automated inspection for factory control at a distance.
Before COVID-19, many companies focused their efforts to automate on specific pain points, like the transportation of fragile materials on the factory floor
In quarantine, for the first time, some of these companies are automating roles beyond taskrabbit robotic arms, wheels, or other rote labor. Dynamic AI has been optimized on the backend to make informed, learned, repeatable decisions in complex manufacturing situations, transforming the routine of the factory floor, and freeing up time for engineers to work creatively and methodically on improving processes and designing new products.
AI for Quality Control and Inspection:
The advances of artificial Intelligence are most beneficial in times of instability and transition. By freeing up manual operations and inspections, fast feedback to manufacturing allows engineers and process owners to optimize and correct issues as they arise. This foundation of artificial intelligence ensures the symbiotic flow of quick transitions when workflows must pivot to build new products or incorporate design changes.
Quality inspections, often ignored in terms of automation, remain an outlying challenge for many large and small American manufacturers. In the past, it seemed apt to have skilled human experts give the final say when it came to the viability of intricate and complex processes like the building of wearable tech, graphene sensors, or photovoltaic cells. But, due to quarantine, many of these same manufacturers are looking to automate all or part of these processes for the first time.
Inspection can require tens of thousands of hours of human laboor, becoming a laborious obstacle to fast production with reduced workforces. If one is required to fully manually inspect 100,000 products a day before any shipments can be made, this creates a bottleneck in times of need. Beyond the concern for time constraints, during times of transition and staggered workforces, the potential for inaccuracies, which could lead to defective products, can increase due to an over-stressed, thinly spread operating team. As we recognize our limitations, data scientists and engineers are building systems around these human blindspots, deploying human-centered solutions to intricate problems. If trained on a rich dataset, classical ML algorithms used in factory control for the automation of inspection tasks can achieve higher accuracy than trained engineers. This frees the engineers to make adjustments in the process and see yield and quality improvements in the final product.
Thanks to a combination of human expertise and Deep Learning and image capturing, recognition, and classification, automated optical inspection solutions have caught up to the most advanced materials and products.
Skilled operators build their knowledge of a system based on experience and formal education; AI builds similar associations through sparse-data training. An AI infrastructure increases production mobility by gathering and processing results simultaneously. This in turn leads engineers to find improved designs by delineating the underlying causes of pain points, while operators focus on keeping production going at full speed.
In today’s changing workplace, AI can be leveraged to maximize the human operator’s flexibility, working hand-in-hand with engineers to alleviate the strain of repeatable tasks. The final goal is to produce at a high yield with very little scrap or waste. If AI aids humans in inspection, factories have a system in place to rely on in times of need, and a foundation to create faster design iteration when they want to transition their production pipeline.
In order to meet demand in the current crisis, manufacturers are re-tooling like never before. Due to social distancing, these improvements require manufacturers to find creative ways to install new pieces of technology.
Generally, the process to install and integrate new equipment into an existing production line can be lengthy and involved–requiring teams of engineers to physically install, program, and train staff on new equipment.
One of the solutions for product integration includes using Mixed Reality (MR) interfaces to demonstrate and test new technology using existing data and processes. MR in these cases is not a high-tech gimmick but an essential component to installation, allowing operators to implement improvements and upgrades to existing production lines at record pace with little need for a physical human presence.
During quarantine, teams are incorporating MR technologies into their daily routine out of necessity. Support teams repair operating equipment through virtual channels, sales teams demonstrate tools, and engineers operate these tools to collect data from a distance. This time of upheaval and the loss of our entrenched habits have provided a testing ground in which virtual, augmented, and mixed reality technologies have more than proven their viability in the field.
Pre-COVID, some factories utilized Mixed Reality to provide instruction and guidance for worker performance. Managing operators demonstrated correct job functions and signaled various warnings or alerts. This adjacent role has expanded rapidly. Now, using mixed reality, teams of engineers control tools in real time to provide accurate results entirely remotely.
Virtually guided machines gather the same data without on-site human support. We’ve demonstrated, during this crisis, that the machines of the future can be controlled entirely remotely, like video games. These are not the lights-out factories of science fiction, but an example of human and technological collaboration
The untold benefits for virtually operated factories will open doors to a new era in manufacturing. We can go from changing gears over six-to-nine months to building a fully operating new pipeline in a matter of weeks. Virtual tools help us to come together to accomplish feats of engineering and production.
Historically, a reduced or partial human workforce forced factories to shutter or substantially decrease production and yields. This in turn lowered final product quality, which spelled out long-lasting economic ramifications.
Today, this is no longer the case. With human insight creating tools for feedback and systems for factory control, economic crises can be detoured through technological innovation. The companies who rise to the challenge of a reduced workforce and bottle-necked supply chains will change the future of manufacturing. We can solve and prevent future manufacturing crises, and in so doing, rewrite history.