Artificial intelligence
The production floors and warehouses face a number of complex issues. Safety always comes first, but efficiency follows. Creating workflows is a long-standing technique for improving both factors, but how can the results be improved? Artificial intelligence (AI) has begun to creep into manufacturing processes in recent years, just like it does in all other business sectors. Advanced in AI, networking and edge devices bring another modern technology into the mix – Augmented Reality (AR). The combination of AI & AR is the latest attempt to increase security and productivity.
Shop floors are crowded and dangerous places. Since the beginning of the industrial revolution, there has always been a tension between the need for efficiency and safety. In the beginning, safety was not so important. However, in the last century that has come to the fore and most companies are leading the way in their messaging with that, even if there are some that only do it there. However, productivity remains important. Advanced technologies of the past decades have often focused on improving both factors at the same time.
When it comes to AI, early applications were at two very different extremes. First, early vision AI would look for simple safety issues, such as missing safety helmets. Second, AI was used to find out optimal plant processes and process flows. Recent developments have allowed systems to begin to integrate these features and others. One way that helps with this is AR.
Popular culture seems to be familiar with virtual reality (VR) as it has been used as a trope in many movies and games have started moving towards both imitating VR and working in it. While AR had a pop culture moment years ago, wearing glasses suggested by a major company, it was a failure, and many haven’t thought much about it since. Simply put, AR is the concept of augmenting reality with technological additions. A simple example is the heads-up display, which has been used in combat aircraft for years and to a lesser extent in some cars in recent years. One industry that has adopted VR is in surgery, with VR, AI, and robots expanding surgeons’ abilities to know more as they work.
A few years ago I was reporting on a company where an inspector on a construction site wore headphones. The images were saved and later compared to blueprints. Part of the fun of this column is seeing the changes over time. Technology has advanced, both in AI and networking. What those advanced ones are starting to achieve is a more direct link between the backend AI systems and the person in the field.
One of the first ways AR helps production is simple. Get a warehouse. When many products are flowing through, checking inventory can be slow. With AR system, a person can look at a group of boxes, the backend system can count the boxes. That’s just a simple visual AI. However, that count can be integrated with inventory and shipping systems, the count can be compared to the expected count and a display can appear, for example that the person has to search for two more boxes that should also have been in the order.
Another example is physical security. It’s one thing to notice if someone is wearing a hardhat. That’s now in the “must have”, more basic visual identification aids. Let’s extend that. The visual system can capture warning signs requiring hard hats, gloves and other safety precautions, or even use the GPS information and database content about the production floor to know what safety precautions are required at each location on the floor. “Gloves are a great example of how the power of AI can be used to improve safety on the production floor,” said Dr. Hendrik Witt, Chief Product
AiStudio analysis gloves
Officer, TeamViewer† “AI can now detect not only whether you are wearing gloves, but also whether you are wearing the right type of gloves for the specific situation, and then immediately notify the employee for a safety correction.”
A final example is another security issue. Background process analyses, using AI or standard analyses, can be performed to estimate the potential for fatigue. A person lifting 10-pound boxes doesn’t tire as quickly as the same person lifting 40-pound boxes. Reminding people when to take breaks is just as important for safety as making sure people work safely.
One aspect of AI/VR that has also slowed down adoption is the need for companies to hire experts in that role, experts who don’t have a lot of resources and who cost more. Companies like TeamViewer are working to simplify training the systems by making them as low-code as possible. This is what other technology cycles have had to do to increase adoption. It’s also something that’s one of my most important soap boxes. Few companies need the magic data scientist. What needs to be done is to build systems that can talk to the line users in the language they know.
“Modern augmented reality is about more than the all-important goal of improving safety,” Witt says. “It’s about understanding process flows, about integrating AR systems with more of the full range of ERP and other backend software, not just AI. These are systems that non-specialists can use to perform their own tasks.”
The aforementioned glasses were all the rage, something cool for a small audience. AR is now being applied to more focused arenas, including manufacturing. That focus can ultimately yield the ROI for an investment in AR and AI that will both be more widely distributed.