EN
Country: EN (United States)
Select country/region

Language

  • English
stage-teaser-pace-KI-Quality-2023-image-webx

Solve tasks with AI in unprecedented quality

A conversation with Dr. Stefan Odermatt, Senior Vice President Research & Development Integrated Automation, and Dr. Shane MacNamara, Senior Vice President Research & Development Autonomous Perception, on the importance of digital solutions for SICK, the role of Artificial Intelligence, and a 20-year-old slogan that’s now more relevant than ever.

Dr. MacNamara, Dr. Odermatt: Thanks to the digital transformation and the rapid advancement of Artificial Intelligence, our world is spinning faster and faster. What do these developments mean for SICK?
DR. SHANE MACNAMARA: SICK has a tremendous portfolio, encompassing roughly 40,000 products and solutions. How can we shape that portfolio in the future without constantly expanding it? In this regard, digitalization plays a key part: software is becoming an integral component of our sensors, opening up new functions and, with them, a completely new dimension of solution development. Digital solutions will help us meet our customers’ needs in a more focused way.

DR. STEFAN ODERMATT: In addition to “Software Defined Sensing,” that is, software as part of our sensors, a further aspect is the use of software at the machine and process level: here we’re coming ever closer to our customers’ IT systems and connecting them to the shop floor, that is, the manufacturing and logistics level, to solve problems in industrial automation.

DR. SHANE MACNAMARA: In short, software solutions make us faster and more flexible, since they can be adapted. That’s also a major advantage for our global business. In this regard, our Business Clusters are responsible for developing building blocks, which are then “put together” in the respective regions. The “last mile customization” is done as close to the customer as possible at one of six globally distributed Engineering Hubs, which are currently being set up. All in all, this approach offers us a level of scalability we’ve never had before.

Let’s talk about Artificial Intelligence. How long has AI been important for solution development at SICK?
DR. STEFAN ODERMATT: Artificial Intelligence, or Deep Learning, grew in popularity in roughly the mid-2010s; since then, we’ve increasingly focused on it. In 2017, SICK even launched a Deep Learning start-up. Here, the focus wasn’t on a single product; rather, the goal was to view the topic comprehensively and integrate AI / Deep Learning in our sensors, but also to solve more complex automation problems using AI on industrial PCs. Even back then, our goal was to achieve cross-divisional collaboration. This evolved into today’s Deep Learning Framework, an element that connects everything else: it’s where you’ll find the majority of our AI experts, and their know-how is available to everyone in the company.

DR. SHANE MACNAMARA: By the way, SICK was the first company to offer “On Device Learning,” long before our competitors: the device learns how to complete small tasks using the AI alone, without any edge device or cloud connection. For larger AI tasks, scalability is possible via an edge device or the cloud. In the last two or three years, access to these AI solutions has become much simpler, both for integrators or plant constructors – our primary customers – and for the end users who apply the solutions, not to mention our own Development and Sales. In a sense, AI has been “democratized.” What that means for us: by using AI solutions, we can solve far more problems, while also becoming relevant for more customers.

So, software and the possibilities that AI offers are also important for SICK’s internal processes?
DR. SHANE MACNAMARA: Software and AI offer valuable support for our Development and Sales. For instance, we can digitally simulate problems involving customers, like a certain size of parcel or variance in a track & trace application. Methods like simulation and digital twins, as well as AI methods, allow us to more easily adopt a customer perspective and select the corresponding sensor solution. This makes people’s jobs easier, not just in Development but also in Sales. As such, digitalization is accelerating the process of finding the right product and a tailor-made solution for the respective customer.

DR. STEFAN ODERMATT: AI also facilitates product and software development in a very concrete way: for instance, last year we rolled out an “AI Copilot” for software development. Initially limited to a pilot project with 200 employees in six countries, the tool is now available to SICK staff worldwide. It provides our software engineers support with finding errors, with generating tests, with explaining software code, and even with writing code. And here, too: AI has massively democratized and simplified these processes. Also, the tools are improving exponentially, which means they’ll offer us even better support in the future. We’re also entering into targeted collaborations and acquiring new solutions to keep pace with the latest developments – in this case, trying to develop everything in-house wouldn’t be productive.

Have customers now come to expect solutions with AI from SICK?
DR. STEFAN ODERMATT: No. It’s more like this: a customer comes to us because they have a problem they want us to solve for them. How we do it is secondary. Sometimes we don’t need AI at all. Though the percentage of AI solutions will surely continue to grow, I’m convinced that, ten years from now, there will still be “classic” solutions with no AI.

Could you give us an example of a typical AI application from SICK?
DR. STEFAN ODERMATT: SICK has installed what are known as thermo portals for tunnel operators in a number of European countries. Located near tunnels on major traffic axes, they can detect overheating vehicles before they enter the tunnel, helping avoid vehicles catching fire in a tunnel and demonstrably improving tunnel safety. These systems have been in use for more than a decade and previously used classic image-processing methods. In the past, we had frequent false alarms, for one thing because, when in doubt, the alarm was sounded.

But such false alarms mean a great deal of time and expense for tunnel operators – from shutting down the tunnel, to calling in the fire department and then sending them away again, etc. For roughly the past two years, these systems have used AI, based on statistical data from these thermo portals’ ten years in operation. And we’ve seen: after just a short time, the AI-based method did the job significantly better than the classic algorithm, which had itself been steadily improved over more than ten years – we’re talking about a 10- to 20-fold reduction in false alarms! AI is enabling us to improve safety for our fellow citizens in unprecedented quality – very much in keeping with our philosophy of “using technology for good.

Another typical AI application from SICK is for the track & trace area, where Artificial Intelligence is used to determine whether there are currently bags or boxes on the conveyor belt – information that is essential for the sorting center. To take advantage of this option, customers can purchase a software upgrade.

Is SICK going to become a software provider?
DR. SHANE MACNAMARA: Today, software developers make up roughly 40 percent of SICK’s Development staff. But of course, that doesn’t mean we’re going to turn into a software provider. Though the digital aspect is extremely important for our company, our sensors continue to be the basis for everything else; software and AI solutions are simply making them more and more intelligent. In fact, you could say that the slogan we introduced in 2004 – “Sensor Intelligence” – is now more relevant than ever.