Wheel loaders maneuver between piles of dirt, dump trucks beep their way into position, dust hangs in the air. Whatever doesn’t fit in the mirror disappears behind the driver’s cab. And somewhere in between, a person is walking around. When machines weighing as much as small houses move through tight work zones, every second counts.
Collisions with mobile machinery are among the most severe accidents that occur on construction sites, in ports, and in the recycling industry. Usually it’s just a little jostle, damaged equipment, bent crash barriers. But each incident costs time, money, and nerves. In environments like these, performance is measured not only in cycle rates and shipment volumes, but in smooth operations as well.
With the Visionary AI-Assist, SICK has brought an assistance system to market that works right where humans, machines, and complex surroundings intersect. The solution combines a robust 3D stereo camera with AI-based person detection in a single device. “We’ve built a true all-in-one device – with full evaluation capabilities right on the camera,” says Ulrich Druzella, Product Manager in the Mobile Perception business unit. For users, this means no additional computing unit and no complex infrastructure.
Evaluation is performed right on the device, shortening reaction times and reducing the number of interfaces. This is particularly important for heavy earthmoving machinery, where seconds are made in fractions of a second. “Evaluation happens right where the event occurs,” explains Druzella. “This makes the system fast and independent.”
In the cab, a 2D live view provides the user with an overview. Machine operators can see what is happening behind or beside them in real time, even in hard-to-see areas. But taken on its own, this would not be a paradigm change. The actual advancement lies in the combination of three-dimensional distance measurements and the ability to distinguish people from objects.
Traditional systems often work either only with camera images or distance measurement. In complex environments, this can easily result in false alarms. Bollards, a pile of dirt, moving equipment – anything can trigger an alarm. Over time, users become desensitized to warning signals. “The Visionary AI-Assist has great person detection capabilities – which is exactly what the market values,” says Druzella. Warning zones can be defined flexibly, with up to 16 evaluations possible simultaneously. A warning is issued only if a predefined zone is entered. This reduces the occurrence of unnecessary signals – and increases acceptance.
Target sectors will not be surprised to learn that the camera is dust-proof, can withstand high-pressure cleaning, and works reliably both in single-digit temperatures and summer heat. Construction sites, mines, and ports are not test labs. “When we launch a solution, it must be able to survive application,” says Druzella. Robustness, easy integration, reliable detection – according to Druzella, these are the issues that customers emphasize again and again.
Originally, the Visionary AI-Assist was intended primarily for heavy construction machinery. But time has shown that the demand goes beyond that. In the port sector, for instance, where cranes and vehicles follow tightly scheduled processes. In the recycling and metal industry, where huge plants require monitoring. Inquiries have even come from the airport sector – for monitoring luggage handling systems, for example, where unauthorized access is a recurring issue. “These are examples of applications that we weren’t initially considering,” comments Druzella. “But it reveals the solution’s potential!”
The response after market launch was correspondingly high. In just a few months, opportunities to the tune of tens of millions poured in. For Druzella, this is another indicator that there had been demand in the market for a long time. “We are bringing AI to collision warning systems, which is a decisive step for SICK,” he says. At the same time, he emphasizes that the product is not safety-related; it is an assistance system. “But we actively help to prevent safety-critical incidents from occurring in the first place.”
For SICK, the Visionary AI-Assist represents a strategic step away from rule-based systems and toward situation-distinguishing sensory perception.
Within the company, the Visionary platform was a lot like a start-up early on. “It took a lot of effort to establish it in the company,” recalls Druzella. But now the solution is a growth engine – and this move outdoors a logical next step.