Knut Sandven: Sound is fundamentally deterministic. That means it obeys the laws of physics and makes it possible to calculate performance in various conditions. It travels at a known speed, reflects in predictable ways, and an echo either comes back within a defined window or it doesn’t. That’s exactly the property you want in a safety function: behavior you can bound and prove. The hard part was engineering a sensor that guarantees a worst-case detection and response time even if something should fail.

Certification doesn’t care about average performance; it cares about the worst case. What’s new is that this kind of certified safety is now available in 3D. That should change how the whole industry thinks about virtual safety barriers. Instead of drawing them on a flat plane, you can define them through a real volume around the machine. And once the protected space is something you can trust in three dimensions, you can finally start taking down the physical fences rather than working around them.

MD: The standards referenced in your press release are International Standards that typically complement rather than replace U.S. and Canadian regulatory and certification requirements. What does it actually mean for marketing the product in the U.S.?

KS: Functional safety speaks a common language. North American machine safety is built on the same ISO/IEC backbone, so a sensor certified to the international functional-safety standards gives a U.S. integrator and their safety assessor the documented basis they need to put it to work. The final deployment sign-off still rests with the integrator and the local authority, but we hand them a certified building block.

READ MORE: The Eyes of Automation: Machine Vision’s Role in AI-Powered Automation

MD: Environmental conditions can degrade the performance of cameras, LiDAR and other optical sensing technologies. In what safety-critical scenarios does certified 3D ultrasonic sensing provide a measurable advantage? How should engineers think about sensor redundancy and fusion when designing collaborative work cells?

KS: Ultrasound doesn’t depend on light, so it performs exactly where optics struggle. Dust, glare, darkness and matte or transparent surfaces that cameras and LiDAR may misread. But the bigger point, once a robot works close to people, is that you have to see the whole person, not a slice of them.

For example, a worker crouches to pick something up beside a moving AMR, and a 2D plane at leg height never sees the head and torso now low in the robot’s path, or an operator reaches into a cobot cell and their arm extends well above the single scan line. If the robot is meant to operate near people, full-body 3D detection shouldn’t be only a nice-to-have.

On redundancy and fusion, my advice is to think in layers, not in one fused brain. Keep the certified, deterministic safety channel independent from the probabilistic perception and fusion stack, so an edge case or failure in the AI can’t compromise the safety function. Real redundancy comes from diversity, not from stacking more of the same sensor.

MD: How does certified real-time 3D spatial awareness enable more flexible human-robot collaboration without compromising compliance with standards such as ISO 13849 and IEC 61508?

KS: Because the safety function has a known, rated performance and a bounded response time, you can calculate safe separation and speed precisely. Add the third dimension and you’re accounting for the whole working volume rather than a single plane. That’s what creates the headroom: You can reduce separation distances, raise speeds and reshape or remove zones while staying firmly inside the determinism those standards demand.