Text
When Smart Systems Hesitate, Things Go Wrong — HERO Makes Them Predictable
In safety‑critical technology, a millisecond can be the difference between success and failure. Autonomous machines, industrial robots, and connected vehicles may make the right decision — but what happens if they make it too late? The HERO research group is turning this challenge into an advantage: exposing unseen delays, making complex systems predictable, and ensuring intelligence holds up under real‑world pressure.
It’s not enough for smart systems to make the right decision—they have to make it in time. When milliseconds matter, even “intelligent” systems can fail.
Autonomous vehicles, industrial robots, and connected machines are designed to act instantly. But what if they hesitate—just for a fraction of a second?
In safety-critical systems, timing is everything. A delay too small to notice can still be large enough to cause failure: a missed brake, a delayed shutdown, a system that reacts—but too late. As modern systems grow more complex—combining AI, multiple processors, and fast communication technologies—the risk is no longer just that they make the wrong decision, but that they make it at the wrong time.
Recent research shows that these timing uncertainties are not rare edge cases—they are built into the very complexity that makes systems powerful. When different technologies interact, delays can emerge in unexpected ways, making behaviour difficult to predict even for experienced engineers.
So the real question is no longer: Can we make systems intelligent?
It is: Can we trust them to act in time?
At Mälardalen University, the Heterogeneous Systems – Hardware–Software Co-Design (HERO) research group is tackling this challenge head-on. Their work focuses on making complex systems predictable before they reach the real world.
By combining hardware–software co-design, model-based engineering, and advanced timing analysis, HERO develops methods that expose hidden delays and performance risks early in development—before they become costly or dangerous failures. In collaboration with industry partners such as HIAB, ABB, Volvo CE, Scania, Arcticus Systems and Westermo, the group has already enabled reliable communication across mixed technologies like 5G and in-vehicle networks, making remote operation of heavy machinery a reality.
At the same time, HERO is pushing the boundaries of AI in safety-critical systems, developing privacy-aware and distributed learning methods that allow systems to collaborate without sharing sensitive data.
The result is a shift in how we think about smart systems:
not just as intelligent, but as predictable under pressure.
Because in the real world, intelligence is not enough.
If a system cannot act at the right time, it cannot be trusted at all.