How We Study What Driver Intoxication Really Looks Like

When we think of intoxicated driving, we often picture the most obvious signs: slurred speech, weaving between lanes, racing way past speed limits. But that’s not always what it looks like in reality, especially early on.

This study focused on how driver behavior changes as impairment begins to set in.

At a closed track in Sweden, Smart Eye and The Swedish National Road and Transport Research Institute (VTI) invited participants to drive under controlled conditions, with varying levels of blood alcohol concentration — and in some cases, added fatigue.

Here’s what that looked like.

Why This Research Matters Now

There’s plenty of research on impaired driving, but far less that captures how behavior unfolds behind the wheel, in real time, under real conditions.

That kind of data is becoming more important as two things accelerate in parallel: new safety requirements, and the growing capability of in-cabin technology. Together, they’re pushing vehicles to do something they haven’t been able to do before — recognize risk as it’s happening, based on how the driver is behaving.

Euro NCAP will include impairment detection in its safety ratings starting in 2026. In the U.S. , lawmakers are working toward a mandate for passive alcohol detection in all new cars. And as sensing systems get smarter — and more widely deployed — what regulators and safety bodies expect from vehicles is changing fast.

But smarter technology doesn’t help much without the right input. If these systems are going to detect risk, they need to know what it actually looks like. That’s where studies like this one come in.

What The Study Set Out to Capture

If we want to detect impairment through behavior, we need to see how it actually shows up behind the wheel.

Participants drove under controlled conditions on a closed track, at blood alcohol levels ranging from 0.2‰ to 1.2‰. Some drove while intoxicated. Others combined intoxication with sleep deprivation. The focus was on what the cameras could pick up: subtle shifts in attention, delayed reactions, changes in how drivers moved, looked, and responded over time.

This kind of real-world, in-vehicle data is still rare, especially when collected under known and repeatable conditions. But it’s exactly the kind of reference material needed to train systems that aren’t just watching for lane departures, but for the quieter signs that something’s not right.