TL;DR

• Current production DMS detects gaze, head pose, eye closure, and hands-on-wheel - point-in-time signals that answer what the driver is doing, not what state they're in
• Longitudinal tracking, emotional state detection, and real-world occlusion robustness are the primary capability gaps in existing systems
• Euro NCAP 2026 is raising the bar: impairment and cognitive distraction detection are now part of the scoring framework
• The hardware to do more already exists in most vehicles - the gap is in the software intelligence running on it

Modern Advanced Driver Assistance Systems (ADAS) has become genuinely impressive at perceiving the world outside the vehicle. Radar tracking objects at distance, cameras distinguishing pedestrians from cyclists in milliseconds, sensor fusion building real-time maps of everything around the car. The external perception stack has come an extraordinary distance in a short time. But point those systems inward - at the person behind the wheel - and the picture changes considerably.

What current ADAS and DMS systems actually detect inside the cabin

Most production vehicles today include some form of driver-facing sensing. What it captures varies by system, but the common baseline across the industry looks like this:

• Head pose - is the driver's head oriented toward the road or turned away?
• Eye closure - are the eyes open, partially closed, or shut? Used primarily to trigger drowsiness alerts
• Gaze direction - broadly, where is the driver looking? Forward, at a screen, at a mirror?
• Hands-on-wheel detection - particularly relevant in vehicles with lane-keeping or partial automation features

These are point-in-time behavioural readings.

They answer one question: what is the driver doing right now?

For basic regulatory compliance - meeting the minimum floor of what the EU GSR ADDW mandate requires - they are often sufficient. Common technologies used in production DMS as of today include near-infrared cameras, steering sensors for hands-on detection, and in some cases radar modules.

The limitation becomes clear when you ask a different question entirely.

The gap: what state is the driver actually in?

What a driver is doing and what state a driver is in are different things. A driver can be looking straight ahead - head forward, eyes open, hands on the wheel - and still be cognitively overwhelmed, emotionally dysregulated, or in the early stages of a fatigue episode that has been building for an hour. Point-in-time behavioural signals will not catch any of that.

Current ADAS systems notably operate without taking into account drivers' states - for example, whether the driver is emotionally apt to drive. There is a recognised lack of advanced driver monitoring capability for ADAS that could increase driving quality and security for both drivers and passengers.

A few specific gaps worth naming clearly:

• Longitudinal state tracking. Most current systems do not compare a driver's behaviour against their own baseline over time. Whether blink rate, facial muscle tension, or micro-expression patterns have shifted during a journey - relative to how they started - is information that existing production systems largely do not capture or use.

• Emotional and physiological state. Beyond drowsiness, a broader range of states affect driving safety: acute stress, emotional dysregulation, chronic fatigue that compounds across days. Conventional driver monitoring systems frequently fail to detect emotional and cognitive states, limiting their potential to prevent accidents. The research direction is clear. Production deployment is still catching up.

• Real-world condition robustness. Systems that perform well in lab conditions frequently underperform in real cabin environments - variable lighting, sunglasses, partial occlusion from scarves or visors, different camera placements across vehicle platforms. Low-latency driver state monitoring and consistent performance across real-world conditions - including occlusion robustness - are becoming essential requirements rather than nice-to-haves.

• Occupant monitoring beyond the driver seat. In many current implementations, in-cabin sensing remains fragmented across ADAS, HMI, and passive safety domains, each with separate data pipelines and performance requirements.Full cabin awareness - not just the driver seat - is an architectural challenge most production systems have not yet resolved.

Why this gap matters now specifically

Impairment and cognitive distraction detection are being added to Euro NCAP ratings from 2026. The regulatory floor is rising. What was optional is becoming a scored requirement, and what was a scored requirement is becoming a determinant of 5-star ratings.

The industry is not short of sensing hardware. Most modern vehicles already carry the camera needed to do significantly more than current systems do. The constraint is not the hardware - it is the software intelligence sitting on top of it, and specifically whether that software can do more than read a moment in time.

That question - what does it take to genuinely understand the human in the cabin - is what the next article in this series addresses directly.

FAQs:

• What does a driver monitoring system currently detect? Most production DMS today detects head pose, eye closure, gaze direction, and hands-on-wheel status - point-in-time behavioural signals that capture what the driver is doing at a given moment, not the underlying state they are in. They meet basic regulatory compliance requirements but fall short of detecting cognitive load, emotional dysregulation, or fatigue that has accumulated over time.
• Why can't current systems reliably detect driver fatigue or emotional state? Because they are built around single-session, point-in-time measurement. Detecting fatigue progression or emotional dysregulation requires comparing a driver's current behaviour against their own historical baseline - a longitudinal approach that most production systems do not currently implement. Real-world conditions including variable lighting, occlusion, and demographic diversity add further technical complexity that lab-validated systems often do not handle well in the field.
• What is the difference between driver monitoring and occupant monitoring? Driver monitoring focuses on the person behind the wheel - attention, alertness, gaze, head pose, and safety-relevant state signals. Occupant monitoring covers the full cabin - detecting presence, position, and characteristics of all passengers. Euro NCAP 2026 includes requirements for both, and they are increasingly being integrated into a unified in-cabin sensing architecture rather than treated as separate systems.