Driver fatigue sensor: how the camera that detects drowsy driving works
From the car’s built-in sensor, which reads the steering wheel, to the infrared camera that watches the eyes: what each one detects, what PERCLOS is and where the technology gets it wrong.
"Fatigue sensor" is the name of two different technologies, and mixing them up leads people to buy the wrong one. One comes from the factory in many cars and watches the steering wheel. The other is a camera that watches the driver’s face. This article explains how each one works, what it detects and where it fails.
The car’s sensor: the steering wheel gives fatigue away
In cars that come with a factory fatigue warning, the system usually analyzes driving behavior: small steering corrections, how consistently the car stays in its lane, how long the trip has gone on without a stop. When the pattern changes in a way typical of tiredness, the instrument cluster suggests taking a break, often with a coffee cup icon.
It is useful, but indirect. It notices the driving getting worse, not the eyes closing. And it tells a fleet manager nothing: the warning stays on the car’s dashboard.
The camera: the eyes give sleep away
The fatigue camera sits on the dashboard, facing the driver. It lights the face with infrared light, invisible to the driver, which is why it works at night and through most eyeglasses. A processor inside the camera itself tracks, frame by frame:
- Eyelid closure. An eye that closes and is slow to reopen, longer than a blink.
- Repeated yawning.
- A head that drops or nods forward.
- Gaze direction, which also serves to detect distraction and phone use.
The best-known eye-based measure of drowsiness is PERCLOS: the percentage of time, within a window, during which the eyelids cover most of the pupil. The higher it is, the drowsier the driver. Cameras combine this idea with the duration of each closure to decide when to alert.
What happens when it detects something
The main alert is in the cab, instantly: a sound, a voice or seat vibration. It does not depend on cellular signal, because detection happens inside the camera. In parallel, the event goes to the fleet platform with the time, the location and, when there is cellular signal, a short video of the moment. The manager sees the pattern: which driver, what time, which stretch of road.
Where the technology gets it wrong
- Sunglasses. Some lenses block infrared and hide the eyes. The camera still picks up head movement and yawning, but loses its main measurement.
- Poorly positioned camera. Wrong angle, a dashboard that vibrates, direct sunlight. Installation is half the result.
- Badly tuned threshold. Too sensitive, and it beeps for nothing until the driver covers the lens with tape. Too loose, and it misses the nod-off. The pilot phase is there to find the middle ground.
- Thinking the camera solves fatigue. It warns. What solves fatigue is rest, and what guarantees rest is the company’s driver schedule.
Is it mandatory?
That depends on the country. In the European Union, Regulation (EU) 2019/2144 made the driver drowsiness and attention warning mandatory on every new car, bus and truck from July 2024, and the advanced driver distraction warning, which watches the driver’s gaze, on every new vehicle from July 2026. It is a good indicator of where the technology is heading.
Where IoTData fits in
We manufacture the camera ourselves: board, detection firmware, integration with the tracker and the platform, alert rules, reporting and support. See IoT Safety.
Sources: Regulation (EU) 2019/2144 (DDAW for new vehicles from 07/07/2024; ADDW from 07/07/2026); literature on drowsiness detection using the PERCLOS measure.