Towards low-cost systems for measuring visual cues of driver fatigue and inattention in automotive applications

Recent studies show that driver fatigue and inattention are major causes of fatal road accidents. A large number of these accidents could be avoided if the vehicles were equipped with a) sensors that reliably could monitor the attention and alertness cues of the drivers, and b) systems that, based on the sensor input, could warn the drivers in time. Today, camera-based systems can measure such attention cues, and are tested in automotive in-cabin installations with two or more cameras. Camera-based solutions are non-intrusive, they can measure head position and orientation as well as gaze direction and eyelid closure, and they can be used to identify the driver for other tasks. However until now, no camera-based system has met the requirements from the car manufacturers on compactness and cost. Here we present the principles behind Smart Eye AntiSleep, which is a compact one-camera system specially designed for automotive in-cabin applications. AntiSleep measures 3D head position, head orientation, gaze direction and eyelid closures. AntiSleep uses one camera and two IR-flashes. With its own illumination, AntiSleep is highly robust to all natural illumination conditions and can efficiently handle disturbing reflexes in eyeglasses. The system is fully automatic and requires no manual intervention during initialisation.