Renault has investigated two approaches for developing an on-boargi system to warn of driver drowsiness. The approach is based on the use of mechanical signals (steering wheel angle, lateral position, etc), while the second is based on analysis of eyelid movements. The aim of this system is to warn the driver sufficiently early of the occurrence of micro-sleeps, because it is during these micro-sleep phases that the driver completely loses control of his vehicle. These moments of loss of control of the vehicle are characterized objectively by the simultaneous analysis of physiological signals and driving incidents recorded during tests on simulator and on the test circuit.

The study of mechanical signals has shown that the latter are too dependent on the geometric characteristics of the road and the kinetic characteristics of the car. According to us, such a system can only function reliably on motorways, on which driving characteristics are normalized to a maximum.

Partial analysis of thirty videos of the face, recorded during alertness tests, showed that certain eyelid parameters change when lapses of alertness occur. These observations are confirmed by numerous studies already performed on the subject. At present, we are studying the relevance of eyelid parameters for predicting driver drowsiness at the steering wheel. The initial results are positive, although they still need to be further detailed and backed up by more widespread testing.

We have developed an on-board system for automatic measurement of eyelid movements. We consider that the feasibility of this measurement is proven, but only in conditions of low luminosity (dawn, dusk, night).

Accident research data shows that a system operating by night on all types of road would be more elîective in terms of warning than a system operating on motorway both by day and by night. This is why Renault, Renault VI and Mack Trucks have favoured this approach and have planned a series of tests to validate the complete system.