One of the visual tasks during night-time driving is the timely detection of potential hazardous objects in the road scene. Often these objects are observed at dusky (mesopic) light levels, especially when they appear in the periphery of the visual field. Since comprehensive methods for the measurement of mesopic light levels (luminances between 0.01 and 10 cd/m²) are not available yet, light levels in a traffic environment are typically measured using light meters provided with spectral sensitivities that are only valid for the photopic region (>10 cd/m²). Therefore methods are developed in the MOVE-project (Mesopic Optimisation of Visual Efficiency) which have proper spectral sensitivity functions for mesopic light levels. As a part of this project we have performed target detection experiment in a driving simulator. This report describes a short review on mesopic literature and the first limited simulator experiment that was meant for testing the experimental method. The experiment was performed in a high fidelity driving simulator. While subjects (3) drove on a winding road they had to respond to randomly presented circular targets, which appeared at various eccentricities (-30, -15, -7.5, +7.5, +15, and +30 degrees). The target diameter was 2 degrees and the (positive) target contrast was 0.2 or 0.8. The background luminance ranged between 0.0015 and 10 cd/m2 with the colours white, yellow, and blue (the same for target and background). The target detection performance was measured in terms of reaction time and the number of missed targets. The driving speed was fixed at two levels (80 or 100 km/h) by a cruise control system. Various measures of driving behaviour, derived from the position of the car at the road and the steering wheel angle, were recorded during the runs. The target detection performance becomes poor for low background luminances, a low contrast, and large target eccentricities. Blue light is more effective than yellow light in conditions of low light levels and large eccentricities. This is in accordance with the results of former target detection experiments in a driving context. There are indications that at a larger driving speed the subjects focus on a smaller part in the centre of the visual field. This study shows that it is possible to perform an experiment in the driving simulator at mesopic light levels and measure target detection performance and driving behaviour. The results can be used for the testing of mesopic models and justify extension of experiments in the driving simulator with more subjects and additional, adapted conditions. (Author/publisher)
Abstract