Driver Perception of Crest and Sag Combinations at the Driving Simulator: Effects on Drivers’s Behaviour

Several researches have pointed out that optical illusions may occur when the horizontal curve is combined with a vertical curve. Crest and sag combinations can cause a significant different driver’s behaviour compared to the behaviour adopted on the horizontal curves on flat grade. In particular the literature reports the hypothesis that horizontal curves appear sharper or flatter when overlapping with crest o sag vertical curves, respectively. This hypothesis was validated from studies on the visual perception of the road through the use of computer animation techniques. On the basis of these researches models were proposed to estimate the horizontal radius perceived on the combined curve as a function of the actual horizontal radius to evaluate the operating speed through 2-D predicting models. A drawback of computer animation techniques though is that they are limited and not interactive. Interactive driving simulation systems, on the contrary, have high potentials for the analysis of the effects of the combined alignment. An experimental survey at the driving simulator of CRISS (Interuniversitary Research Center on Road Safety) was carried out in order to assess the driver’s behaviour on crest and sag combinations, compared to the behaviour on horizontal curve with the same radius but on flat grade. Two road alignments for the experiment were designed according to the technical Italian guidelines on the coordination of horizontal curves overlapping with vertical curves. One alignment had a flat longitudinal grade, the other had three crest combinations and three sag combinations but with the same horizontal alignment. The vertical grade on the approach tangent of each combined curve was flat in order to avoid the influence of the grade on driver’s behaviour. 35 drivers carried out two driving sessions at the simulator (one for each road alignment) during which the local speeds and the local lateral placement were recorded. On the crest combinations the result of statistical analysis of the speeds was entirely consistent with the hypothesis of perception on crest combinations: at beginning of the circular curve on the reference curves the speeds are greater than those on the correspondent crest combinations. Furthermore the outcomes of the unilateral Z-test showed that the differences between the speeds on reference curves and those on crest combinations tend to decrease as the curve radius increases. Also the influence of the crest vertical curve on the trajectory was clear: the mean deviations from the ideal trajectory at the beginning of the circular curve with radii medium and great were significantly higher on the crest combinations than the mean deviations recorded on the reference curves. On the sag combinations the speeds were not significantly higher (at 5% significance level) than those on the correspondent reference curves. Such result does not seem to strengthen the hypothesis of perception on sag combinations and suggests that the models to estimate the horizontal radius perceived on the sag combinations as a function of the actual horizontal radius to evaluate the operating speed through 2-D predicting models are inappropriate. No significant effect of the sag vertical curve was observed on the trajectory.