Driving simulator for road safety design: a comparison between virtual reality tests and on-field tests

Virtual reality simulations conducted by driving simulators represent a methodology to assess both the quality of road design and road safety in a safe, controlled, and replicable environment. Nowadays, there are numerous studies that use driving simulators to analyze the driver's response when specific road safety treatments are planned before these are implemented. This approach allows the road designer/scientist to estimate the potential safety effectiveness of the countermeasure/design configuration considered. However, although virtual reality simulations are potentially extremely useful in the evaluation of road configuration design and treatments effectiveness, they also have cons. The two most important are the limitations in the reproducibility of the real world environment and the difference in drivers’ behavior due to the awareness that they are conducting a test. In this context, our research aims to overcome these limitations through a robust validation procedure designed to demonstrate that the results obtained in virtual reality are reliable and can be exploited to design safer roads. According to the state of the art, the statistical procedure adopted allows the comparison between data collected during virtual reality experiments with those collected in on-field experiment using an instrumented vehicle. The on-field test was conducted a few years after the implementation of the safety measures evaluated by means of a driving simulator experiment. Statistical analyses were conducted to compare the results of the two experiments to determine if the differences between them are more likely to arise from random chance or not, to demonstrate the reliability of the virtual simulations and to identify the main limitations in exploiting the results. The procedure was repeated on the road section affected by the reconfiguration intervention analyzed in virtual reality, using the same validation procedure usually adopted to validate a driving simulator before the implementation of a safety measure. The procedure demonstrates the relative validity of the virtual reality experiment and, in some road segments, also the absolute validity of the results obtained. It confirms the ability of the driving simulator to be used in the preliminary assessment of the effectiveness of the designed safety solutions.