Abstract

In recent years the way in which road safety is being analyzed is changing. It is intended to make designers and drivers share responsibility for it. Thus, self-explaining roads, where drivers are expected to be induced by the road itself and adopt behavior consistent with design and function, arise. Effective measures to adopt these design criteria would imply a reduction in the incidence risk resulting in safer roads. This paper focuses on consistency measures used to evaluate the consistency between the design and the specified speed of different roads. The disadvantages presented by some of the existing measures are analyzed, in particular, a dimensional analysis of the formulas used is performed. The properties that they should have are discussed and new measures, based on a Bayesian network model for probabilistic road safety analysis that integrates different variables, such as the driver's attention and tiredness, speed, road geometry and traffic intensity are proposed. Finally, some examples of application are given to illustrate the proposals.