Road design influence on driving behaviors: The influence of curve design, a case study

Road geometry has always been a key feature for road safety concerns. It will become more crucial in the context of future transportation, especially with the advent of Connected and Autonomous Vehicles (CAVs). In fact, recognizable and intuitive road alignments would simplify the driving tasks for both humans and CAVs (independently from the rate of automation). Thus, not only building consistent and self-explaining roads is fundamental for new and old vehicles, but also adjusting the existent ones, operation that seems even harder. Since most of the existing roads would need massive interventions, policy makers and road designers might choose between making adjustments being compliant with the current regulations in toto or adopting countermeasures supported by specific safety assessments to make existing roads safer, also in the perspective of future changes. In this optic, the present study tries to investigate a typical geometric design issue of existing roads, i.e., the presence of a long segment followed by a sharp curve without transition curves on undivided two-way two-lane rural roads. This alignment does not reflect the current recommendations for road alignment, so it was investigated the effect of such a design on users and safety for a specific testbed. The users’ behavior was investigated recording the kinematic parameters of the traveling vehicles. This data collection was run using radar traffic counters, placed on the roadside throughout the entire layout of the investigated segment-curve, to get speed and acceleration. The data were elaborated to investigate driving behavior in free-flow conditions. A K-means cluster analysis was run to characterize the users’ behaviors in terms of speed and acceleration. Hard braking was found to be strongly related to high speed, as well as ongoing deceleration on curve was detected for all the vehicles with high speeds on the segment. Results about users’ behaviors were compared to the available crash dataset to understand the possible implications of human factors on occurred crashes and to simulate the decision process of safety-related adjustments of existing roads.