Performance indexes for characterizing driving behavior

In infrastructure environments, the need for continuous monitoring of road safety has become paramount. However, current road geometric design rules often rely on theoretical assumptions that do not accurately reflect real-world driving dynamics. For example, the calculation of lateral acceleration and steering is performed assuming that the driver follows the center line of the lane slavishly. It is known, however, that vehicle trajectories are always very different from the axis line and the deviation from this makes many variables linked to vehicle dynamics critical for safety. This paper addresses this gap by proposing and validating 14 new indexes related to different performance variables (steering angle, steering speed, longitudinal acceleration and speed) in some critical points of a curve based on a comparative analysis between the ideal trend of a specific performance variable and their observed values. The study utilized a driving simulator to replicate a rural road (type F, according to Italian standards), facilitating data collection. Alongside calculating the performance indexes, some two-way ANOVAs were applied to better comprehend the significance of the proposed indexes. The aforementioned indexes have shown statistical significance in relation to the geometric characteristics of the curves (direction and radius) and are suitable for summarizing the driving behaviour in great detail. These performance indexes hold potential interest for infrastructure managers as they prove valuable and user-friendly in identifying particularly vulnerable road sections, thereby contributing to enhanced user safety. • Road design assumptions often differ from actual driving behavior, causing safety concerns. • Driving simulators replicate real roads, providing risk-free statistical analysis. • 14 new performance indexes assess workload at critical curve points for safer roads. • The procedure offers road managers a reliable tool for better transportation planning.