Enhancing Safety-Related Lane-Change performance of novice drivers at freeway exit diverges using UAV-Calibrated pavement guidance.
Abstract
Freeway diverge areas require complex decisions under tight spatiotemporal constraints, often resulting in late-initiated maneuvers among novice drivers. To evaluate how pavement guidance markings influence diverging behavior, this study proposes a two-layer lane-change evaluation framework combining longitudinal spatial feasibility, represented by the Dynamic Minimum Diverging Length (DMDL) model, with lateral trajectory consistency, represented by a fitted trajectory envelope. The framework was calibrated using naturalistic unmanned aerial vehicle (UAV) trajectories, followed by a driving-simulator experiment with novice drivers under different lane configurations and speed limits. Results show that pavement guidance markings improved lateral vehicle control, with significant reductions in lateral-acceleration magnitude and variability and steering-wheel-angle magnitude and variability, while changes in longitudinal speed behavior were limited. Linear mixed-effects modeling further showed a shift from late, concentrated corrections toward earlier and more gradual adjustments, providing greater longitudinal space for trajectory stabilization. Sensitivity analysis yielded a maximum DMDL estimate of 675.6 m within the examined parameter ranges, supporting approximately 700 m as a conservative, scenario-specific reference for advance guidance placement.