Coordinated adaptive front-lighting and road-lighting control for driver visibility optimisation on curved roads: A vehicle-engineering simulation framework

Night-time visibility on lit roads depends on the combined contributions of road lighting and vehicle headlamps, although these systems are generally designed and controlled separately. This study develops a driver-centred simulation framework coupling measured LM-63 roadway-luminaire photometry, a declared representative low beam, the Adrian small-target visibility model, closed-loop vehicle dynamics on curved roads, and coordinated adaptive front- and road-lighting control. Safety is assessed using detection distance relative to stopping sight distance. For the declared GE installation and low-beam model, target and background luminance balance at 66.1 m, causing visibility to collapse. A beam-shape sweep shows that this distance depends mainly on the low beam’s vertical extent: reducing either vertical Gaussian width by 20%–30% moves the balance point to approximately 47 m, whereas beam magnitude, horizontal width and cut-off-gradient changes have little effect. Contrast inversion persists in all tested perturbations. Across a 20-cell radius-speed campaign, coordinated model-predictive control increases the representative-cell above-threshold fraction from 10.9% to 43.2% while reducing lighting energy by 59%. At 40 km·h −1 , it eliminates unsafe instants (minimum margin +1.1 m) with 63% lower energy. A same-installation ablation shows that monitoring visibility only at the stopping point reduces detection distance by 17%, whereas a multi-range objective avoids this effect. The controller solves in 19.3 ms per 10 Hz cycle, remains stable over 0–500 ms latency and 0%–10% packet loss, and outperforms the baseline above-threshold fraction in 75.4% of 1000 Monte Carlo draws. Benefits are metric-dependent: threshold availability and minimum-margin ranking improve in most draws, but mean unsafe-time fraction does not. These results are simulation-based and conditional on the declared assumptions and installation.