Towards self-explaining spiral tunnels: road-prototype-based facility design and visual-perception fluid assessment.

With increasing demand for spiral tunnels, drivers face spatial cognitive challenges due to strong geometry transitions. To address this issue, a self-explaining-road-based framework is developed, focusing on the system-level optimization of four categories of tunnel environmental facilities, including visual guidance facilities, traffic signs, overall environment, and roadway. These elements are systematically combined under a road prototype design approach to induce expected driving behavior. In addition, a visual-perception fluid (VPF) model is proposed to evaluate drivers' visual-perception performance in spiral tunnels. Four semantic categories were extracted via environmental semantic segmentation to compute equivalence mass of environmental information, information flow velocity, cognitive resistance, and composite VPF index. A driving-simulator experiment with 12 scenarios yielded 48 valid samples, and a hierarchical paired permutation test evaluated design effects. The prototype design increased composite VPF index by 1.9 (4.6%) relative to ordinary design, with significant improvements in structural feature parameters (p < 0.05), supporting the assessment capability of the model. Prototype designs consistently enhanced key semantic information (visual guidance facilities and roadway) and reduced cognitive resistance by 2.2% and showed favorable medium-term and long-term stability. Simultaneously, engineering optimization should prioritize entrance areas and the saliency design of visual guidance facilities and traffic signs.