An improved car-following model considering visual occlusion in heterogeneous traffic at signalized intersections

Large vehicles at signalized intersections often obstruct traffic signal visibility, which forces small vehicle drivers to rely solely on preceding vehicle movements and increases collision risks. Existing car-following models inadequately address this visual occlusion phenomenon. This study develops a modeling framework by introducing three key innovations into the full velocity difference (FVD) model. First, a binary occlusion function S based on sight distance triangles is established to determine when visual obstruction occurs. Second, the vehicle driving strategy with the yellow light phase in the dilemma zone is integrated into the car-following model by using two binary functions D and W. Third, a spacing-dependent speed difference adjustment mechanism is proposed, where drivers modify their following behavior according to four decision rules based on vehicular spacing and relative velocity conditions by γ. For the purpose of quantifying the impact of this mechanism on traffic safety at signalized intersections, we developed an FVD model that incorporates the occlusion phenomenon (FVD-OP) for simulation analysis. To mitigate safety risks, an FVD-based spacing guidance strategy model (FVD-SG) is further developed that maintains safe following distances by adjusting acceleration through the parameter μ when occlusion is detected. Simulation results reveal that large vehicle penetration rates above 15 % significantly destabilize traffic flow, with traditional models showing DRAC reaching 6.09 m/s² under 20 % penetration. The proposed FVD-SG model reduces collision risks by up to 56.65 % while maintaining traffic efficiency, and can significantly reduce the percentage of vehicles trapped in a dilemma zone from 14.92 to 2.21 %. Sensitivity analysis identifies optimal parameter ranges (γ≤1.0, μ≤0.2) that ensure system stability across varying traffic conditions, demonstrating the model's effectiveness in addressing visual occlusion challenges in heterogeneous traffic flow.