Uncertainty Risk Assessment and Autonomous Vehicle Car-Following Control Considering Road Adhesion Performance

A decrease in the road adhesion coefficient significantly increases the rear-end collision risk of autonomous vehicles. However, most existing safety assessment indicators and car-following models ignore this adverse effect, making them unable to adapt to low-adhesion road scenarios. To address this issue, this study conducts research on uncertainty risk assessment and autonomous vehicle car-following control considering road adhesion performance. First, an Uncertainty Risk Indicator (URI) integrating the road adhesion coefficient and braking performance is constructed, and the worst-case scenario assumption is adopted to ensure the reliability of risk assessment. Second, combined with the Gipps model and risk homeostasis theory, a φ-Gipps car-following model adaptable to different adhesion roads is developed, which can balance safety and comfort through personalized risk expectations. Finally, numerical simulation tests are carried out under four different adhesion road scenarios (dry, wet, snowy, and icy) to verify the performance of the indicator and the model. The results show that compared with Time-to-Collision (TTC) and Deceleration Rate to Avoid the Crash (DRAC), the URI can maintain a reasonable safety threshold in all adhesion road scenarios, with the required deceleration within the range of road adhesion capacity and smooth motion responses. The φ-Gipps model is highly stable and reliable, which can ensure the car-following safety of autonomous vehicles on different adhesion roads. The research results provide a new theoretical method and technical support for the car-following safety control of autonomous vehicles on low-adhesion roads.