隧道段自动驾驶车道偏移预测方法OA

To address the issue of onboard vision perception caused by dynamic lighting conditions in tunnels,this paper proposes a lane departure prediction method for autonomous vehicles based on an adaptive Kalman filter.A coupled vehicle kinematics-vision perception model is established,integra-ting time-series data of multiple state variables such as lateral displacement and heading angle.Kalman filter is employed to achieve dynamic prediction.An adaptive adjustment mechanism is further designed to optimize filter parameters in real time,mitigating measurement noise induced by light flicker.Tests were conducted in tunnel scenarios constructed using the Prescan-Matlab/Simulink co-simulation platform.The results indicate that,compared with conventional detection methods,the proposed method provides earlier warning,higher correction efficiency,and adaptive capability to lighting fluctuations,demonstrating the robustness of the Kalman filter in complex tunnel lighting envi-ronments.