In environments with limited satellite signal availability,LiDAR Odometry (LO) can continuously provide stable pose estimation.However,its errors accumulate over time,affecting long term localization accuracy.To address this issue,an optimization algorithm for LiDAR localization based on road network constraints is proposed in this paper.Using LO trajectories and road network maps as core inputs,the method effective trajectory to map correspondences are constructed by establishing data associations for both straight and turning road segments.Based on a Kalman filter framework,the matched pairs are treated as observations,and the observation weights are adaptively adjusted according to the matching confidence,thereby correcting LO drift errors.This method features low cost and lightweight implementation,facilitating deployment on embedded front end systems.Experiments on public datasets and real vehicle platforms show that the proposed method can achieve lane level localization accuracy without relying on satellite signals,and significantly improves localization performance compared with traditional LO and other map matching methods.