Abstract
Hidden Markov Model (HMM)-based map matching algorithm presents several significant limitations. First, the candidate road segment generation process heavily depends on geometric features while largely ignoring semantic attributes and spatiotemporal context of the road network. Second, the probability modeling phase often fails to account for drivers' individualized road selection preferences. To address these issues, this study proposes an improved HMM-based map matching algorithm that incorporates drivers' personalized road selection preferences (PP-HMM). In the candidate road segment generation stage, a multi-dimensional fused scoring function is constructed by integrating spatial distance, directional similarity, road segment semantic attributes, and temporal factors, enabling more accurate ranking and selection of candidate segments. Moreover, by extending the state transition and observation probabilities of the HMM framework, the proposed method integrates various drivers' personalized road selection preferences observed during actual route selection, including basic route attribute preferences, road network structural characteristics, driving behavior traits, and temporal dynamics, thereby establishing a more comprehensive transition probability model. Experimental comparisons with traditional ST-HMM algorithms demonstrate the enhanced performance and robustness of the proposed approach across diverse road network environments.