Search papers, labs, and topics across Lattice.
This paper introduces SceneGTMM, a novel dual-graph interaction framework that integrates a GNN and Transformer to enhance map matching by addressing challenges in noise robustness and cross-regional transfer. By employing a conformal mapping-based scene relative strategy, the framework constructs trajectory-centric local coordinate systems, enabling dynamic updates and reducing reliance on specific training road networks. Experimental results demonstrate that SceneGTMM achieves over 80% accuracy on multi-source trajectories, significantly outperforming existing methods in cross-city transfer scenarios while improving interpretability through attention mechanisms.
Achieving over 80% accuracy in map matching, SceneGTMM not only outperforms traditional methods but also enhances interpretability, paving the way for more robust autonomous driving systems.
Map matching is a key technology connecting positioning data with high precision road networks, but it faces challenges in noise robustness, cross regional transfer, and interpretability. To addr ess the limitations of existing methods in local global fusion, dynamic road network adaptation, and reliance on black box mod els, this paper proposes SceneGTMM, a transferable GNN Transformer dual graph interaction map matching framework based on a conformal mapping based scene relative strategy. 1) Conformal mapping based scene relative strategy: constructs trajectory centric local coordinate systems to reduce dependence on the training road network, supporting cross regional transfer and dynamic road network updates; 2) GNN Transformer dual graph interaction architecture: a GNN modeled road graph captures local topological constraints, while a Transformer modeled trajectory graph captures global temporal dependencies, and cross graph attention achieves noise suppression and semantic alignment; 3) CRF enhanced structured prediction: combines the global context of the Transformer with the topological transition constraints of CRF to improve path connectivity and robustness. Experiments show that SceneGTM achieves over 80% accuracy on multi source trajectories with positioning errors of 16 50 meters, representing a 5.3% improvement over HMM. In cross city transfer scenarios, it outperforms MTrajRec, GraphMM, and TMM, and enhances interpretability through attention and relative coordinate visualization. This study provides a new paradigm for high precision, transferable map matching for real time traffic perception and autonomous driving path planning.
