Cooperative Localization in Vehicular Networks Based on Multimodal Information Fusio

To enhance vehicle localization accuracy in vehicular networks, this paper proposes a cooperative localization method based on multimodal information fusion. A hierarchical fusion localization framework is established by deeply integrating LiDAR point cloud data, camera visual information, and Global Navigation Satellite System (GNSS) signals. The method employs a dual fusion mechanism: (1) at the sensor fusion level, the three-dimensional spatial information from LiDAR is combined with the two-dimensional visual features from cameras to achieve robust vehicle detection and precise inter-vehicle distance measurement, effectively overcoming the perceptual limitations of a single sensor in adverse weather conditions (such as LiDAR attenuation in rain or fog) and the depth estimation errors of cameras; (2) at the localization fusion level, the inter-vehicle distance information is integrated with GNSS data to enhance vehicle localization accuracy further The mathematical model based on the maximum likelihood strategy can be transformed into a quadratic non-convex programming problem, and an iterative optimization algorithm is utilized to solve an approximate quadratic programming problem with linear constraints at each iteration, thereby obtaining the global optimal solution for the original problem. Experimental results demonstrate that the multimodal fusion approach significantly improves localization accuracy compared with traditional single-modal methods.