Research on Map-SLAM Fusion Localization Algorithm for Unmanned Vehicle

Vision-based localization techniques and detection technologies are key algorithms for the navigation of unmanned vehicles. Especially in scenarios where GPS signals missing, Simultaneous Localization Mapping (SLAM) that rely on vision, inertial system (INS) other sensors have important applications. Among them, vision combined with IMU SLAM has advantage realistic scale, which is lacking monocular computational power compared to multi-visual so it suitable application an vehicle system. In this paper, we propose a fusion algorithm combines visual-inertial map road information, processing information under structured roads, detecting lane lines locating its local position by camera, applying strategy prediction update map-SLAM localization. It solves problem accumulating errors pure without loopback provides accurate global-local positioning results navigation.