VAC:Enhanced Visual-LIDAR Fusion SLAM Framework using Square Marker

Warehouses, as a unique degraded scene, pose challenges for accurate localization due to their spaciousness and lack of distinctive features. In this paper, we propose Visual localization assisted Cartographer(VAc), a LIDAR-visual-Inertial SLAM system aim to address issues in existing methods, including map distortion, loop closure detection failures and localization loss in long corridors. Our system improves the mapping and localization performance in feature-scarce scenarios by facilitating collaboration between the visual system based on Aruco Markers and the LIDAR system. We propose an optimization-based localization algorithm utilizing Markers and a monocular camera. Additionally, we introduce a multi-sensor pose fusion method, this method leveraging pose increment, integrates the visual Marker localization system into the LIDAR framework, jointly optimizing visual residuals and LIDAR residuals, forms a coherent representation of the environment. This fusion approach significantly improves overall robustness and accuracy.