PVL-Cartographer: Panoramic Vision-Aided LiDAR Cartographer-Based SLAM for Maverick Mobile Mapping System

The Mobile Mapping System (MMS) plays a crucial role in generating accurate 3D maps for a wide range of applications. However, traditional MMS that utilizes tilted LiDAR (light detection and ranging) faces limitations in capturing comprehensive environmental data. We propose the PVL-Cartographer SLAM (Simultaneous Localization And Mapping) approach for MMS to address these limitations. This proposed system incorporates multiple sensors to yield dependable and precise mapping and localization. It consists of two subsystems: early fusion and intermediate fusion. In early fusion, range maps are created from LiDAR points within a panoramic image space, simplifying the integration of visual features. The SLAM system accommodates both visual features with and without augmented ranges. In intermediate fusion, camera and LiDAR nodes are merged using a pose graph, with constraints between nodes derived from IMU (Inertial Measurement Unit) data. Comprehensive testing in challenging outdoor settings demonstrates that the proposed SLAM system can generate trustworthy outcomes even in feature-scarce environments. Ultimately, our suggested PVL-Cartographer system effectively and accurately addresses the MMS localization and mapping challenge.

Automated summary

The Mobile Mapping System (MMS) plays a crucial role in generating accurate 3D maps for various applications. However, traditional MMS with tilted LiDAR faces limitations in capturing comprehensive environmental data. To address these limitations, we propose the PVL-Cartographer SLAM approach for MMS, which incorporates multiple sensors for reliable and precise mapping and localization. The proposed system consists of two subsystems: early fusion and intermediate fusion, which simplify the integration of visual features and merge camera and LiDAR nodes using a pose graph, respectively. Comprehensive testing shows that the proposed SLAM system can generate trustworthy outcomes in feature-scarce environments.