Vision-based Situational Graphs Generating Optimizable 3D Scene Representations

Abstract-- 3D scene graphs offer a more efficient representation of the environment by hierarchically organizing diverse semantic entities and the topological relationships among them. In the context of Visual SLAM (VSLAM), especially when the reconstructed maps are enriched with practical semantic information, these markers have the potential to enhance the map by augmenting valuable semantic information and fostering meaningful connections among the semantic objects. In this regard, this paper exploits the potential of fiducial markers to incorporate a VSLAM framework with hierarchical representations that generates optimizable multi-layered vision-based situational graphs. The framework comprises a conventional VSLAM system with lowlevel feature tracking and mapping capabilities bolstered by the incorporation of a fiducial marker map. When the goal is to incorporate semantic data, it becomes possible to enrich VSLAM Employing vision sensors for Simultaneous Localization with high-level information about the environment [2], [3] and Mapping (SLAM) applications can bring about several Nonetheless, many of these approaches do not integrate merits, including the ability to achieve rich visual information valuable relational information among pertinent semantic using a low-cost hardware setup, making them attractive entities. This variant of SLAM systems is graphs from underlying SLAM.