Semantic Communities from Graph-Inspired Visual Representations of Cityscapes

The swift development of autonomous vehicles raises the necessity semantically mapping environment by producing distinguishable representations to recognise similar areas. To this end, in article, we present an efficient technique cut up a robot’s trajectory into consistent communities based on graph-inspired descriptors. This allows agent localise itself future tasks under different environmental circumstances urban area. proposed semantic grouping utilizes Leiden Community Detection Algorithm (LeCDA), which is novel and method low computational complexity exploits topometric information from observed scenes. presented experimentation was carried out dataset city Xanthi, Greece (dubbed as Gryphonurban dataset), recorded RGB-D, IMU GNSS sensors mounted moving vehicle. Our results exhibit formulation map with visually coherent realisation effective localisation mechanism for environments.