A Hierarchical Map Building for SLAM Used in Ruins Environments

As to the morphological characteristics of the seismic damage of the interior ruins, this paper presents a map building mechanism by using the analytic hierarchy process and proposes a hierarchical simultaneous localization and mapping (SLAM) algorithm that is based on the hybrid metric-topological map. The architecture of the ruins-oriented SLAM system consists of three layers. In the intermediate layer, the global environment is partitioned on the basis of the spectral clustering that can reduce the computational cost and ensure the global consistency. A semantic map is built in the top layer where the node identification and logic location are included. According to the characteristics of the seismic destruction, an aggregation degree of features is proposed as an identification method for building topological nodes to improve the interactivity. In the bottom layer, details of the morphology of the destruction are described as grid maps that ensure the environmental adaptability. Through experiments, the ability of environmental description and the availability at an artificial ruins environment are verified.