Expression and Construction of a Multi-layer Map for Indoor Human-Robot Co-existing Environments

Aiming at the demand for map expression in the human-robot co-existing environments, a multi-layer map including semantic grid map, semantic topological map and semantic concept map is proposed and constructed. Firstly, 2D LIDAR and RGB-D sensors are used to construct object semantic grid map combined with SLAM and object semantic segmentation method. Semantic topological map is then constructed using virtual door detection and Bayesian estimation method based on room and object containment relationships. Finally, according to the semantic grid and the semantic topological map, a seven-step ontology creation method is used to build a semantic concept map. The effectiveness of the proposed multi-layer map is verified by testing in multiple scenarios.