Semantic Region Aware Autonomous Exploration for Multi-Type Map Construction in Unknown Indoor Environments

Abstract: Mainstream autonomous exploration methods usually perform excessively-repeated explorations for the same region, leading to long exploration time and exploration trajectory in complex scenes. To handle this issue, we propose a novel semantic region aware autonomous exploration method, the core idea of which is considering the information of semantic regions to optimize the autonomous navigation strategy. Our method enables the mobile robot to fully explore the current semantic region before moving to the next region, contributing to avoid excessively-repeated explorations and accelerate the exploration speed. In addition, compared with existing au?tonomous exploration methods that usually construct the single-type map, our method allows to construct four types of maps including point cloud map, occupancy grid map, topological map, and semantic map. The experiment results demonstrate that our method achieves the highest 50.7% exploration time reduction and 48.1% exploration trajectory length reduction while maintaining >98% exploration rate when comparing with the classical RRT (Rapid-exploration Random Tree) based autonomous exploration method.