Building a grid-point cloud-semantic map based on graph for the navigation of intelligent wheelchair

Abstract

The automatic navigation of the intelligent wheelchair is a major challenge for its applications. Most previous researches mainly focus on 2D navigation of intelligent wheelchair, which loses many useful environment information. This paper proposed a novel Grid-Point Cloud-Semantic Multi-layered Map based on graph optimization for intelligent wheelchair navigation. For mapping, the 2D grid map is at the bottom, the 3D point cloud map is on the grid map and the semantic markers are labelled in them. The semantic markers combine the name and coordinate value of object marker together. For navigation, the wheelchair uses the grid map for localization and path planning, uses the point cloud map for feature extraction and obstacle avoidance, and uses the semantic markers for human-robot vocal interaction. A number of experiments are carried out in real environments to verify its feasibility.