Drivable Area Extraction from Three-Dimensional Map Considering Step Climbing

Abstract

This paper proposes a method of extracting areas where wheelchair users can drive from a three-dimensional map of the environment. The proposed method is characterized by its ability to detect steps that can be overcome by wheelchairs as the drivable areas in addition to flat ground. By an experiment using a drone and LiDAR, it is demonstrated that the proposed method can detect normal steps as drivable areas and convex-shaped steps, which are insurmountable by wheelchairs, as non-drivable areas. The proposed method can be easily applied to path planning for vehicles, mobile robots, and pedestrians besides wheelchairs.