Online 3D Coverage Path Planning Using Surface Vector

Abstract

This paper presents an online coverage path planner for mapping the surface of three-dimensional structures. Recent works on mapping three-dimensional structural surface rely on random walks, volumetric cellular decomposition, and viewpoint optimization. Unfortunately, the approaches that utilize the volumetric cellular decomposition or the viewpoint optimization cannot perform their work online due to their heavy computational complexity, and online coverage path planners based on random walks show the inefficient spatial visiting, which leads to the failure of the coverage. In this paper, a novel non-randomized real-time coverage path planner that utilizes the newly updated surface is proposed to solve these problems. The proposed coverage planner detects the newly updated surface with an occupancy map and calculates the surface normal vector with k-nearest neighbors (K-NN) and principal component analysis (PCA). The ratio between the exploitation and map gathering can be balanced properly by planning the paths that consider the newly updated surface and surface normal vectors, which guarantees the increased pose estimation performance. By intensively simulating the proposed algorithm in challenging environments, the efficiency and the performance of the algorithm are validated in terms of coverage completeness.