A New Clustering-based View Planning Method for Building Inspection with Drone

TL;DR

The paper tackles drone-based building inspection by introducing a clustering-based two-step view planning framework. It generates high-quality candidate viewpoints through spectral clustering and local potential field corrections, then solves a Set Covering Problem with a genetic-based hyper-heuristic (GA-HH) to select a minimal yet sufficient subset. Empirical results across multiple buildings and sensor configurations show the method reduces viewpoint requirements by around 20% while achieving full coverage and maintaining practical runtimes, outperforming random sampling. The approach advances practical 3D visual coverage by effectively leveraging geometry-aware candidate generation and robust search strategies, with implications for more efficient inspection workflows and future path-planning extensions.

Abstract

With the rapid development of drone technology, the application of drones equipped with visual sensors for building inspection and surveillance has attracted much attention. View planning aims to find a set of near-optimal viewpoints for vision-related tasks to achieve the vision coverage goal. This paper proposes a new clustering-based two-step computational method using spectral clustering, local potential field method, and hyper-heuristic algorithm to find near-optimal views to cover the target building surface. In the first step, the proposed method generates candidate viewpoints based on spectral clustering and corrects the positions of candidate viewpoints based on our newly proposed local potential field method. In the second step, the optimization problem is converted into a Set Covering Problem (SCP), and the optimal viewpoint subset is solved using our proposed hyper-heuristic algorithm. Experimental results show that the proposed method is able to obtain better solutions with fewer viewpoints and higher coverage.

Figures (10)

• Figure 1: The proposed view planning method.
• Figure 2: Processing the target model as a set of triangular meshes.
• Figure 3: Spectral clustering of a set of triangular meshes.
• Figure 4: Generated candidate viewpoints.
• Figure 5: The visibility model.