A Novel View Planning With Joint Optimization for Efficient 3D Building Inspection
Tenglong Zhang, Guoliang Liu, Guohui Tian
Abstract
With the development of drone technology, drone-based large-scale building facade inspection has become a key method for urban management and building safety assessment. Existing view planning methods suffer from redundant viewpoints and fail to adequately consider the coupling between viewpoint planning and path planning, resulting in low inspection efficiency. To address these limitations, we propose a view planning method combining hybrid clustering and projection-covering optimization. Our approach first clusters 3D model facets based on normal vectors and spatial density, then projects them onto 2D planes to generate candidate viewpoints through inverse projection, ensuring uniform coverage with fewer viewpoints. We introduce a multi-objective evolutionary framework (MAEF) to jointly optimize viewpoint selection and path planning, effectively solving the Set Covering Traveling Salesman Problem (SCTSP). Experimental results demonstrate that our method improves coverage completeness, reduces flight path length, and enhances inspection efficiency compared to conventional approaches.
BibTeX
@inproceedings{ral2026_anovelviewplanni,
title = {A Novel View Planning With Joint Optimization for Efficient 3D Building Inspection},
author = {Tenglong Zhang and Guoliang Liu and Guohui Tian},
booktitle = {RA-L 2026},
year = {2026}
}