Intelligent UAV inspection for the full lifecycle of deep underground engineering:State-of-the-art,challenges,and future directions
ZHANG Ru
LYU You
ZHANG Zetian
HEI Xinhong
TU Haiyan
SONG Xiaogang
ZHANG Anlin
ZHENG Xiujuan
XIAO Kun
Abstract：Compared with shallow engineering,deep underground engineering is characterized by complex,coupled geological conditions,including high in-situ stress,high geothermal tem-perature,and high seepage pressure.Under the influence of engineering geological activities,these conditions exhibit long-term dynamic evolution,resulting in frequent hazards with high energy release that are difficult to control.Such challenges persist throughout the entire pro-ject lifecycle,spanning investigation,construction,operation and maintenance,and remedia-tion,and therefore call for a comprehensive inspection system covering the full lifecycle of deep underground engineering.In this context,Unmanned Aerial Vehicles(UAVs),leveraging three-dimensional mobility and integrated capabilities in environmental perception,localization and mapping,and flight control,can provide an intelligent solution for integrated management of safety,quality,and cost.Accordingly,this paper systematically reviews the research pro-gress in intelligent UAV inspection technologies for deep underground engineering,and the main findings are summarized as follows.Intelligent UAV inspection is systematically applied to perform five categories of tasks,namely modeling and geological analysis,structural health monitoring,environmental monitoring,production progress monitoring,and emergency re-sponse.An imaging,three-dimensional,and temporal data processing framework is also sum-marized.Through within-domain processing and cross-domain fusion,raw data are trans-formed into state indicators and risk levels that support engineering decision making.A sensor configuration logic driven by environmental factors is established,and a complete preprocess-ing chain covering spatiotemporal alignment,data cleaning,and quality assessment is summa-rized.This provides a solid foundation for maintaining stable perception of intelligent UAV in-spection systems in degraded environments.The reliability of UAV localization and mapping is emphasized to depend on a full-chain technical system capable of degradation self-diagnosis and global consistency maintenance.Robustness should be embedded in registration,odometry,loop closure detection,and back-end optimization,and geometric and semantic information should be integrated to suppress drift and mismatches.Multiple technical pathways are sys-tematically reviewed,ranging from macroscopic path search to microscopic real-time obstacle avoidance.It is indicated that a successful planning system should deeply integrate environ-mental uncertainty,dynamic constraints,and task semantics.Based on insights into current technical barriers,this paper further proposes specific development directions centered on plat-form sensing,localization and mapping,planning and obstacle avoidance,intelligent algo-rithms,computing collaboration,and communication and energy,aiming to achieve break-throughs through synergistic advances across multiple technological chains and to promote the application and expansion of fully autonomous intelligent inspection paradigms for deep under-ground engineering.
Keywords：deep underground engineeringintelligent UAV inspectionmultisensor fusionlocalization and mappingpath planning and obstacle avoidance
Publication Date：2025-11-30
Online Publishing Date：2026-01-09(First online date of this platform, not the publication date of the document)
Pages：31( 1326-1356 )