The Use of Drones for Monitoring the Condition of Railway Infrastructure—Techniques, Applications, and Implementation Barriers

Unmanned aerial vehicles (UAVs) have become an important platform for acquiring data on the technical condition of railway infrastructure, enabling assessment without track possessions or exposure of personnel to traffic. This article critically reviews the state of knowledge on UAV-based monitoring of railway infrastructure in 2020–2026, assesses the technological maturity of individual applications and identifies research gaps and implementation barriers, with particular attention to Polish conditions. The review is based on a structured literature search and covers 62 sources: 36 journal articles indexed in Journal Citation Reports (JCR) and Scopus, eight book chapters, two peer-reviewed conference papers, one doctoral dissertation, and 15 institutional and normative documents; maturity was rated on the technology readiness level (TRL) scale. Perception tasks—track detection and segmentation, Global Navigation Satellite System (GNSS)-free visual navigation and 3D modeling of engineering structures—reach TRL 5–6, with deep-learning models achieving F1 scores of 76–96% and autonomous flight along the track already demonstrated. Metrological tasks remain at TRL 4–5: the best reported airborne track-geometry accuracy, a root mean square error (RMSE) of 0.8–1.6 cm, still does not meet diagnostic tolerances. Nine research gaps were identified, chiefly the absence of shared multi-line, multi-season reference datasets and validation protocols. UAV technology is today a mature documentation tool but only an emerging diagnostic one.