A bayesian network approach for systemic risk analysis in unmanned aerial vehicle (UAV) operations

Abstract Unmanned Aerial Vehicle (UAV) operations confront complex systemic risks that challenge traditional analytical methods. This paper develops a hierarchical Bayesian Network (BN) to quantitatively model these risks. Our model establishes causal pathways from foundational drivers to key performance indicators (KPIs): Safety, Mission Success, and Third-Party Risk. The baseline risk assessment reveals significant operational vulnerabilities. It identifies degraded pilot performance, evidenced by a 54% probability of ‘Poor’ Decision Making, as a primary contributor to a 56% baseline probability of an ‘Accident’. However, a comprehensive sensitivity analysis demonstrates a more critical insight: the operational environment, specifically ‘Adverse Weather’ and ‘Terrain & Obstacles’, constitutes the single most dominant risk driver across all KPIs. This finding underscores the strategic importance of rigorous pre-flight environmental assessment over in-flight reactive measures. Furthermore, the analysis reveals the necessity for differentiated mitigation strategies; Mission Success exhibits unique sensitivity to ‘Signal Interference’, a factor less critical for direct safety outcomes. This framework provides a data-driven, causal tool to support UAV operators in resource prioritization and systemic resilience enhancement within a complex operational landscape.