Information Theory Based Quantitative Complexity Analysis for Uncrewed Aerial Vehicle Safety Testing Scenarios

Uncrewed Aerial Vehicles (UAVs) are becoming increasingly prevalent, raising significant challenges in ensuring their safe and reliable operation through scenario-based testing. Traditional methods for evaluating scenario complexity often rely on basic metrics, limiting their effectiveness when predicting UAV interactions in complex and dynamic environments. A critical open research challenge is the accurate quantification of scenario complexity to enable efficient and targeted testing of UAV systems. In this paper, we introduce a novel complexity estimation methodology based on an energy-based model, incorporating a quadcopter kinematics approach to generate realistic future flight trajectories. Our approach represents trajectories as polyhedrons, effectively modeling the spatial envelopes occupied by each UAV actor within test scenarios. Validated through experimental scenario testing, this method demonstrates improved accuracy and utility in quantifying complexity compared to traditional metrics. By precisely identifying and assessing complex interactions among multiple actors, our approach allows researchers and engineers to prioritize critical scenarios, significantly enhancing testing efficiency and UAV operational safety. This methodology not only advances scenario-based UAV validation, but also provides a broadly applicable framework for complexity estimation in other automated system contexts. More broadly, our research facilitates safer integration of UAVs into increasingly complex airspace, supporting the future scalability and regulatory acceptance of autonomous aerial operations.