Rapid Identification and Detailed Evaluation of UAS Encounter Scenarios through Integrated Simulation

Ensuring reliable collision-avoidance (CA) performance is a critical challenge for the safe integration of Unmanned Aircraft Systems (UAS) into increasingly dense and dynamic airspace environments. This paper presents a flexible, scalable multi-fidelity simulation framework for evaluating CA algorithms across a broad range of encounter geometries and operating conditions. The framework couples a lightweight, configuration-driven particle-based simulation with a high-fidelity environment capable of modeling realistic sensing effects, dynamic constraints, and actuator behavior. The low-fidelity module enables rapid exploration of thousands of head-on horizontal encounter scenarios by varying drone speed, horizontal turn rate, and response distance while automatically identifying cases that violate horizontal or vertical separation minima. To validate the fidelity transition and assess model limitations, representative scenarios were re-simulated in Isaac Sim using a various response distance and increasing drone speeds. The low-fidelity simulation predicted separation violations in less scenarios than the high-fidelity, which indicated more violation cases under identical initial conditions. This shift demonstrates that low-fidelity kinematic models may overestimate safety margins by simplifying sensor latency, turn-rate limitations, and timing effects that become increasingly significant at higher speeds. The results highlight the importance of multi-fidelity validation for UAS CA development and demonstrate the framework’s capability to efficiently identify edge cases while providing realistic insight into their viability under physical constraints. The proposed approach supports more reliable safety envelope definition, informs system performance requirements, and offers a promising foundation for pre-certification evaluation of autonomous UAS operations.