Benchmarking of Optimization and Heuristic Conflict Resolution Maneuvers for UAS Detect-and-Avoid Systems

Abstract

Uncrewed aircraft systems operating in shared airspace require detect-and-avoid capabilities that can systematically reduce the probability of mid-air conflict while remaining compatible with existing separation practices. As traffic density increases and mission profiles diversify, both optimization-based and heuristic conflict resolution maneuvers are being integrated into detect-and-avoid architectures to support timely and interpretable decision making under uncertainty. This paper examines a benchmarking methodology for such maneuvers in structured and unstructured airspace, focusing on horizontally and vertically coordinated resolutions subject to performance envelopes and surveillance latencies. The study defines a consistent encounter modeling framework, introduces several representative resolution algorithms, and evaluates their behavior over a wide set of geometrically diverse and operationally plausible conflicts. The benchmark emphasizes repeatable assessment of feasibility, computational burden, robustness to state and intent uncertainty, and sensitivity to sensing and communication constraints. Results are reported in terms of separation maintenance, maneuver aggressiveness, and path efficiency rather than in terms of any singular optimality claim. The analysis is constructed to highlight differences between convex and nonconvex optimization formulations and between rule-based and randomized heuristic schemes, under homogeneous comparison conditions. The overall objective is to contribute a technically transparent basis for comparing candidate detect-and-avoid maneuvering logics and to outline where different classes of algorithms exhibit consistent strengths, weaknesses, and trade-offs across a large ensemble of encounter situations.