Machine-optimized parametric computer assisted design (CAD) of a stable flying wing UAV glider

Abstract Designing a statically stable flying-wing UAV glider is a highly nonlinear problem in which small geometric changes cause large shifts in stability and performance. While prior work has applied gradient-based and gradient-free methods to flying-wing aerodynamic optimization, most approaches lack direct parametric CAD integration, ignore stability constraints, or rely on low-fidelity aerodynamic models, and the resulting designs are rarely built and validated in flight. We present an automatic gradient-descent optimizer that directly drives a FreeCAD parametric model of a flying-wing UAV, evaluating aerodynamics at each iteration with Reynolds-Averaged Navier–Stokes (RANS) CFD in OpenFOAM. Longitudinal stability is treated as the primary design driver, while lift balance and endurance act as guardrails that keep the design flyable. After ten iterations the optimizer reduced the total objective from 102.0 to 10.3, meeting the stability targets with a static margin of $$11.02\\%$$ 11.02 % and a trim moment of $$8.22 \\times 10^{-2}\\,\\text {Nm}$$ 8.22 × 10 - 2 Nm , while the secondary power and lift targets were not fully met. Most importantly, a prototype built from the optimized geometry was flight-tested and demonstrated both static and dynamic longitudinal stability, with a well-damped short-period mode and a lightly damped but stable phugoid, experimentally confirming the predicted stability.