RESEARCH · RESEARCH · #1249
GeoWind2Plan: mission-time 3D urban wind prediction for energy-efficient UAV planning
GeoWind2Plan is a geometry-to-wind-to-planning framework that uses a localized geometry-conditioned neural operator to predict mission-relevant 3D urban wind patches from a background wind vector and 3D building geometry, stitches them into a local wind field, and optimizes a feasible 3D trajectory and speed profile with a UAV energy model. The paper (arXiv:2609.36056v1) reports corridor-localized inference in about 3 seconds versus roughly 8 hours for CFD, and under CFD evaluation trajectories planned with GeoWind2Plan reduce energy by 6.9% (tailwind), 12.7% (headwind), and 4.5% (crosswind) compared with wind-agnostic planning, recovering 87.9%, 85.7%, and 75.0% of CFD-reference savings respectively.
KEY POINTS
- GeoWind2Plan is a geometry-to-wind-to-planning framework that uses a localized geometry-conditioned neural operator to predict mission-relevant 3D urban wind patches from a background wind vector and 3D building geometry, stitches them into a local wind field, and optimizes a feasible 3D trajectory and speed profile with a UAV energy model.
- The paper (arXiv:2609.36056v1) reports corridor-localized inference in about 3 seconds versus roughly 8 hours for CFD, and under CFD evaluation trajectories planned with GeoWind2Plan reduce energy by 6.9% (tailwind), 12.7% (headwind), and 4.5% (crosswind) compared with wind-agnostic planning, recovering 87.9%, 85.7%, and 75.0% of CFD-reference savings respectively.
- Fast, corridor-localized 3D wind prediction that is decision-useful (rather than CFD-perfect) enables wind-aware UAV route and speed planning at mission time, cutting energy use without expensive CFD runs.
WHY IT MATTERS
Fast, corridor-localized 3D wind prediction that is decision-useful (rather than CFD-perfect) enables wind-aware UAV route and speed planning at mission time, cutting energy use without expensive CFD runs.