No-Flow-Sensor Wind Estimation for Multirotor UAS Using Kalman Filtering
Keywords
Wind estimation, Kalman Filter, No-flow sensor, Uncrewed Aircraft Systems
Presenter Abstract
This work describes the design and testing of a wind estimator for multirotor uncrewed aircraft systems based on Kalman filtering and standard onboard navigation measurements. The method estimates wind from GNSS-derived velocity, vehicle attitude, and body specific force, without relying on a dedicated flow sensor. The work presents the filter equations and the reasoning behind the estimator structure.
A main result of the work is the establishment of a 2D Kalman filter baseline for horizontal wind estimation. That baseline was evaluated through maneuver-based simulation tests and sensor-noise studies, and provides a practical reference solution for the horizontal wind components. The work summarizes the RMSE behavior and shows how noise testing was used to define the baseline performance of the filter.
Using the same framework, the work also examines the extension to a 3D Extended Kalman Filter. The 3D case is more difficult because the vertical component depends more strongly on force-model assumptions, especially when direct thrust measurements are not available. The work shows why the horizontal estimation problem can be made reliable sooner, and why the full 3D problem remains a more difficult estimation task within the same Kalman filtering framework.
Presentations
Presented in Session 5: Sensors I
No-Flow-Sensor Wind Estimation for Multirotor UAS Using Kalman Filtering
This work describes the design and testing of a wind estimator for multirotor uncrewed aircraft systems based on Kalman filtering and standard onboard navigation measurements. The method estimates wind from GNSS-derived velocity, vehicle attitude, and body specific force, without relying on a dedicated flow sensor. The work presents the filter equations and the reasoning behind the estimator structure.
A main result of the work is the establishment of a 2D Kalman filter baseline for horizontal wind estimation. That baseline was evaluated through maneuver-based simulation tests and sensor-noise studies, and provides a practical reference solution for the horizontal wind components. The work summarizes the RMSE behavior and shows how noise testing was used to define the baseline performance of the filter.
Using the same framework, the work also examines the extension to a 3D Extended Kalman Filter. The 3D case is more difficult because the vertical component depends more strongly on force-model assumptions, especially when direct thrust measurements are not available. The work shows why the horizontal estimation problem can be made reliable sooner, and why the full 3D problem remains a more difficult estimation task within the same Kalman filtering framework.