Real-time shape estimation for a flexible unmanned air vehicle via kalman filtering

As modern aircraft designs with flexible airframes become susceptible to dynamic coupling between rigid body and structural modes, challenges in aircraft design, modeling, and control increase significantly. Active modal suppression control is required to ensure dynamic stability across desired flight envelope. Closed-loop shape control, which takes advantage of the airframe flexibility to optimize aerodynamic shape for minimal drag, is also an important technology to be developed. One of the critical pieces of technology required for structural mode and shape control is shape estimation of the structure in real-time, which can serve as a feedback signal. In this paper, a Kalman filter-based shape estimation approach for a small flying wing unmanned air vehicle (UAV) is described. The UAV features a set of distributed sensors including small, light-weight inertial measurement units (IMUs) along its wings and center-body, as well as cameras that record and process visual information on wing-tip deflections. A linear Kalman Filter is designed using a linear aeroelastic vehicle-dynamic model for state propagation and IMU measurement data for measurement updates. The filter estimates wing-tip deflection and twist while camera data, which is available at a different sampling rate and is independent of the theoretical model, is used for validating the estimation. Data obtained from flight tests conducted for system identification purposes are used to validate the performance of the filter. Finally, blended estimates of wing-tip heave and twist are obtained via weighted averaging of filter estimates and visual data from the cameras.