This research presents an application of support vector machine for the control of a fixed-wing unmanned aerial vehicle. Neural networks have been broadly used in developing this type of application. Unlike neural networks, support vector machine is mathematically proven to generate global solutions. Support vector machine regression models are developed using both off-line and on-line learning. The data required for the training was obtained using the flight-validated dynamics model of the airplane. The trained models are implemented in open- and closed-loop systems. Simulation results are shown.
展开▼
