Neural network based position control of an underactuated flapping wing aircraft considering the aerodynamic damping

The aerodynamic damping induced by the body motion is significant in flapping wing aircraft due to the low Reynolds number flight and the flapping motion. It can influence the flight stability and is coupled with active control. This study proposes a position control strategy for a self-designed underactuated flapping wing aircraft by considering the aerodynamic damping. Inspired by the pendulum-like dynamics, the dominant aerodynamic damping is theoretically modelled under linear damping assumption and the damping parameters are identified by two specially designed experimental platforms. It is experimentally found that the flapping motion does increase the damping compared with the non-flapping case. The damping influence is thus non-negligible in flapping wing aircraft. The outer position controller and the inner attitude controller are designed by including the modelled linear damping terms, and the unmodelled nonlinear little damping terms and other nonlinearities are estimated via the radial based function neural network. The kinematic solve is used to deal with the under-actuated characteristics which can solve the expected roll and pitch angles based on the position controller. The algorithm is compared with traditional methods under two different cases. It is proved that by using human prior knowledge of the aerodynamic damping model to compensate the damping influence, the desired angles computed by the outer position controller are more targeted. The attitude tracking control accuracy is also improved. These two points contribute to the speed and accuracy of the position tracking control. This research proves the necessity of including the aerodynamic damping in the active control of flapping wing aircraft as well.