Developing machine learning control of a non-axisymmetric x-wing flapping drone

Micro sized flapping wing vehicles are inherently difficult to control. To overcome its instability, the method of machine learning is proposed. We focus on a palm-sized, nonaxisymmetric flapping x-wing vehicle that is able to perform level flight with sufficient speed and payload. In the scaled conjugate gradient framework, we investigate the effect of different sets of neurons used. Results are compared with its actual flight over three time scales.