Backstepping control for quadrotor with BP neural network based thrust model

This paper presents a nonlinear backstepping control scheme for quadrotor with BP neural network based thrust model. The thrust in quadrotor system is difficult to calculate or measure, so the paper build a BP based thrust model to approximate the mapping between thrust with factors of altitude, voltage and the high level time of PWM. Through the model, the control input for each rotor can be calculate accurately to gain the desired thrust. The controller is designed by backstepping method and verified by Lyapunov stability theorem. Hovering experimental results are presented to show the effectiveness of the proposed controller.