Adaptive Neural Network Position Control for Quadrotor with Attitude Unavailable

In this paper, the problem of position control for the quadrotor is solved when attitude information is not available. Firstly, the quadrotor model is successfully transformed into a third-order state space formulation by introducing a first-order low-pass filter. Secondly, when attitude information is unavailable, they can be treated as unknown time-varying parameters. A method for time-varying parameter extraction is employed to separate these parameters from the position model of the quadrotor, resulting in an unknown boundary function based on state variables. Subsequently, a neural network (NN) is used to approach the unknown function, and the weight of the NN is adjusted by adaptive technology, so that the designed controller can ensure the uniformly ultimately bounded performance of the closed-loop system. Finally, the Lyapunov Second Method is utilized for stability analysis, and the effectiveness of the proposed approach is validated through simulations.