Self-Organizing Adaptive Robust Fuzzy Neural Attitude Tracking Control of a Quadrotor

In this paper, a novel self-organizing adaptive robust fuzzy neural control (SARFNC) scheme is proposed for the attitude tracking of a quadrotor. Compared with traditional approaches, the distinct features of the SARFNC scheme are as follows: 1) The self-constructing fuzzy neural network (SCFNN) can online self-organize fuzzy neural structure by generating and pruning fuzzy rules automatically, and thereby resulting in accurate approximation. 2) By virtue of improved projection-based adaptive laws, the parameter drift and singularity in membership functions can be avoided simultaneously. 3) A robust supervisory controller is designed to compensate approximation error and enhance the robustness of overall SARFNC control scheme. 4) The SARFNC scheme can not only achieve remarkable tracking performance but also ensure tracking error and their first derivatives are globally uniformly ultimately bounded (GUUB). Simulation studies are performed to demonstrate the remarkable performance of the SARFNC scheme in terms of tracking error and online approximation.