Online Takagi-Sugeno Fuzzy Identification of a Quadcopter Using Experimental Input-Output Data

This paper presents a sequential learning machine based on the Takagi-Sugeno (TS) fuzzy inference system to model the dynamics of a MIMO nonlinear quadcopter using experimental data. Unlike conventional TS-fuzzy systems, all the antecedent and consequent parameters of our proposed TS-fuzzy model are updated using the gradient descent-based back-propagation algorithm. After extensive numerical simulations, the accuracy of the proposed model is validated and compared with the Fuzzy C-Means clustering (FCM) algorithm and also with the ARMAX linear model identification technique. This paper leverages the advantages of model-free systems, which can incorporate various uncertainties such as noise, wind gusts, etc. The learning capability using back-propagation method is also suitable to represent the nonlinear dynamics of our quadcopter.