Trajectory Tracking of a Quadcopter Using Fuzzy Logic and Neural Network Controllers

Abstract

In this work, the trajectory tracking control of an Unmanned Aerial Vehicle (UAV) has been realised using fuzzy logic and neural network based controllers. Parrot AR.Drone 2.0 has been selected as the test platform. For simulated and real-time experimental studies, a square shaped reference trajectory has been generated, and the discrepancies from this trajectory in x-and y-directions along with their derivatives have been employed as the input signals to the proposed controllers. The update rules for the neural network have been derived based on the variable structure systems theory to enable stable online tuning of the parameters. The obtained results indicate that both fuzzy logic and neural network controllers can be applied effectively to the trajectory tracking of a drone, and particularly neural networks with variable structure systems theory based learning algorithms exhibit a highly robust behaviour against disturbances.