URED Observer-Based Feedback Linearized Neuro Adaptive SMC for a Twin Rotor MIMO System: Design and Experimental Study

Unmanned Aerial Vehicles (UAVs) are highly nonlinear and sophisticated systems that demand precise trajectory tracking in environments with uncertainties and disturbances. This research presents advanced nonlinear, adaptive, and artificial intelligence-based control strategies for UAVs. Beyond simulation, the strategies are experimentally evaluated on a coupled Two Degree of Freedom (2-DOF) Twin-rotor MIMO System (TRMS). The proposed strategies include Sliding Mode Control (SMC), Super Twisting (ST), BackStepping (BS), and Neuro-Adaptive SMC (NNSMC), all designed using a feedback linearized mathematical model of the system. System performance is enhanced by decoupling the TRMS into horizontal and vertical subsystems through Lie derivatives and diffeomorphism principles. A Uniform Robust Exact Differentiator (URED) estimates rotor speeds and recovers missing derivatives, while a nonlinear state feedback observer improves system observability and mitigates uncertainties and external wind gusts. Furthermore, ST and NNSMC-based laws reduce high-frequency oscillations in the control input of the first-order SMC law, resulting in improved transient response. The experimental results reveal that NNSMC significantly outperforms ST and BS in terms of trajectory tracking accuracy, transient performance, and integral performance indices for both pitch and yaw angles. These findings underscore the superior convergence performance and robustness of NNSMC, establishing it as a promising solution for precise TRMS control in real real-world environment.