Date of Award

Fall 12-14-2023

Access Type

Thesis - Open Access

Degree Name

Master of Aerospace Engineering


Aerospace Engineering

Committee Chair

Dr. Kadriye Merve Dogan

First Committee Member

Dr. Richard J. Prazenica

Second Committee Member

Dr. Hever Y. Moncayo

College Dean

James Gregory


This work addresses the joint tracking problem of robotic manipulators with uncertain dynamical parameters and actuator deficiencies, in the form of an uncertain control effectiveness matrix, through adaptive control design, simulation, and experimentation. Specifically, two novel adaptive controller formulations are implemented and tested via simulation and experimentation. The proposed adaptive control formulations are designed to compensate for uncertainties in the dynamical system parameters as well as uncertainties in the control effectiveness matrix that pre-multiplies the control input. The uncertainty compensation of the dynamical parameters is achieved via the use of the desired model compensation–based adaptation, while the uncertainties related to the control effectiveness matrix are dealt with via two fundamentally different novel adaptation methods, namely with bound-based and projection operator-based methods. The stability of the system states and convergence of the error terms to the origin are proven via Lyapunov–based arguments. Extensive numerical studies are performed on a two–link planar robotic device, and experimental studies are preformed on Quansers QArm to illustrate the effectiveness of both adaptive controllers. In the experimental validation of the theory, both adaptive controllers demonstrate remarkable resilience, maintaining control of the Quanser QArm even with up to an 80% control input deficiency. After tuning the gains, both joints satisfactorily tracked the desired trajectories. When evaluating the entire experiment, the norm of the square of the total error is averaged. The bound-based controller exhibited an average error of 2.816◦ across all cases, while the projection operator-based controller had a reduced average error of 1.012◦ across all cases. Furthermore, over time, there is a noticeable decrease in error for both joints. These results underscore the robustness and effectiveness of the proposed adaptive controllers, even under substantial actuator deficiencies. The results highlight the significance of achieving near-perfect system knowledge and the careful selection of controls for desirable system performance