Image-based visual servoing (IBVS) control is increasingly popular in autonomous underwater vehicles (AUVs) due to its reliance solely on camera systems, making it suitable for the complex and dynamic underwater environments where traditional navigation systems may falter. However, conventional IBVS typically lacks robust constraint handling capabilities, limiting its application scope.

To address these challenges, this thesis introduces a model predictive control (MPC) strategy designed to optimize the control performance of AUVs using IBVS. MPC explicitly handles system constraints but traditionally involves solving complex optimization problems at each time step, which incurs significant computational overhead. To solve this problem, we propose two novel nonlinear model predictive control (NMPC) formulations for IBVS in AUVs that significantly reduce computational time compared to conventional NMPC approaches, enhancing real-time applicability while ensuring compliance with dynamic constraints.

Furthermore, the inherent nature of underwater IBVS systems, reliant on camera imagery, complicates frequent recalibration, particularly in inaccessible environments where information such as depth may be unavailable. To address this, we develop an uncalibrated IBVS framework, thereby adapting to changing underwater conditions without the need for constant recalibration.

Additionally, to handle disturbances and model uncertainties that are introduced by the uncalibrated IBVS settings, a robust MPC strategy is developed. The robust MPC IBVS controller computes the optimal control sequence under the worst case scenario, significantly enhancing the robustness and reliability of AUV operations.

Overall, the enhancements and methodologies developed in this thesis provide a comprehensive solution to the practical challenges of implementing effective and efficient visual servoing in AUVs, potentially transforming autonomous underwater navigation and task execution.