Vision-Based Underwater Inspection With Portable Autonomous Underwater Vehicle: Development, Control, and Evaluation.

This paper presents the development of a portable autonomous underwater vehicle (AUV) named Shark, along with the design of its hovering control system for vision-based underwater inspection. Firstly, the mechanical structure design, electrical system configuration, and software system development of the Shark AUV are introduced, highlighting its cost-effectiveness, compact size, and streamlined torpedo shape to enhance the flexibility of underwater inspections. Secondly, the dynamic model of the Shark AUV is established, and the hydrodynamic coefficients are numerical estimated utilizing computational fluid dynamics (CFD) simulations. Thirdly, considering that the developed Shark AUV is underactuated in the sway degree of freedom, a novel heading control model is developed that reasonably tackles the motion coupling between sway and yaw. This model facilitates the design of a robust $\mathcal {H}_{\infty }$ controller to regulate the heading angle of the Shark AUV for vision-based underwater inspections. Finally, the effectiveness of the designed hovering control method and vision-based underwater inspection system are validated in both a 3D simulation platform and a laboratory test pool. The experimental results demonstrate that the Shark AUV prototype can achieve satisfactory vision-based underwater inspection performance within a pixel error of 5 × 5, and the proposed $\mathcal {H}_{\infty }$ heading controller consistently provides superior performance in handling motion coupling, parametric uncertainty, and external disturbances.