Intelligent vision control system for underwater robots based on coupled optical models

Underwater robots are widely used in engineering domains such as resource exploration and environmental monitoring. However, during underwater operations, severe absorption and scattering in water reduce visibility for optical imaging. Traditional approaches often relies on experience-based lighting schemes, which suffer from hard-to-quantify lighting parameters, low energy efficiency, and unstable image quality, resulting in limiting adaptability across different water qualities and operating environments. This paper establishes a coupled optical model integrating the Beer - Lambert law with a camera optical response model and proposes a method for light-source parameter inversion and illumination configuration optimization. It also presents a Python-based software system for light-source parameter design and illuminance visualization. The method rapidly generates lighting schemes under varying operating conditions. Compared with empirical lighting approaches, it features clear optical modeling and a implementable system architecture, and can provide useful references for the quantitative design and optimization of underwater-robot illumination systems.