RI: Small: Computational Imaging for Underwater Exploration

The underwater environment poses great challenges for vision sensing due to the refraction, absorption, and scattering that occurs in the water column. Although a variety of manned vehicles, remotely operated vehicles (ROVs), and autonomous underwater vehicles (AUVs) have been developed for various underwater missions, the vision sensing capacity of these underwater vehicles still stays at a limited level.

This project aims at developing novel computational imaging solutions to facilitate underwater robotic tasks, such as autonomous navigation and in-detail sea floor mapping. The project is expected to produce new imaging systems, scene reconstruction algorithms, and control theories that are tailored for underwater vehicles. The project benefits marine research, oil & gas industry, and the military by developing underwater vision systems that decrease the costs, challenges, and risks associated with ocean exploration.

The project tightly integrates research with education by introducing new curricula on computer vision and marine robotics. The project will provide training and research experience to both undergraduate and graduate students, in particular women and underrepresented minorities.

This project focuses on three specific research objectives: 1) angular ray sampling for nonlinear light transport analysis; 2) underwater reflectance modeling and geometry estimation; and 3) visual perceptive tracking for underwater vehicles. A critical component of this project is to develop a novel angularly sampled imaging system that strategically emits and collects light rays, which allows the analysis of nonlinear light paths through water.

Specifically, the team will first investigate the underwater light transport model in presence of refraction, scattering, and absorption. Based on the light transport model, the team will develop new surface reflectance models and three-dimensional shape reconstruction algorithms that follow the physics of underwater ray optics. Finally, the team will integrate the imaging system with underwater vehicles and develop a novel visual perceptive tracking navigation system that leverages the angular ray sampling scheme.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.