Advancing Underwater Robots in Complex Environments

Complex underwater environments, such as underwater caves, flooded mines, offshore wind farms, and under-ice waters have significant societal and scientific impacts. For example, surveying underwater cities and exploring underwater caves provide evidence about our history, and the health of underwater infrastructure is extremely important for reliable energy and the blue economy.

While these complex environments are challenging and risky for human divers to perform repeated surveys, underwater robots must be considered as viable candidates. This award supports fundamental research in robot design, control, and perception to overcome the challenges faced in complex underwater environments. A highly maneuverable underwater robot will be designed, fabricated, and tested for moving in tight complex spaces as proof of concept.

The robot will be made accessible and customizable with the aim to grow and broaden the underwater robotics research community. Additionally, an outreach effort to non-STEM students from the School of Design will be conducted for convergent education.

The overarching goal of the project is to increase the functionality and reliability of underwater robots in complex environments. To this end, the project will make contributions to the fundamental research in robotics in three aspects. First, a novel hydrobatic (a term derived from hydro and acrobatic) robot capable of performing challenging maneuvers, such as turn-in-place, hovering, and vertical descent, will be designed.

Second, a novel adaptive neural-network learning control scheme will be developed to enable the robot to efficiently track a desired trajectory precisely under dynamic uncertainty and time-varying environmental disturbances. Meanwhile, it will accurately identify/learn the nonlinear uncertain robot dynamics through online closed-loop control with provable guarantees in both control and learning performances.

Finally, a learning-based terrain reconstruction algorithm will be created for wide aperture sonars. This new algorithm would overcome the elevation ambiguity problem which causes errors when reconstructing underwater environments from a single sonar image. The improved terrain reconstruction result is expected to advance other fundamental aspects in robot localization, navigation, and path planning.

This project is supported by the cross-directorate Foundational Research in Robotics program, jointly managed and funded by the Directorates for Engineering (ENG) and Computer and Information Science and Engineering (CISE).

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.