Long-Term Underwater Autonomy for Surveillance and Manipulation
Abstract
The aim of this project is develop new capabilities for long-term autonomy and supervisory control of underwater robots operating in highly complex environments that include: poor or irregular communication, coordinating multiple agents performing surveillance, inspection and maintenance tasks, and interacting with humans in multiple roles (divers underwater and supervisors on the surface). To achieve our objectives, we will develop algorithms for learning manipulation of underwater objects that can adapt to new situations and conditions from experience, self-adaptation, and interactions. Communication between the vehicle and human supervisors is particularly challenging due the low bandwidths available with underwater communication. Additionally, we will develop new methods of aggregating important information to support supervisory control of multiple AUVs by one human. To close the loop with humans, a noveluser interface will be developed with a model of the system that provides human supervisors with relevant information as appropriate for the scenario and predicted behavior of the system. The AI will be built to engender appropriate trust to avoid over-reliance or underutilization. Finally, our work will focus on the development of an integrated software stack, as well as performance metrics and evaluations methods.Approved for public release
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 24, 2023
- Source ID
- N000142312744
Entities
People
- Holly Ann Yanco
Organizations
- Office of Naval Research
- United States Navy
- University of Massachusetts