Sense-Understand-Decide Chains to Support Dynamic Mission Replanning for USVs
Abstract
The new generation of autonomous surface vessels presents many challenges from a mission planning and safety perspective. Their combination of small size and wide range of operating environments means that vessel motions and secondary loads will determine vessel operational performance and safety. However, existing engineering techniques are unlikely to be able to predict these responses withsufficient accuracy at the design stage to ensure mission success. Furthermore, lacking a human crew, the vessels have no way of improving as operational experience is gained. This work will address safety-critical digital twin systems that can quickly learn in changing conditions and perform dynamic mission replanning based on their learning. To do so, three primary technical objectives willbe explored:1. Improved Methods for Short-Term Extreme Response Prediction: By blending existing statistical methods with few/zero-shot machine learning and Bayesian approaches, calculation frameworks for quickly assessing vessel safety in previously unknown conditions will be explored.2. Correlated Response Monitoring: Not all responses, especially structural responses, can be sensed cost-effectively. Approaches using a basket of partially correlated, easily-sensed responses instead of difficult-to-sense responses will be explored, expanding the framework pioneered in the first research objective.3. Dynamic Mission Replanning: Using the learning fromobjectives one and two, dynamic mission replanning under uncertainty with multiple objectives and adjustable risk acceptance will be demonstrated.At the completion of this project, the results of the algorithms and frameworks explored at a 6.1 level will be documented in a series of reports and archival journal papers. Additionally, one Ph.D. student will be trained on this project through the completion of their Ph.D. Approved for Public Release.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Nov 09, 2024
- Source ID
- N000142412571
Entities
People
- Matthew Collette
Organizations
- Board of Regents of the University of Michigan
- Office of Naval Research
- United States Navy