Novel Computational Framework for Optimal Multi-Agent Control
Abstract
A novel computational framework for the optimal control of multi-agent systems is developed. The framework is applicable in a variety of U.S. Navy applications including undersea, surface, and air vehicles. New methods for generating discrete approximations of the multi-agent optimal control problem are developed using adaptive Gaussian quadrature collocation, leading to a largesparse nonlinear programming problem (NLP). This sparse NLP is then solved using a distributed framework approach for nonlinear optimization. The discrete approximation is developed such that the NLP decreases in size as the horizon shrinks, thereby making it possible to solve the NLP in real time. The methods developed in this research are tested using a team of aerial and ground robots. The framework developed in this research is ideally suited to systems where information is acquired and needs to be used asynchronously. In addition, the approach developed in this research assumes little to no communication between the vehicles, thereby making it applicable in systems where the information available to any agent in the system is limited.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 20, 2019
- Source ID
- N000141912543
Entities
People
- Anil V. Rao
Organizations
- Office of Naval Research
- United States Navy
- University of Florida