Hybrid dynamical systems: modeling and design for robust decision making, control, and optimization
Abstract
The proposed work will advance the state of the art regarding the use of hy-brid dynamical systems to design feedback algorithms for decision-making tasks, high-performance control, and e cient optimization. Here, by “hybrid”, we mean mechanisms and systems that combine incremental change, including that experi-enced by the physical states, with more extreme change, including that experienced by easy-to-modify digital states. One thrust of our work will be on the use of reset-ting mechanisms to improve performance and robustness of control and optimization algorithms. This thrust will cover the multi-agent setting, developing problem formu-lations and algorithms that require a hybrid modeling framework. Another thrust will be on the combination of resetting mechanisms with stochasticity to produce e cient algorithms for non-convex optimization and also for decision making in dis-tributed multi-agent systems. Also consistent with a hybrid systems framework, we will develop new results for multi-agent systems that employ hybrid feedback and that must cope with a persistently switching task, asset, or environment. In this setting, we will give a rigorous characterization of the “steady-state” behavior of the multi-agent system. More broadly, we aim to bring novelty and uniformity to the design and analysis of hybrid mechanisms in single- and multi-agent systems and to advocate for creative hybrid mechanisms that may be able to improve performance of these systems.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 21, 2022
- Source ID
- FA95502110452XX0
Entities
People
- Andrew R. Teel
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- University of California, Santa Barbara