Duality Theory for Stochastic Filtering and Control

Abstract

The major theoretical thrust is to develop duality theory for stochastic filtering and control. Specifically, new methods for the analysis of the stochastic nonlinear filter will be developed, by making use of the dual optimal control techniques. A major algorithmic goal is to develop duality-inspired interacting particle algorithms (e.g., the ensemble Kalman filter (EnKF) and the feedback particle filter (FPF)) for solving the stochastic optimal control problem. Given the past successes and impact of EnKF in the data assimilation applications involving large-scale systems, the proposed work can be a game-changer in problems of recent research interest at the intersection of stochastic optimal control and and reinforcement learning. The proposed research will leverage several recent breakthroughs by the PI, specifically, on topics related to duality and the feedback particle filter.

Document Details

Document Type
DoD Grant Award
Publication Date
Feb 29, 2024
Source ID
FA95502310060

Entities

People

  • Prashant G. Mehta

Organizations

  • Air Force Office of Scientific Research
  • United States Air Force
  • University of Illinois Urbana–Champaign

Tags

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Ocean-Atmosphere Mesoscale Modeling, Data Assimilation, and Flux Boundary Layers
  • Operations Research

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms