A Novel Computational Framework for Real-Time Optimal Control

Abstract

A novel computational framework is developed for the real-time solution of constrained-nonlinear optimal control problems with appli,cations to guidance and control. The approach-developed employs novel methods for discretizing continuous optimal control problems t,ogether-with novel methods for solving large sparse nonlinear optimization problems (NLP). The-framework is designed to have good lo,cal convergence properties so that it can be used in a-model predictive control framework where the control is constantly recomputed, during a series-of guidance cycles to take into account new data, and it is expected to have good global-convergence properties so, as to quickly compute the starting control in cases where the initial-guess may be poor. In order to solve the NLP as efficiently a,s possible, novel methods for-efficient generation of derivatives are developed. These methods will include the development of-novel, approaches for algorithmic differentiation that provide highly accurate derivatives while-simultaneously generating these derivativ,es in a computationally efficient manner. The methods-developed in this research will be tested on a variety of challenging benchmar,k optimal control-problems both from the open literature and new problems that may be of interest as the research-progresses.

Document Details

Document Type
DoD Grant Award
Publication Date
May 16, 2022
Source ID
N000142212397

Entities

People

  • Anil V. Rao

Organizations

  • Office of Naval Research
  • United States Navy
  • University of Florida

Tags

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Operations Research