Tractable Computational Methods for Optimal Control Via Fast Dynamic Programming
Abstract
Outcomes delivered by this project include advances in the representation of general solutions of dynamic programming equations and Hamilton-Jacobi-Bellman (HJB) partial differential equations (PDEs) using idempotent fundamental solution semigroups [P2,8,1012,2224,40], new adaptive idempotent methods for solving nonlinear regulator problems [P10, 11, 13], new game representations for non-quadratic regulator problems [P1, 3, 20, 30, 32, 33, 40], a preliminary theory of stationary control [P4, 8, 15, 21, 23, 27, 31, 36], and a preliminary new approach to curse-of-dimensionality attenuated optimization-based optimal feedback policy synthesis [P9, 1719]. The project has also supported auxiliary advances in extremum seeking feedback control for dynamical systems evolving on manifolds[P16], networked control and observer design [P6, 7, 14, 26, 35], quantum control [P28, 37, 38], and stability theory [P5, 25, 29, 34, 39]. These outcomes have been communicated via 40 scholarly publications in international journals and conference proceedings, and 19 invited presentations at national and international meetings. The project contributed direct funding support for the training of two postdoctoral research fellows in optimal control, via consecutive one year appointments, and partial support for travel associated with participation in conferences and US-based research collaboration (with Professor W.M. McEneaney at UCSD). It facilitated the organization of 4 international minisymposia on the topics of idempotent methods and stationary control [A1, 2, 4, 7], and 3 multi-day meetings on applied dynamics in Australia [A3, 5, 6].
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 12, 2023
- Accession Number
- AD1224956
Entities
People
- Peter M. Dower
Organizations
- University of Melbourne