Temporal Parallelization of the HJB Equation and Continuous-Time Linear Quadratic Control

This article presents a mathematical formulation to perform temporal parallelization of continuous-time optimal control problems, which can be solved via the Hamilton-Jacobi-Bellman (HJB) equation. We divide the time interval of the control problem into subintervals, and define a control problem in...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on automatic control Vol. 70; no. 6; pp. 3755 - 3770
Main Authors: Sarkka, Simo, Garcia-Fernandez, Angel F.
Format: Journal Article
Language:English
Published: New York IEEE 01.06.2025
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Associativity
Central processing units
Continuous-time control
Costs
CPUs
Dynamic programming
graphics processing unit (GPU)
Graphics processing units
Hamilton–Jacobi > Bellman (HJB) equation
linear quadratic tracking
Mathematical models
Microprocessors
multicore
Numerical methods
Operators (mathematics)
Optimal control
Optimization
parallel computing
Reviews
Symmetric matrices
Time complexity
Time optimal control
Trajectory
Viscosity
ISSN:0018-9286, 1558-2523
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This article presents a mathematical formulation to perform temporal parallelization of continuous-time optimal control problems, which can be solved via the Hamilton-Jacobi-Bellman (HJB) equation. We divide the time interval of the control problem into subintervals, and define a control problem in each subinterval, conditioned on the start and end states, leading to conditional value functions for the subintervals. By defining an associative operator as the minimization of the sum of conditional value functions, we obtain the elements and associative operators for a parallel associative scan operation. This allows for solving the optimal control problem on the whole time interval in parallel in logarithmic time complexity in the number of subintervals. We derive the HJB-type of backward and forward equations for the conditional value functions and solve them in closed form for linear quadratic problems. We also discuss numerical methods for computing the conditional value functions. The computational advantages of the proposed parallel methods are demonstrated via simulations run on a multicore central processing unit and a graphics processing unit.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0018-9286
1558-2523
DOI:10.1109/TAC.2024.3518309