Saved in:
Bibliographic Details
Main Authors: Lam, Victor Truong Thinh, Lazar, Mircea
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.03581
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915183159410688
author Lam, Victor Truong Thinh
Lazar, Mircea
author_facet Lam, Victor Truong Thinh
Lazar, Mircea
contents This paper presents a new fast active-set quadratic programming (QP) solver based on inverse matrix updates, which is suitable for real-time model predictive control (MPC). This QP solver, called imuQP (inverse matrix update QP), is based on Karush-Kuhn-Tucker (KKT) conditions and is inspired by Hildreth's QP solver. An extensive convergence and optimality analysis of imuQP, including infeasibility detection, is presented. The memory footprint and computational complexity of imuQP are analyzed and compared with qpOASES, a well-known active-set QP solver in literature. Speed and accuracy of imuQP are compared with state-of-the-art active-set QP solvers by simulating a chain of spring-connected masses in MATLAB. For illustration, MPC with integral action is used, to remove offset when tracking a constant reference, with a small sampling period of Ts = 4 ms. Simulation results show that imuQP is suitable for fast systems - with small Ts - and is competitive with state-of-the-art active-set QP solvers.
format Preprint
id arxiv_https___arxiv_org_abs_2503_03581
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle imuQP: An Inverse-Matrix-Updates-Based Fast QP Solver Suitable for Real-Time MPC
Lam, Victor Truong Thinh
Lazar, Mircea
Optimization and Control
This paper presents a new fast active-set quadratic programming (QP) solver based on inverse matrix updates, which is suitable for real-time model predictive control (MPC). This QP solver, called imuQP (inverse matrix update QP), is based on Karush-Kuhn-Tucker (KKT) conditions and is inspired by Hildreth's QP solver. An extensive convergence and optimality analysis of imuQP, including infeasibility detection, is presented. The memory footprint and computational complexity of imuQP are analyzed and compared with qpOASES, a well-known active-set QP solver in literature. Speed and accuracy of imuQP are compared with state-of-the-art active-set QP solvers by simulating a chain of spring-connected masses in MATLAB. For illustration, MPC with integral action is used, to remove offset when tracking a constant reference, with a small sampling period of Ts = 4 ms. Simulation results show that imuQP is suitable for fast systems - with small Ts - and is competitive with state-of-the-art active-set QP solvers.
title imuQP: An Inverse-Matrix-Updates-Based Fast QP Solver Suitable for Real-Time MPC
topic Optimization and Control
url https://arxiv.org/abs/2503.03581