Stability analysis of complementarity systems with neural network controllers

DOI10.1145/3447928.3456651MaRDI QIDQ6120660zbMATH OpenOpenAlexWikidataFDO

Authors Mahyar Fazlyab, Manfred Morari, Michael Posa

Publication date 21 February 2024

Published in Proceedings of the 24th International Conference on Hybrid Systems: Computation and Control (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2011.07626

zbMATH Keywords

contact dynamicscomplementarity systemsstability verification of neural network controllers

Mathematics Subject Classification ID

Formal languages and automata (68Q45) Specification and verification (program logics, model checking, etc.) (68Q60) Control/observation systems governed by functional relations other than differential equations (such as hybrid and switching systems) (93C30)

Abstract: Complementarity problems, a class of mathematical optimization problems with orthogonality constraints, are widely used in many robotics tasks, such as locomotion and manipulation, due to their ability to model non-smooth phenomena (e.g., contact dynamics). In this paper, we propose a method to analyze the stability of complementarity systems with neural network controllers. First, we introduce a method to represent neural networks with rectified linear unit (ReLU) activations as the solution to a linear complementarity problem. Then, we show that systems with ReLU network controllers have an equivalent linear complementarity system (LCS) description. Using the LCS representation, we turn the stability verification problem into a linear matrix inequality (LMI) feasibility problem. We demonstrate the approach on several examples, including multi-contact problems and friction models with non-unique solutions.

Cites work

• scientific article; zbMATH DE number 1036629 (Why is no real title available?)
• A receding-horizon regulator for nonlinear systems and a neural approximation
• Analysis of Lur'e dominant systems in the frequency domain
• Cone-Copositive Piecewise Quadratic Lyapunov Functions for Conewise Linear Systems
• Difference Methods for Differential Inclusions: A Survey
• Equivalence of hybrid dynamical models
• Introduction to Piecewise Differentiable Equations
• Introduction to the theory of differential inclusions
• Linear Complementarity Systems
• Linear Matrix Inequalities in System and Control Theory
• Lyapunov Stability of Complementarity and Extended Systems
• Nonlinear systems.
• Nonsmooth mechanics. Models, dynamics and control.
• Predictive control for linear and hybrid systems
• Safety Verification and Robustness Analysis of Neural Networks via Quadratic Constraints and Semidefinite Programming
• Sampling algebraic varieties for sum of squares programs
• Sherlock - A tool for verification of neural network feedback systems
• Some perspectives on the analysis and control of complementarity systems
• Stability Analysis Using Quadratic Constraints for Systems With Neural Network Controllers
• Stability Analysis and Control of Rigid-Body Systems With Impacts and Friction
• Stability and Instability Matrices for Linear Evolution Variational Inequalities
• The Linear Complementarity Problem
• The explicit linear quadratic regulator for constrained systems
• Verification of piecewise deep neural networks: a star set approach with zonotope pre-filter
• Warm Start of Mixed-Integer Programs for Model Predictive Control of Hybrid Systems
• Well-posedness and output regulation for implicit time-varying evolution variational inequalities

Cited in

(1)

• Closed-loop stability analysis of deep reinforcement learning controlled systems with experimental validation

This page was built for publication: Stability analysis of complementarity systems with neural network controllers

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6120660)