Formal verification of hybrid systems

Abstract

In formal verification, a designer first constructs a model, with mathematically precise semantics, of the system under design, and performs extensive analysis with respect to correctness requirements. The appropriate mathematical model for embedded control systems is hybrid systems that combines the traditional state-machine based models for discrete control with classical differential-equations based models for continuously evolving physical activities. In this article, we briefly review selected existing approaches to formal verification of hybrid systems, along with directions for future research. Categories and Subject Descriptors General TermsVerification. KeywordsHybrid systems, Control systems, Model checking, Formal methods REVIEW OF CURRENT APPROACHESModel-based design offers a promising approach for detecting and correcting errors in early stages of system design [33,37,49]. In this methodology, a designer first constructs a model, with mathematically precise semantics, of the system under design, and performs extensive analysis with respect to correctness requirements before generating the implementation from the model. Embedded systems, such as controllers in automotive, medical, and avionic systems, consist of a collection of interacting software modules reacting to a continuously evolving environment. The appropriate Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. EMSOFT'11, October 9-14, 2011, Taipei, Taiwan. Copyright 2011 mathematical model for design of embedded control systems is hybrid systems that combines the traditional models for discrete behavior with classical differential-and algebraicequations based models for dynamical systems. Such models can capture both the controller -the system under design, and the plant -the environment with continuously evolving physical activities in which the system operates. Given that (1) automated verification tools have recently been successful in finding bugs in "real-world" hardware protocols and device drivers [13,15], (2) tools such as Stateflow/Simulink are commonly used in automotive and avionics industry for modeling, and (3) high assurance is a necessity in safetycritical applications that deploy embedded software, formal verification of hybrid systems has been a vibrant research area for the past 20 years. This article is an overview of current research directions, aimed at providing an introductory "roadmap" rather than a comprehensive survey. ModelingIn early 1990s, formal models for discrete reactive systems were integrated with models for dynamical systems [2,41]. The model of hybrid automata [1,2,29] has emerged to be a popular choice. A hybrid automaton is an extended fin...

Kutatási prioritások a megbízható és hasznos mesterséges intelligencia létrehozásáért

“…Az adaptív szabályozás elméletére (Åström-Wittenmark, 2013), az úgynevezett kiberfizikai rendszerekre (Platzer, 2010) vagy a hibrid és robotrendszerek ellenőrzésére (Alur, 2011;Winfield-BlumLiu, 2014) irányuló munka szintén rendkívül releváns, de ugyanezekkel a nehézségekkel kell szembenéznie. És természetesen mindezek a kérdések a sztenderd problémára rakódnak, miszerint be kell bizonyítani, hogy az adott szoftver-kezdemény megfelelően implementálható például a megerősítő tanulás algoritmusának egy kívánt típusára.…”

Safety Verification Utilizing Model-based Development for Safety Critical Cyber-Physical Systems

“…Amongst the varied model checking technologies, there exists an approach for hybrid system model checking which can express both discrete and continuous aspects [3], [4], [5], [6], [7]. However, hybrid system modeling is often not easy for most engineers because they are only conversant with their own design aspects, such as software or mechanical design.…”

Computation of polytopic invariants for polynomial dynamical systems using linear programming

“…in the Poincaré Bendixon theorem [Sas99]); they have played an important role in control theory for the analysis of performance, robustness or practical stability [Bla99]. They are also of great interest for reachability analysis of continuous and hybrid systems, especially for verification of safety properties where the goal is to prove that trajectories of a system starting from a given set of initial states will never reach a specified set of unsafe states [Alu11]. This can be done by exhibiting an invariant set, containing the set of initial states, and whose intersection with the set of unsafe states is empty.…”