Description du projet
Synthèse formelle des logiciels de commande intégrés
Les logiciels de commande intégrés jouent un rôle central dans un éventail d’applications complexes, dont les transports autonomes, les réseaux électriques et l’aérospatiale. Ces applications représentent des exemples de systèmes cyber-physiques (SCP), qui se font de plus en plus omniprésents en raison des avancées rapides au niveau du calcul, de la communication et de la mémoire. Le projet AutoCPS, financé par l’UE, développera un processus de conception innovant visant à synthétiser les logiciels de commande systématiquement intégrés, afin de réduire les coûts de conception tout en garantissant leur exactitude. Pour remédier à la complexité de conception sous-jacente des SCP à grande échelle, ce projet préconise l’adoption d’une stratégie du «diviser pour mieux régner» pour la synthèse des logiciels de commande. Cette méthode repose sur la combinaison de techniques de composition issues de l’informatique et de la théorie de la commande.
Objectif
Embedded Control software plays a critical role in many safety-critical applications. For instance, modern vehicles use interacting software and hardware components to control steering and braking. Control software forms the main core of autonomous transportation, power networks, and aerospace. These applications are examples of cyber-physical systems (CPS), where distributed software systems interact tightly with spatially distributed physical systems with complex dynamics. CPS are becoming ubiquitous due to rapid advances in computation, communication, and memory. However, the development of core control software running in these systems is still ad hoc and error-prone and much of the engineering costs today go into ensuring that control software works correctly.
In order to reduce the design costs and guaranteeing its correctness, I aim to develop an innovative design process, in which the embedded control software is synthesized from high-level correctness requirements in a push-button and formal manner. Requirements for modern CPS applications go beyond conventional properties in control theory (e.g. stability) and in computer science (e.g. protocol design). Here, I propose a compositional methodology for automated synthesis of control software by combining compositional techniques from computer science (e.g. assume-guarantee rules) with those from control theory (e.g. small-gain theorems). I will leverage decomposition and abstraction as two key tools to tackle the design complexity, by either breaking the design object into semi-independent parts or by aggregating components and eliminating unnecessary details. My project is high-risk because it requires a fundamental re-thinking of design techniques till now studied in separate disciplines. It is high-gain because a successful method for automated synthesis of control software will make it finally possible to develop complex yet reliable CPS applications while considerably reducing the engineering cost.
Champ scientifique
Programme(s)
Thème(s)
Régime de financement
ERC-STG - Starting GrantInstitution d’accueil
80539 MUNCHEN
Allemagne