Automating Timed Automata Design

Obiettivo

Cyber-physical systems (CPSs) are everywhere, from autonomous vehicles to medical devices to smart buildings. Designing such CPSs that achieve complex tasks is a tedious and error prone process. During the design, high-level specifications describing desired functionalities, safety measures, physical properties and restrictions, and optimality criteria have to be considered. Furthermore, as people’s life may depend on CPSs, their correctness is of critical importance. This IF project aims at developing new theory and novel academic tools to make possible the design of such systems in an automated way from a set of given descriptions and specifications with optimality and correctness guarantees.

In the era of CPSs, automating the design process is crucial. This project will investigate the design aspect focusing on subsets of Timed Automata (TA) as a modeling formalism and Metric Linear Temporal Logic (MITL) as a high-level specification language. The main objectives are (1) to generate the “optimal” timed automata from a parametric partial model by both tuning its parameters and synthesizing controllers at the same time, and (2) to develop a new framework for the automated construction of a partial TA from system descriptions and specifications. The optimization of this partial TA will be achieved by (1).

The first objective involves combining two problems, namely optimal control and parameter synthesis for TA, that are currently studied separately in literature. Despite earlier un-decidability results, the parameter synthesis problem has been popular due to the expressivity of TA as a modelling formalism and difficulties in designing TA from complex specifications. This project aims at developing a new framework to generate optimal timed automata by exploring all design directions. This framework will be tested on a practical example that will be generated with an SME via the host organisation.