Didier Lime takes up the position of Professor in Real Time Systems

Didier Lime is a lecturer/researcher in the Laboratory of Digital Sciences Laboratory (LS2N) in the Real Time Systems team, which he headed up until 1 January 2022, and is now deputy head. He is also programme supervisor of the Embedded Real Time Systems (CORO ERTS) Master's programme and teaches in the Compuer Science specialisation of the engineering programme.

New courses in computer science and artificial intelligence to meet company needs

Through this professorship, Didier Lime intends to set up new courses to meet company needs in the field of computer science (there is already strong student demand for the current Computer Science specialisation).

An engineering programme specialisation in computer science for artificial intelligence is set to open in the 2022/23 academic year. It will focus on artificial intelligence in the broadest sense, including statistical learning but also important topics such as planning, game theory and logic. Other more conventional topics will be covered to ensure efficient implementation of artificial intelligence algorithms: object programming, parallel programming, advanced algorithmics and data structures, etc.
The aim is for students to have a clear understanding of how algorithms work and their potential problems.

In the longer term, Didier Lime would also like to develop a new Erasmus Mundus Joint Master Degree based on the Embedded Real Time Systems programme.

Improving models to check real-time systems

As a specialist in the verification of complex systems (in particular real-time embedded systems), Didier Lime conducts research into the modelling of systems whose behaviour is observed through well-identified events. These systems can be biological systems (e.g., transitions between gene expression modes), production systems (e.g., assembly lines) or even computer systems.

Together with his LS2N team, Didier Lime will develop a research project on the creation of models along three main lines:

• Introduction of parameters, whose value will be determined automatically, to model quantities such as durations or numbers of processes as early as possible in the development cycle
• Taking energy into account in the systems to optimise energy expenditure
• Correct-by-construction control synthesis, from a specification of their properties and their environment.

The originality of this research approach lies in the combination of these three aspects and the use of these models for embedded systems that use artificial intelligence.

A longer term objective is to use statistical methods, which generally scale up better, to verify that the systems will work well. The techniques used in this approach are similar to those used in the field of reinforcement learning.

This research work will strengthen the partnerships already established within the LS2N teams and Nantes University, but also with other national partners such as IRISA in Rennes, LMF in Paris-Saclay and LORIA in Nancy.