HDR Defense of Thierry MOYAUX

Date
Thursday 08 January 2026
Type
Defense

M. Thierry MOYAUX soutient son HDR le Jeudi 8 janvier 2026 à 14h, intitulée : "Quand, où et comment (dé)centraliser la prise de décisions pour une
Industrie 4.0 durable". La soutenance a lieu sur le campus Doua.

Le jury sera composé de :

  • Mme ALPAN Gulgün, Grenoble INP - UGA, Présidente du jury
  • Mme GOEPP Virginie, INSA Strasbourg, Rapporteure
  • M. FRAYRET Jean-Marc, Ecole Polytechnique de Montréal, Rapporteur
  • M. MATHIEU Philippe, Université de Lille, Rapporteur,
  • M. TRENTESAUX Damien, Université Polytechnique Hauts-de-France, Membre
  • Mme COSTA-AFONSO Roberta, ISAE supméca Membre,
  • M. MORGE Maxime, Université Claude Bernard Lyon 1, Membre
  • M. CHEUTET Vincent, INSA Lyon, Membre

 

Abstract:

(De)centralising the organisation of decision-making to a greater or lesser level should be based on quantified evidence. I am particularly interested in the (de)centralisation of decision-making organisation (rather than its physical or informational structuring) in production systems of goods and services, $e.g.,$ how many decision-makers should be there and how should they interact?
This is an old question that I address with recent digital tools.
From a technical point of view, this organisational issue corresponds to the choice of decision support technologies (de)centralised to a more or lesser degree.
Indeed, Industrial Engineering employs various such decision-making methods, initially centralised and often based on Mixed Integer Linear Programming (MILP, or PLNE in French) and, more recently, decentralised through MultiAgent Systems (MASs, or SMAs in French). Each of both approaches is developed independently by its community with little overlap. These approaches seem complementary to me. For instance, MILP finds the optimal decision if it has sufficient computation time available, whereas MASs are not only more reactive but also take the autonomy of decision-makers into account.
In this context, the core of my research aims to optimise the organisation of decision-making by studying when, where and how to use MILP within a MAS.
To this end, I quantitatively compare, using simulation, the efficiency of more or less (de)centralised decision-making organisations when agents are more or less selfish, hence my inspiration from game theory.

This basic research contributes in multiple ways to support the making of decisions in Industry 4.0.
A first example is my ongoing work on applying agent-based simulation as an engine for a digital twin.
Next, I am starting to work on the transport of people and goods in sparsely populated areas, taking my inspiration from the physical internet (interconnection of logistics networks).
In the longer term, I would like to study how to improve the sustainability of production systems in our increasingly VUCA (Volatility, Uncertainty, Complexity \& Ambiguity) environment thanks to the lack of a single point of failure in a MAS.