PostDoc Project
Digital twins for cyber-physical production systems: from data to cognition
2024/02/01 - 2025/01/05
AI4C2PS focuses on setting continuous interoperability to collaborative enterprises systems composed of Cyber-Physical Systems (CPS) and Humans, working together for a common objective. Exploiting the Digital Twin (DT) concept in a decentralised way where each CPS and Human is twinned, AI4C2PS seeks to bridge Knowledge Representation and Reasoning with Deep Reinforcement Learning (DRL) techniques to maintain semantic interoperability within the resulting system-of-systems. This is obtained by building an emulation environment where DTs embed semantic reasoning, DRL and explainable AI capabilities to learn from the interactions between entities. DTs continuously enrich their knowledge base from simulations and from actual interoperations at runtime, with semantically enhanced axioms extracted from the learning process. The goal is to form cognitive CPS from the couples , able to interoperate with other CPS and humans in a seamless way whatever happens during their interactions.
The post-doctorate will WP1 addresses the design and assessment of one (or several) model(s) of humans in their roles in the CP entreprise to be integrated in a DT that will represent humans the same way CPS are twinned, and which we call Human Digital Twin (HDT). We will focus mainly on two roles: (1) human-as-a-worker (workers on the field interacting with machines and CPS, supervisor monitoring a global CPS...), and (2) human-as-a-sensor (data provider). Building on state-of-the-art models and studies, the human model will consider different dimensions like, competencies and skills, interests, and preferences, physiological or cognitive characteristics, as well as behavioural patterns, linked to situations identified from the use-cases. Using Human as data provider will also come with some issues that need to be addressed specially the data heterogeneity, the quality/accuracy of information, the confidence. Any aspect that can have an influence on Human-CPS collaboration will be considered, focusing on the pilot application cases, while keeping the overall design at a generic level. In this context, safety elements will be considered as well. The goal of the modelling is to build an HDT that can simulate humans in their interactions with the CPS elements, so that they can take part in the simulations conducted by the DTs and be used to facilitate cognitive interoperability and Human-CPS collaboration.

More specifically, the post-doc will be responsible for :
- State-of-the-art studies on holonic architectures and cognitive ADMMs, as well as software development tools and technical environments.
- Formalization of usage scenarios based on AIPL S.MART equipment (cobot UR5 and YUMI and AMR OMRON).
- Integration of developments carried out by LIST engineers and the ORISON platform for proof-of-concept on an AIPL S.MART use case.
Digital TWIN, Cognitive SMA, Humain in the loop, Holonic Architecture
Modeling and Control of Industrial Systems