Job offer

European Commission; France; Sur place; EUR 24 000 - 35 000; Plein temps

Résumé du poste

Un institut de recherche de premier plan en France propose une thèse visant à développer des conceptions de supports de câblage pour l'intégration électrique d'avions. Ce projet innovant nécessite une bonne maîtrise de l'IA et de la modélisation 3D, avec la possibilité d'avoir un impact direct sur l'industrie aéronautique. Les candidats doivent avoir un master en ingénierie ou dans des domaines connexes, avec un fort intérêt pour l'automatisation et l'optimisation.

Qualifications

• Master requis dans un domaine pertinent.
• Intérêt marqué pour l'IA, la modélisation 3D et l'automatisation.
• Expérience en optimisation ou modélisation 3D est un plus.

Responsabilités

• Développer une méthode d'IA pour la conception de supports de câblage.
• Intégrer des contraintes de fabrication et de conception dans un modèle génératif.
• Valider les conceptions générées selon des normes d'installation d'avions.

Connaissances

Intérêt pour l'IA

Modélisation 3D

Conception automatisée

Formation

Master en Aéronautique, Mécanique, Informatique

Description du poste

Organisation/Company Arts et Métiers ParisTech Aix en Provence Research Field Engineering Computer science » Informatics Computer science » Digital systems Researcher Profile Recognised Researcher (R2) Leading Researcher (R4) First Stage Researcher (R1) Established Researcher (R3) Country France Application Deadline 11 Jul 2025 - 22:00 (UTC) Type of Contract Temporary Job Status Full-time Offer Starting Date 3 Nov 2025 Is the job funded through the EU Research Framework Programme? Not funded by a EU programme Is the Job related to staff position within a Research Infrastructure? No

Offer Description

Context: Aircraft electrical harness integration involves a two-step process. First, the routing phase focuses on designing the optimal path for the harnesses within the aircraft structure. This step ensures that the harnesses are efficiently and safely positioned to avoid interference with other systems. Next, during the mounting phase, appropriate brackets are selected to securely attach the harnesses to the aircraft structure, ensuring stability and reliability throughout the aircraft's use.

Currently, electrical designers have access to a library of standard components for mounting. However, routing in some aircraft structures requires the development of new types of brackets to meet integration constraints. Moreover, these new components must comply with material, mechanical and design constraints to validate the design.

While the routing phase has been automated, the mounting phase still requires manual intervention. To fully automate the harness integration design process, new tools are needed to automate this critical phase.

Objectives of the thesis: This research aims to develop and evaluate an AI-driven method for the automatic generation of optimized 3D bracket designs for aircraft electrical harness integration (see Figure 1). The proposed system will leverage the 3D environment of the aircraft structure and the predefined harness path to ensure accurate routing and seamless integration. Furthermore, it will incorporate manufacturing process constraints and design rules to generate solutions that comply with mechanical and structural requirements.

Methodology: Starting with an analysis of the state of the art, the first step will focus on formalizing the problem and requirements, with particular attention to the functional, structural and mechanical requirements of the brackets. The modeling of the predefined harness path will also be studied in detail. Furthermore, the manufacturing constraints and design rules will be systematically formalized.

Building on SAFRAN’s extensive experience in designing such brackets, data will be col-lected to create and populate the database required for training. This dataset will include representative 3D environment data from aircraft, covering mounting interfaces and sur-rounding structures, as well as existing bracket designs. To enhance robustness, the da-taset will be augmented with synthetic variations of bracket geometries and preprocessed to standardize and prepare the data for training.

Given the nature of CAD models, graph-based and sequence-based representations will be explored, as these have recently shown promising results in learning tasks. Other rep-resentations, such as point clouds, voxels, and meshes, will also be considered and may serve as intermediate representations during the training phase.

The development of the generative AI model will then proceed, leveraging approaches such as conditional GANs, variational autoencoders, or diffusion models. These models will be conditioned on environmental and functional constraints to generate viable bracket geometries. To ensure compliance with mechanical and manufacturing requirements, con-straint-based optimization techniques (such as physics-informed neural networks or to-pology optimization) will be integrated into the generative pipeline.

Depending on the adopted AI model, the output geometries may need to be converted into either editable or dumb CAD models that are ready for manufacturing. Once trained, the proposed approach will be validated against known feasible solutions using quantita-tive metrics (such as weight, stiffness, and harness path compliance) and will also involve expert qualitative assessments to ensure the generated bracket designs meet aircraft in-stallation standards.

Funding category: Cifre

SAFRAN et ANRT

PHD title: Doctorat de l'Ecole Nationale Supérieure des Arts et Métiers

PHD Country: France

Master’s degree in Aerospace, Mechanical Eng., Computer Science, or related field
Strong interest in AI, CAD, and design automation
Experience with optimization or 3D modeling is a plus

Additional Information

Work Location(s)

Number of offers available 1 Company/Institute Arts et Métiers ParisTech Aix en Provence Country France City Toulouse Geofield