Development of a massive point cloud engine for industrial infrastructure h/f

Depuis 50 ans, nous sommes animés par la conviction que les ingénieurs, par leur audace, leur expertise et leur capacité à faire bouger les lignes sont les architectes du monde de demain.
Tous les jours, nos 5 700 collaborateurs innovent au contact de leurs clients. Ils accompagnent les plus grands projets de l'ingénierie mondiale dans les domaines de l'énergie, de la santé, des sciences de la vie et des transports.
Groupe international en forte croissance, nous sommes présents dans 14 pays (Europe, Moyen-Orient, Asie, Afrique).

Description du Poste
Context

Building Information Modelling (BIM) is a powerful approach used by industrial building managers in order to monitor the state of their buildings, anticipate any potential failure or malfunction, and plan for maintenance operations. A BIM model should reflect the state of the building as accurately as possible (as-is modelled), which means that the initial model must be accurate, and which also means that there should be procedures in place which guarantee that this model is always up-to-date with the building status.

Creating a BIM of a newly built building is relatively an easy task since all the plans of new buildings are digital and could be integrated easily into an efficient and accurate BIM model. However, for existing buildings, this task is much more difficult and time consuming and most often, the results are limited due to the compromises made between cost and level of details.

Because of the importance of such models, many studies and industrial solutions are under development in order to facilitate the process of generating an as-is BIM model of existing buildings. The most promising is scanning the building using a mobile laser scanner (MLS), which generates a huge number of points, contained in a cloud of points, which is laterprocessed in order to reconstruct the details of the scanned building. The progress of the technologies are allowing the production of point of cloud with an amazing resolution and aspect ratio (+/- 5mm for a nuclear reactor).

Internship Description

The objective of this internship is to implement a solution that could transform a massive cloud of points into a 3D model with the least amount of information loss as possible. This solution should be also able to identify the objects within the installation automatically, by using an object library as a reference for classification.

In this context, the trainee will be doing the following tasks:

1. Conduct a state of the art on the subject of clustering

2. Generate a list of existing solutions of Scan2Bim

3. Propose a method to process the cloud of points with minimum information loss that will incorporate the most advance image processing techniques combined with artificial intelligence algorithms




Profil
Skills and Knowledge

1. Studying for Bachelor or Master degree in Computer Science

2. Strong research and analytical skills

3. Strong skills in clustering and image processing

4. Big data skills will be appreciated

5. A good level in English

Duration

The expected duration is 6 months. He/She will be preferably work 5 days a week from our site in La Defense/Courbevoie.






Chez Assystem, chacun est acteur et entrepreneur de sa carrière. Nous vous offrons l’opportunité et les moyens d’atteindre votre plein potentiel.

Alors rejoignez-nous et faites avancer à grand pas l’ingénierie de demain !