AI for automatic visual inspection

Development of AI methodologies based on Deep-Learning to extend the implementation of usual approaches to complex industrial applications (little/no data, potentially of poor quality). The work is carried out along two major axes.

Generation / Exploration of data

  • Methodologies for the generation of synthetic images

  • Unsupervised learning methods / clustering for the exploration and the analysis of visual data

New algorithmic approaches

  • Transfer Learning approaches

  • Development of methodologies for the progressive deployment of AI technologies according to the amount of available data and their level of qualification.


Vision based localization

Development of vision-based localization tools with the objective of determining the sensor’s 3D pose. For this, several technologies are developed:

  • Trajectory estimation

  • Geolocalization

  • Registration

  • 3D object tracking


In the field of manufacturing, this work is implemented in several applications.

Augmented reality


Development of SLAM algorithms combined with model-based constraints to meet industrial requirements.

Picking


Development of innovative 3D registration approaches for robotic guidance

Localization


Development of multi-sensor SLAM software for the localization and guidance of operators and/or autonomous robots (AMR, UxVs).

VIDEO IA Manuf – Loc

Natural Language Processing (NLP)

By exploiting NLP technologies, applications allow a natural interaction of the operator with his tool as well as a better understanding of user expectations.

Discovery, Structuration et Knowledge Management (DMS 2.0)


We integrate data-mining tools into Electronic Document Management (EDM) technologies.

Voice dictation for semi-structured reports


NLP tools allow to simplify the report writing process of maintenance reports, by automatically structuring dictated reports for integration into automated processes.