 |
||
|
||
Home > TERATEC FORUM > Workshops > Workshop 6
Workshop 05 - 09:00 am to 12:30 am
Applications of AI in research and industry
Chaired by Stéphane Requena, Director Innovation & Technology, Genci and Patrick Fabiani, AI Roadmapping & Advanced Scientific Studies, Dassault Aviation
Foundation Models on Wheels: Large Scale Self-Supervised Learning for Autonomous Driving
By Florent Bartoccioni, AI specialist, Valeo
What if cars could learn from every drive, every moment on the road? Traditional approaches to autonomous driving have long relied on predefined ontologies and detection methods. While these systems are semantically rich, they often falter under the unpredictable conditions of real-world driving. This limitation arises because these traditional methods are built upon static definitions— thus, what is not defined does not exist.
This rigid structure, combined with the high costs and intensive labor required for manual data annotation, presents significant scalability and efficiency challenges. In contrast, the field of self-supervised learning offers a compelling solution.
Such methods learn useful and transferable features directly from vast amounts of unlabeled video data, circumventing the need of costly manual annotations. The resulting models, often referred to as `foundation models`, serve as a versatile base from which specific applications can be developed.
At valeo.ai, we are at the forefront of developing these foundation models that redefine how automated vehicles perceive and react to their surroundings.
In this presentation, I will share with you several innovative methods developed and currently being refined at valeo.ai, showcasing our progress and future directions in autonomous driving technologies.
 |
Biography : Florent Bartoccioni is a researcher at Valeo.ai, a research lab for autonomous driving. He began his AI journey at École normale supérieure de Rennes, completing a Master’s in computer science before joining Valeo.ai for his PhD. Currently, his research intersects computer vision, robotics, and large-scale self-supervised learning. Florent also acts as a scientific advisor, guiding R&D teams across multiple sites, effectively translating his advanced research on dynamic scene understanding and 3D perception into practical AI-driven solutions within the automotive industry. |
|---|
Register now and get your badge here >>>