Valeo.ai

Connected to the whole academic world worldwide, our Artificial Intelligence Research Center is committed to cutting-edge automotive applications. We are spearheading ambitious research in AI, especially in assisted and autonomous driving. Leveraging state-of-the-art AI, we pioneer advances that redefine the future of automotive.

Scientific research

Valeo.ai tackles the key challenges that autonomous vehicles encounter in everyday driving. Advanced Driving Assistance Systems sometimes fall short in accuracy and reliability, particularly under complex scenarios of malfunctioning traffic lights, missing lane markings, adverse weather conditions, and other road users behaving abnormally.

Our mission is to overcome those obstacles and enhance automated driving, enabling safer and more efficient autonomous travel in any environment, worldwide.

Team presentation

The valeo.ai center spearheads AI research and applications applied to the automotive industry. With excellent skills, our teams include experts in generative AI and multimodal understanding, computer vision and scene interpretation, machine learning (Core Machine Learning) and predictive and uncertainty modeling.

Collaborative projects

Confiance.ai

Confiance.ai is a technological research program aimed at securing, certifying, and enhancing the reliability of artificial intelligence (AI) systems. The program, launched by the Innovation Council, focuses on developing methods and tools for industrial players to engineer and deploy AI-based systems. With a strong ambition to break down barriers associated with AI industrialization, Confiance.ai addresses the scientific challenges of trustworthy AI and provides tangible solutions for real-world deployment. The program adopts a strategy of progressive advancement, starting with data-based AI solutions and gradually moving to more complex problems and industrial use cases.

Multitrans

The project MultiTrans aims to accelerate the development and deployment of autonomous vehicles (AVs) by addressing the challenges of perception, decision, and control in open environments. The project focuses on vision-based embedded systems and proposes a novel approach to transfer learning and domain adaptation, enabling AVs to operate safely and reliably in a wider range of situations. The expected impacts and benefits of the project include advances in transfer and frugal learning, multi-domain and multi-source computer vision, and the development of a robotic autonomous vehicle model demonstrator combined with a virtual world model.

ASTRA

Within the joint lab with Inria, we study 2D vision and 3D perception for robust scene understanding. Our research focuses on relaxing the use of abundant data and supervision, stepping towards weak-/un-supervised vision algorithms, while providing models that are more interpretable. We primarily address autonomous driving but our research expands to a variety of indoor and outdoor applications.

ELSA

The ELSA project aims to establish a virtual center of excellence on safe and secure AI technology to address fundamental challenges hindering the deployment of AI. The project will develop a strategic research agenda focusing on technical robustness, privacy, and human agency, and will tackle three grand challenges: robustness guarantees, private collaborative learning, and human-in-the-loop decision making. The initiative builds on the ELLIS network of excellence and will connect over 100 organizations and 337 fellows and scholars to drive the development and deployment of AI technology that promotes European values.

EXA4MIND

The EXA4MIND project aims to democratize access to and enable connectivity across EU supercomputing centers, allowing for innovative solutions to complex everyday problems and addressing challenges in data analytics, Machine Learning, and Artificial Intelligence at scale. The project will build an extreme data platform that combines large-scale data storage systems with powerful computing infrastructures, enabling integration with diverse data sources and supporting advanced data analysis pipelines for knowledge extraction.

ELLIOT

Leveraging Europe’s AI talent and supercomputing infrastructure, the EU-funded ELLIOT project will develop a family of open trustworthy Multimodal Generalist Foundation Models: AI systems designed to learn general knowledge and patterns from massive amounts of data of various types – from videos, images and text to sensor signals, industrial time series and satellite feeds – and efficiently transfer the knowledge gained to different downstream tasks.