From Foundation Models to Physical Intelligence: Embodied, Wearable, and Edge AI

This workshop brings together researchers and practitioners working on foundation models, embodied intelligence, biosignal processing, and efficient embedded AI. It will explore recent advances in whole-body control and vision-language-action models for humanoid robots; self-supervised foundation models for EEG and other biosignals; and practical frameworks and techniques, such as the AIfES library, for deploying machine learning on resource-constrained embedded platforms.

Particular attention will be given to efficient model architectures, pre-training strategies, model compression and quantization, hardware-aware optimization, and ultra-low-power inference. Through invited talks, technical demonstrations, and open discussion, the workshop aims to identify common scientific challenges, foster collaboration between AI and hardware researchers, and explore opportunities for future research projects in robotics, wearable intelligence, TinyML, and edge AI.

The workshop will be of interest to researchers, engineers, doctoral students, and industry practitioners working on machine learning, deep learning, foundation models, embedded systems, signal processing, robotics, and AI hardware.

The event is taking place at Télécom Paris, Amphitheatre 4. A zoom link will be provided for people who wants to attend remotely.