Project Aria @Meta

Excited to share our latest collaboration with the @ComputerHistory Museum and Lighthouse Guild (@lighthousegld ) to redefine accessible cultural experiences! 🏛️🤖

Through an early prototype using Project Aria research glasses, we’re exploring how Conversational AI Guides and Spatial Audio Navigation can make museum tours deeply engaging and truly inclusive—especially for blind and low vision visitors.

By grounding technological innovation in community-led research, we’re building a more accessible future for spatial computing.

Read the full update on our blog: computerhistory.org/blog/ai-guided…

Teaching vision models how actions feel, not just how they look. 👓💪

Congratulations to our partners at the University of Maryland on their paper accepted to #NeurIPS2026!

Introducing MotorSense, the first large-scale multimodal dataset pairing egocentric video with surface EMG. Spanning 100 hours of unscripted hand manipulation captured using Project Aria glasses and bimanual EMG armbands, MotorSense bridges the gap between visual perception and physical motor intent.

Explore the dataset and paper: 🔗 cs.umd.edu/~edessale/moto…

Exciting work from the #EgoXperience team! 🚀 They are turning human experience into world-grounded representations for embodied AI.

By utilizing both simulated data from the Aria Synthetic Environments dataset and real-world data like the Aria Digital Twin dataset, their 1.2B parameter EgoX model achieves impressive single-pass joint reconstruction of scenes, hands, and objects from just monocular observations. 🕶️🌐

Read the paper and check out the data: ropedia.github.io/EgoXperience/

We’ve partnered with the Computer History Museum to build an AI-guided museum tour experience! 👓✨ Read the full blog post to learn more about the future of accessible museum experiences:
computerhistory.org/blog/ai-guided… youtu.be/RlXisxLm8lM?si…

Check out CoMind, an ECCV 2026 dataset exploring how people collaborate, from two participants; perspectives!

Captured with Project Aria glasses, CoMind combines synchronized egocentric video, gaze, audio, and 3D scans to study shared attention, anticipate interactions, and predict handovers. A rich resource for building AI that understands and supports human collaboration.

Fantastic work by the CoMind team! 📷 @xiwang1212