Building Low Power XR/AI Glasses

This talk shares our journey in building an open-source prototype of low-power XR/AI glasses—from early design sketches to a working system that supports real-time spatial perception and interaction. Unlike smartphones or headsets, XR glasses face tight constraints on size, power, thermal envelope, and compute. We tackle these challenges through careful hardware-software co-design: selecting off-the-shelf components, integrating microcontrollers and sensors, designing lightweight firmware, and developing spatial applications that can run within milliwatt budgets. I'll walk through the architecture of our glasses prototype, including its communication with a companion device, custom 3D-printed enclosures, and support for AI tasks like object recognition and pose estimation. The goal of this effort is to democratize access to XR systems research—by offering a reproducible, hackable platform that lowers the barrier to entry for researchers, students, and developers alike.