Pri4L

Smart glasses anywhere.
Spatial intelligence at home.

Mobility is not a feature if you're standing still. Commercial AR optimizes for the pocket — a compute puck with a beefy AI coprocessor, active cooling, and 5G. That's the right answer when you're moving. At home, in a fixed workspace, you're paying weight and thermal penalties for constraints that don't exist. Wall power removes every limit. The glasses become a lightweight, mobile-first display peripheral. The room becomes the brain that serves robust spatial experiences when in range.

BLE / Direct pair WebSocket / ROS2 AR Spatial Hub SLAM · LLM · Map Glasses thin client Phone personal bridge
About

I'm a Product Designer at Meta working on AR and wearables, with recent contributions to Ray-Ban Meta Display. Before Meta, I was Product Design Lead at Magic Leap (2016–2019), where I was awarded six US patents on spatial interface design. Pri4L is my open-source personal infrastructure for hub-and-spoke AR.

What's Shipped
Phase 1 — Complete
  • Live depth feed via RealSense D435i
  • ROS2 topic publishing
  • Live map building via RTAB-Map
  • Map survives restart and relocalizes
  • Android client with ARCore alignment and anchor overlay
In Progress
Phase 2
  • IMU access via Android SensorManager
  • Android app streams camera to hub via RTSP or WebRTC
  • Hub relocalization returns pose and anchor data
  • 360 degree room coverage
The Stack

RealSense SDK — hardware interface
ROS2 Humble — sensor middleware
RTAB-Map — spatial mapping
Ollama (llama3.2) — local LLM inference
ARCore — device-side 6DOF
Kotlin / Jetpack Compose — Android client
rosbridge — WebSocket bridge

The Process

10 ADRs — numbered design decisions
Written threat model — LAN, physical, transport
Gated rollout — phased plan with 010 gates
Privacy POC — documented boundary conditions

Control Center