Meridian — A Navigation OS
Redesigning spatial awareness for the next generation of autonomous systems
Overview
Meridian is a comprehensive navigation operating system built for environments where GPS signals are absent or unreliable — underground transit, dense urban canyons, and indoor industrial complexes. The project spanned three years of research, prototyping, and deployment across twelve cities.
The Problem
Existing navigation systems fail catastrophically in GPS-denied environments. Transit operators, emergency responders, and logistics teams rely on outdated paper maps or expensive proprietary solutions with poor accuracy. Meridian set out to solve this with a hybrid sensor-fusion approach.
Research Phase
We conducted contextual interviews with 84 transit operators across six countries, observing how they navigate signal loss and how errors compound over time. The core insight: people needed confidence signals as much as accurate data.
Design Decisions
The interface prioritises three things: ambient awareness, quick correction, and graceful degradation. When sensor confidence drops below a threshold, the UI shifts into a distinctive low-confidence mode — the design communicates uncertainty rather than hiding it.
Technical Architecture
Meridian fuses data from inertial measurement units, barometric pressure sensors, WiFi fingerprinting, and machine-learned dead reckoning. The entire stack runs on-device, processing at 120Hz with under 4ms latency.
Outcome
Deployed in three metro systems. Achieved sub-metre accuracy in 94% of tested corridors. Average operator trust score: 8.7/10 — up from 3.2 with previous systems.