Beyond GPS Accuracy: Why Centimeter-Level Navigation Is the Missing Infrastructure for Autonomy

A self-driving car traveling at highway speed needs to know its position to within a few centimeters. Not meters — centimeters. The difference between the center of a lane and its edge is roughly one meter. The difference between a safe following distance and a collision can be measured in tens of centimeters. The margin between a vehicle on the road and a pedestrian on the sidewalk is often less than a meter. For autonomous systems operating in the physical world, positioning accuracy is not a nice-to-have metric — it is a safety-critical requirement.

GPS, the positioning system that everything else depends on, provides accuracy of approximately 3–5 meters under optimal conditions. In urban canyons — where buildings reflect and block signals — accuracy degrades to 10 meters or worse. Under tree canopy, in parking garages, near bridges, and in any environment where the already-weak GPS signal encounters obstacles, accuracy can degrade further or the signal can drop entirely. This is the fundamental bottleneck that has limited the deployment of autonomous systems across multiple industries.

The Accuracy Gap

The table reveals a consistent pattern: virtually every autonomous or precision application requires positioning accuracy that is one to two orders of magnitude better than what standard GPS provides. The gap between what GPS delivers and what autonomy requires has been the defining constraint on the deployment of self-driving vehicles, autonomous drones, precision farming equipment, and automated construction and logistics systems.

How the Industry Has Worked Around GPS

The autonomous vehicle industry has spent billions of dollars developing workarounds for GPS inadequacy. Lidar systems create detailed 3D maps of the surrounding environment. Cameras use computer vision to identify lane markings, signs, and obstacles. Inertial measurement units (IMUs) track vehicle motion between GPS fixes. HD maps provide centimeter-accurate reference data that vehicles use to localize themselves by matching sensor observations to pre-mapped features.

These workarounds are expensive, complex, and fragile. A lidar-equipped autonomous vehicle carries sensors costing tens of thousands of dollars. HD maps must be continuously updated as road conditions change — a single moved construction barrier can invalidate a map that took weeks to build. Camera-based lane detection fails in rain, snow, fog, and darkness. IMUs drift over time and must be periodically corrected by an absolute positioning source — which brings the problem back to GPS.

What Centimeter-Level LEO Navigation Changes

LEO navigation constellations like Xona's Pulsar deliver centimeter-level accuracy natively, without requiring base stations, HD maps, or expensive sensor suites. The stronger signal — 170 times more powerful than GPS — means it works in environments where GPS fails: urban canyons, under tree canopy, inside parking structures, and near bridges. The signal authentication built into Pulsar prevents spoofing, ensuring the position data is trustworthy. And because the system operates independently of GPS, it continues to function even when GPS is jammed.

For autonomous vehicle developers, this changes the architecture of self-driving systems. Instead of relying on a complex fusion of lidar, cameras, IMUs, and HD maps to compensate for GPS inadequacy, vehicles could use centimeter-accurate satellite positioning as the primary localization source, with other sensors providing confirmation and obstacle detection rather than positioning. This simplification could reduce the cost and complexity of autonomous vehicle systems while improving reliability in challenging environments.

Precision Agriculture: The First Scaled Market

Precision agriculture may be the first industry to adopt LEO-based centimeter navigation at scale. Modern farming equipment — planters, sprayers, harvesters, tillage implements — already uses GPS guidance for straight-line driving and controlled traffic patterns. RTK corrections provide the centimeter accuracy needed for row-level planting and variable-rate application. But RTK requires base station infrastructure that is unavailable in many agricultural regions, and the corrections degrade with distance from the base station.

A LEO navigation system providing native centimeter accuracy without base stations would eliminate the need for RTK infrastructure in agriculture, extending precision farming capabilities to regions where base stations are uneconomical to deploy. The economic value is substantial: precision planting can improve yields by 5–10%, and precision spraying can reduce chemical inputs by 10–20%. At the scale of global agriculture, these improvements translate to billions of dollars in additional value.

Drone Delivery and Urban Air Mobility

The drone delivery industry — pursued by companies like Wing (Alphabet), Amazon Prime Air, and Zipline — faces a positioning accuracy problem that mirrors autonomous vehicles. A delivery drone must navigate to a precise landing point, often in a backyard, on a rooftop, or on a designated pad in a commercial area. GPS accuracy of 3–5 meters is insufficient for safe landing in constrained spaces. Urban environments degrade GPS further, and the near-ground environment where drones land is exactly where GPS performance is worst.

Urban air mobility — the broader vision of electric vertical takeoff and landing (eVTOL) aircraft operating in cities — faces even more stringent requirements. Landing on a vertiport pad requires meter-level accuracy at minimum, and regulatory frameworks are likely to require better. The Federal Aviation Administration's evolving rules for advanced air mobility emphasize precise navigation as a safety prerequisite. Centimeter-accurate LEO navigation could provide the positioning foundation that enables both drone delivery and urban air mobility to scale beyond demonstration programs.

The Infrastructure Layer for the Autonomy Economy

The common thread across autonomous vehicles, precision agriculture, drone delivery, construction automation, port operations, and railway positioning is that each requires centimeter-level positioning that GPS cannot reliably provide. Each industry has developed expensive workarounds — RTK, lidar, HD maps, inertial navigation — that partially solve the problem but add cost, complexity, and failure modes.

LEO navigation constellations have the potential to become the foundational infrastructure layer for the entire autonomy economy — the positioning substrate that every autonomous system depends on, much as GPS became the positioning substrate for the mobile internet. If Xona and its competitors can deploy operational constellations and achieve the centimeter accuracy they promise, they will not just be competing with GPS. They will be enabling an entirely new class of applications that GPS was never capable of supporting.

Frequently Asked Questions