What Is a Geofence? — AI Glossary

What it is: A geofence is an invisible virtual boundary drawn around a real-world area, defined by GPS coordinates. Software watches whether a device or vehicle is inside or outside the boundary and triggers actions based on that. Used in robotaxis, delivery drones, marketing apps, fleet tracking, and prison ankle monitors.
Who it is for: Anyone trying to understand why Waymo only drives in 11 specific US cities, why a Tesla Robotaxi only works inside Austin, or why a marketing app pings you when you walk past a Starbucks.
Best if: You want a plain-English answer for what a geofence is and how it shapes the AV industry.
Skip if: You already write GIS software. Want one practical AI explainer every morning? Subscribe to our free daily newsletter.

What is a Geofence?

A geofence is a virtual perimeter drawn around a physical location. It is defined by a set of GPS coordinates — sometimes a simple circle around a point, sometimes a detailed polygon following streets and city blocks. Software running on a phone, a vehicle, or a server checks the device’s current location against that boundary. When the device crosses the line, an action fires: a marketing notification, an automated check-in, a vehicle handoff, an alarm.

The word combines geography + fence. The simplest way to picture it: an invisible chalk line around a neighborhood. Cross it and something happens.

Why does a Geofence matter?

Geofences matter because they are how every self-driving company defines where their cars are allowed to drive in 2026. Waymo operates inside detailed geofences in 11 US cities: Phoenix, San Francisco Bay Area, Los Angeles, Austin, Atlanta, Miami, Dallas, Houston, San Antonio, Orlando, and Nashville. Cross the geofence boundary and the car will not pick you up. Tesla’s Robotaxi launched in Austin in June 2025 inside a five-mile geofence; by May 2026 it had grown to 173 square miles. Baidu’s Apollo Go runs inside geofences in 26 cities globally, with Wuhan’s covering about 3,000 square kilometers. The geofence is the single most important constraint on whether a robotaxi service is available where you live.

Geofences exist because autonomous-driving software has to be tested, mapped, and certified for each specific area. Inside the geofence, the operator has driven millions of practice miles, knows where every fire hydrant and unusual intersection is, has cleared regulatory approval, and has remote operators on standby. Outside the geofence, none of that is true.

How does a Geofence work?

A geofence is a polygon or radius stored as a list of latitude/longitude coordinates. A device (phone, vehicle, drone) tracks its own GPS position, often supplemented by Wi-Fi triangulation, cellular tower data, and inertial sensors for accuracy. Software runs a constant point-in-polygon check: am I inside the perimeter or outside? When the answer changes, the boundary “fires” an event. Practical implications for beginners:

• Robotaxi geofences are dynamic. Waymo regularly expands its boundaries city by city, adding freeways, airports, or new neighborhoods as the testing data builds up. The geofence in San Francisco in May 2026 is much larger than the one in 2023.
• Geofences fail at edges. GPS accuracy drops near tall buildings, in tunnels, and indoors. A robotaxi or delivery drone near a geofence boundary may behave erratically — refusing a pickup, refusing a drop-off, or rerouting.
• Marketing apps use geofences too. Starbucks, Target, and most retail apps fence their stores; when you walk inside the fence, the app fires a push notification or unlocks loyalty features.
• Fleet operators use them for compliance. A rental car company can ban its cars from leaving a state. A logistics company can verify that a driver actually arrived at the delivery zone.

One subtle point that matters for the autonomy debate: a geofenced autonomous vehicle is not the same thing as a fully autonomous one. Inside its fence, a Waymo is meaningfully driverless. Outside the fence, the same hardware and software is useless. This is the practical meaning of the SAE Level 4 designation — “high autonomy, but only inside a defined operational design domain.” (See our Special Report on the state of self-driving in 2026 for the full deployment map.)

Related terms

• Lidar
• Robotaxi
• Artificial intelligence
• Machine learning
• Computer vision

Learn more on Beginners in AI

• The Real State of Self-Driving in 2026 — our Special Report
• Waymo Explained
• Tesla FSD Explained
• Computer Vision in Autonomous Vehicles
• A Short History of Self-Driving Cars

Sources and further reading

• Waymo — service area expansion blog
• NHTSA — autonomous-vehicle crash reporting (location data)
• Geofence — Grokipedia

Last reviewed: May 2026. Robotaxi geofence boundaries are continuously expanded by operators — check Waymo, Apollo Go, and Tesla service maps directly for the current scope where you live.