The next AI revolution is not just happening on screens. It is happening in the physical world.
Robots, drones, autonomous vehicles, and smart machines all need one thing before they can operate safely: Precise location.
Traditional GPS can be off by several meters. For humans, that may be fine. For autonomous machines, it can be dangerous.
GEODNET is building the spatial intelligence layer for Physical AI, using RTK positioning to improve accuracy from meters to within one centimeter.
Over 20,000 base stations.
170+ countries.
11,000+ cities.
130+ enterprise customers.
Physical AI needs real-world infrastructure.
That infrastructure is being built now.
Great to show centimeter-level RTK at Commercial UAV Expo alongside @HyfixAI. Watch the full recap 👇
Commercial UAV Expo 2026, recapped. 🚁
The H1 autonomous system chip.
The GEODASH ag spray drone.
A live GEO-SWARM demo.
And the world's largest RTK network, @GEODNET, all at one booth!
Full recap in the first comment.
Cameras and lidar see what's around a machine.
RTK tells it where it is on Earth.
A robot can spot a fence, a row or a curb with its sensors. But it can't know which field it's in, where the boundary runs or where it was yesterday without absolute positioning.
Sensors see. RTK locates. Physical AI needs both.
GEODNET is that positioning layer: The Rails for Physical AI.
Vidalia, Georgia: home of the sweet onion! 🧅
Vidalia sits in farm country, where planting, spraying, mapping and autonomous equipment all depend on knowing exactly where a machine is. GPS alone can be off by meters. RTK brings it down to centimeters.
Vidalia, GA is now on the GEODNET map!
The network now spans 176 countries with 22K+ stations, one more town at a time. 📍
How RTK works, simply put:
Satellite signals travel about 20,000 km to reach a receiver. On the way, the atmosphere bends and delays them. A standalone receiver can end up meters off.
RTK fixes that:
1. A base station sits at a known, fixed location. It knows exactly where it is.
2. It compares that position to what the satellites report. The gap shows how much error is in the signal right now.
3. That correction streams to your machine over the internet in real time.
4. Your receiver applies it and lands within centimeters.
Meters become centimeters. That's the difference between "somewhere in the field" and "on the row."
The closer the base station, the better the correction. That's why GEODNET keeps adding them, every week.
22,000+ base stations. 175 countries. The world's largest RTK network.
"The autonomous lawnmower is the same price as the standard one." 🌿
Remove the handlebars and the controls, and the machine still has to know exactly where it is.
Would you pick the robot if it cost the same? 👇💭
Check out Part 2 of ROBOSTRATEGY x GEODNET. Link in first comment!
GEODNET retweeted
🌍 Interested in RTK GNSS, precision positioning or DePIN?
Check out GEODNET StableHexes and explore how GEODNET is developing its global GNSS reference station network.
🗺️ Explore the StableHex Map:
consolebeta.geodnet.com/map
#GEODNET #StableHex #GNSS #RTK #DePIN
Physical AI moves atoms, not just bits. 🤖 Mike Horton on what sits underneath it: the sensing infrastructure.
Full interview with @DePINConnection in the comments 👇
Physical AI Moves Atoms, Not Just Bits
Mike Horton (@mikeahorton) of @GEODNET says the key to physical AI is the first word: these are machines doing things in the real world, and networks like GEODNET provide the sensing infrastructure underneath.
Watch the series here → youtube.com/playlist?list=PL…



