For any autonomous drone, it starts with one question: Where am I?
HYFIX H1 answers it in one chip: positioning, navigation and compute together, with RTK corrections from @GEODNET.
Great breakdown of why positioning comes first. 👇
Every autonomous machine has to answer 3 questions before it can do anything:
Where am I?
What’s around me?
Where am I going?
Cameras, lidar, radar and software handle a lot of 2 and 3. Question 1 is a different kind of problem. It starts with satellites, and standard GNSS is only accurate to a few meters.
A few meters is:
🚁 a drone that isn’t where it thinks it is, mid-flight
🌱 a mower on the wrong side of the lawn edge
🚗 a car in the wrong lane (a highway lane is about 3.7 m wide)
RTK brings it down to centimeter-level, using corrections from a nearby base station. GEODNET provides those through 22K+ stations across 177 countries, the world’s largest RTK network.
Which of the three do you think is hardest for machines to get right? 👇
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.
75 drones. Built by hand.
Jimmy brought the smallest one to the office. It fits on a quarter and flies in HD.
The full conversation about drones, autonomy, and the H1 chip is on our YouTube. Worth the watch. 🎬
Mike sat down with a guy who’s built over 75 drones by hand.
The smallest one fits on a quarter. The biggest one will “take your head off.”
Ep. 1 of the HYFIX H1 Series is live. Full video on our YouTube channel! 🎬
How precise is the HYFIX H1P heading?
Heading error sounds small in degrees, so here is what it means. Fly a drone 500 m on a heading that is off by 2°, which is easy to get from a magnetic compass near motors and wiring, and you end up about 17 m to the side of where you aimed. At 0.2°, it's about 1.7 m.
We tested the H1P next to two other leading dual-antenna GNSS receivers. Each was stationary and compared with a reference: the average heading measured with antennas 2 m apart.
With antennas 1 m apart, H1P's typical heading error was 0.18°, about 1.6 m of sideways offset over 500 m. At 50 cm spacing it was 0.27°, about 2.4 m. H1P sits in the same class as the other receivers we tested. At 20 cm spacing every receiver widens to around 1°, because a shorter baseline turns the same measurement noise into a bigger angle.
Why it matters: a compass points to magnetic north, and anything magnetic nearby can pull it away. Dual-antenna GNSS heading doesn't use the magnetic field at all. It's measured from the satellites, it's already true north, and it works while the drone sits still.
The plot shows each receiver's heading spread around the reference direction. A narrower arc means steadier heading.
@HyfixAI H1P now tracks Xona PULSAR X1 and X5 signals live.
Xona's satellites orbit about 520 km up, far closer than GPS, so their signals arrive much stronger. In a live, clear-sky pass on 27 September 2026, the H1P tracked Xona for about 5 minutes and measured:
Mean signal: X5 52.8 dB-Hz and X1 49.7 dB-Hz (peaks 62.6 and 58.1)
Both are about 7x stronger than the matching GPS signal on average (+8.6 dB), and up to about 90x (+20 dB) at the best moment
Above 40 dB-Hz for 94% (X1) and 97% (X5) of the pass
Doppler swings smoothly from +33 kHz to −37 kHz, the signature of a fast-moving LEO satellite
Why it matters: a stronger signal is easier to keep where GPS struggles, such as city streets and under trees, and it tends to be harder to jam. Each +10 dB is 10x more signal power to spare.
The plot shows C/N0 (top) and Doppler (bottom) for X1 and X5 through the pass. The grey lines are the GPS L1 (solid) and L5 (dashed) average from the same recording, with the shaded band marking the min–max across the five satellites that have both bands.
New office. Same obsession with precision. 🏢
@HyfixAI Spatial Intelligence is growing. More room to build the chip powering the next generation of drones and autonomous systems.
R&D lab. Podcast studio. Shipping room. Main floor.
Welcome to the new HQ.
📍 Santa Clara, CA
@KryptoInsider1@sopranos_ak Same person building the chip and the network underneath it also means fewer integration gaps, which matters as much as sourcing when you're trying to compete on reliability, not just cost!
@KryptoInsider1 Good catch! Positioning is the one piece DJI still sources externally, and that's the exact gap @GEODNET 's RTK network is built to close, sub centimeter accuracy on decentralized U.S. infrastructure. @HyfixAI is putting it to work at the chip level!
@KryptoInsider1 This is exactly the gap we built the H1 chip for. Flight control, AI, and secure comms in one chip, with @GEODNET's RTK network built in for sub centimeter positioning. No separate GNSS chip needed, and no dependency on foreign hardware!
90% of global drones rely on foreign supply chains. The @HyfixAI H1 solves this. 🇺🇸🚁
Made in America, the H1 chip delivers low-cost, sub-centimeter RTK spatial intelligence for autonomous drones & robotics—powered by @GEODNET.
Securing the supply chain for Physical AI. 🌐
📺 Full breakdown video 👇
#MadeInUSA