As robots, drones, autonomous vehicles, and machines move deeper into the physical world, knowing exactly where they are becomes foundational. Point One Navigation is building the precision location infrastructure designed to make that possible at massive scale.
Aaron Nathan, CEO and founder of Point One Navigation, joins Greg to unpack why traditional GPS is not precise enough for many autonomous systems, how his team can take positioning accuracy from roughly five meters down to just a few centimeters, and why location will become a critical layer of infrastructure for the physical AI economy.
Aaron’s path into autonomy started long before self-driving cars became mainstream. As a student at Cornell, he helped build the university’s DARPA Grand Challenge vehicle during an era when autonomous vehicles required racks of computers, foundational software had to be written from scratch, and the supporting robotics ecosystem barely existed. That experience taught him an enduring lesson: no matter how sophisticated perception or control becomes, a machine that does not reliably know where it is cannot operate safely.
That insight eventually led to Point One Navigation, where Aaron and his team are building the location layer for everything from autonomous lawnmowers and agricultural equipment to drones, vehicles, and large robotic fleets.
Greg and Aaron explore:
Aaron’s early robotics journey, from soccer-playing robots at Cornell to helping build an autonomous vehicle for the DARPA Grand Challenge
Why the hardest problems in robotics often live at the intersection of software and hardware, rather than inside either discipline alone
How lessons from Coherent Navigation, cybersecurity, privacy, and large-scale software platforms shaped the architecture and philosophy behind Point One
Why Aaron chose to build foundational infrastructure for autonomy instead of creating another autonomous vehicle company
How Point One makes GPS roughly 100 times more accurate, taking positioning from meters of uncertainty down to centimeter-level precision
Why high-precision location matters for machines even when conventional GPS works perfectly well for humans
The economics and outdated infrastructure that kept precision positioning inaccessible for years, despite the underlying technology already existing
What it takes to build and operate a global corrections network, including thousands of physical stations and a massive real-world sensor infrastructure
Why Point One is expanding beyond corrections into a broader location platform with positioning engines, hardware such as Atlas, APIs, and sensor fusion
How sensor fusion allows robots to keep navigating when GPS conditions deteriorate under trees, between buildings, or in other difficult environments
Why different machines, including cars, drones, and boats, require different approaches to positioning and sensor fusion
The edge cases that appear when robotics systems move from small pilots to fleets of thousands or even hundreds of thousands of machines
Why Aaron believes physical AI has an unavoidable problem that software does not: physics is always on
How global positioning, SLAM, vision, lidar, and other sensing technologies can work together instead of competing for the same role
Why reliable infrastructure and the "boring" problems of robotics matter more than impressive demos when companies begin deploying at scale
How large AI models are opening new robotics use cases by allowing machines to operate in increasingly uncertain and dynamic environments
What needs to happen for millions or billions of robots to safely coordinate across factories, warehouses, roads, ports, yards, and airspace
Why Point One’s biggest constraint on growth is now hiring across engineering, R&D, customer success, and go-to-market teams
Aaron’s advice for technical founders: solving the technical problem is only one part of building a successful company, and understanding customer need, value, communication, and willingness to pay is equally important
Lessons from the DARPA era that still guide Aaron today, especially the value of keeping individual components simple enough to understand when something inevitably fails
Aaron also shares the thinking behind Fuse, Point One’s conference focused on localization and real-world system design. Rather than creating a traditional product conference, the team is bringing together companies that have successfully deployed physical systems to discuss what actually works, what breaks, and what engineers can learn from operating at scale. The next Fuse event takes place November 19 in San Francisco.
For robotics teams moving beyond demos and into real-world deployment, this conversation offers a detailed look at one of the least visible but most important layers of physical AI: making sure every machine knows exactly where it is, even when the environment does not cooperate.
Conference website: https://fuse.pointonenav.com/sf
FUSE discount code: MACHINEMINDS30
Learn more about Point One Navigation: https://www.pointonenav.com/
Connect with Aaron Nathan: https://www.linkedin.com/in/aaron-nathan/
Connect with Greg Toroosian: https://www.linkedin.com/in/gregtoroosian/