On 6 October, Kybera attended FUSE Toulouse 2026, a one-day conference on location technology for Physical AI.
Discussions ranged from GNSS resilience and LiDAR-based localisation to train navigation, low-latency networking and industrial robots, with autonomous machines running live in the exhibition hall.
Three broad themes clearly stood out on the day:
Interest in Position/Map-as-a-Service is growing
Localisation is the key that unlocks Physical AI at scale. First, a position is only half an answer unless the system also says how far to trust it. That matters most when GNSS drops out and a vehicle is running on dead reckoning. Second, a shared global map lets a whole fleet work from the same picture of the world, what PointOne called "multiplayer".
When deploying autonomous fleets at scale, positioning and maps delivered as a service, each position with an associated percentage confidence figure, are an attractive alternative to building everything in-house. The market is still early, but the interest is clearly growing.
GNSS spoofing is an increasing threat.
Spoofing and jamming of GNSS systems used to be mainly a military concern. It is evidently clear the risk is now growing for civilian users too, and resilient receivers are becoming a mainstream topic, including (over?)-reliance on satellites to enable LiDAR-based localisation for GNSS-denied environments, and safe train localisation using perception and machine learning.
The lesson for autonomous vehicles is to treat GNSS as one input among several. A system that cross-checks satellite positions against LiDAR, inertial and wheel data can spot a bad fix instead of following it.
Real-world robustness is where the hard work lies
This theme ran through the whole day, and it is the focus FUSE sets itself. A design that works in the Lab can fail in the field, where conditions keep changing and no single sensor sees everything.
In seaport container terminals and industrial sites, the operational environment itself keeps changing. Containers are stacked and moved, cranes travel, industrial vehicles come and go. A useful map has to separate permanent structure from temporary operational requirement, and localisation has to stay reliable either way.
What this all means for Kybera
All three themes sit close to our own work here at Kybera.
We are building multi-sensor mapping and localisation capability to support autonomous horizontal transportation in seaport container terminals and industrial sites. We fuse LiDAR, inertial, GNSS and wheel data into one map, check each GNSS fix before trusting it, and validate our maps against surveyed control points.
We have returned from FUSE fresh with ideas to add to our development roadmap, including a percentage confidence figure for each and every position and maps that stay reliable as operational environments evolve.
If you were also at FUSE Toulouse 2026, or are working in autonomy in seaport container terminals or industrial sites more broadly, we would love to compare notes. Drop us a line at hello@kybera.com
