Most deployed localization systems operate on the assumption that the environment is essentially flat. The vehicle moves in a plane, the map is a 2D representation, and height is either ignored or treated as a minor correction. This assumption works in single-floor facilities and outdoor environments where grade changes are gradual. It fails in buildings, parking structures, and any facility that connects multiple levels with ramps or elevators.
The problem is vertical ambiguity. A column on the second floor of a parking structure produces a nearly identical local LiDAR observation to the same column on the third floor. An open section on level 2 looks, from the inside, like the open section on level 4. Two locations that are meters apart vertically may be perceptually indistinguishable from local sensor data alone.
A 2D localization system cannot resolve this ambiguity by design. It has no representation of vertical structure in the map and no mechanism to distinguish floor levels. When the vehicle transitions between floors via a ramp, the system observes changes in the sensor data that it cannot interpret correctly. In practice this produces incorrect floor associations that are locally consistent but globally wrong, making them difficult to detect until the vehicle attempts to navigate to a destination on the wrong level.
Parking structures are a useful test case for this because they concentrate the problem: multi-story, ramp-connected, similar architecture on every level, and regular operation involving vehicles that traverse all levels. A system that maintains correct floor association reliably in a parking structure under real traffic conditions has genuinely solved the vertical ambiguity problem, not just handled the easy version of it.
If your deployment involves multiple levels or elevation changes, we can show you directly how 3D map-based localization handles floor disambiguation in your specific facility type.
Book a demo →