Localisation and maps

Position without satellites

Between tall buildings or under cover, GPS dies. We record with LiDAR, camera and IMU; generate HD maps for autonomous vehicles; the vehicle localises itself by matching live point clouds to that map in real time. It does not rely on satellites alone.

The vehicle locates itself on the orthophoto by fusing its LiDAR with the map. Orthophoto: PNOA courtesy of © Instituto Geográfico Nacional.

What localisation and maps are

Localising means working out the vehicle's pose by comparing live LiDAR with a high-definition map. An HD map describes lanes, edges and drivable areas so the vehicle can decide on its own. We distinguish three: navigation, localisation and perception.

The problem it solves

GNSS fails under roofs, between tall buildings or beside stacked containers, exactly where precision matters. Hand-made maps are slow, costly and go stale when the site changes.

Illustrative diagram

How we do it

01

Record

We drive through the area with the sensor-equipped vehicle.

02

Map

We generate the map automatically from the drives and review it before use.

03

Derive

From the map we work out where the vehicle can drive.

04

Localise

With the map and the vehicle's sensors we calculate its position, even without satellite.

sv sol own localizacion

Localisation on georeferenced LiDAR.

Questions

How is it different from a satnav?

A satnav describes streets for people; a high-definition map details lanes so that a vehicle can decide.

How is it kept up to date?

Each new pass is compared with the previous one and changes are updated.

Can we use our own maps?

Yes. If they are georeferenced, we make use of them.

Let's talk

Tell us your case and we will look at it with your data.

Next

Navigation and control