Skip to main content

Listed

, home
FREN

Localization

In French:
Localisation

Definition

Estimation of a robot’s position and orientation within an already known map. A common method, the particle filter, maintains many position hypotheses and discards those that do not match what the sensors see. In Nav2, this is the job of the AMCL module (Adaptive Monte Carlo Localization).

Checked on against the references below. · Written with AI · checked against the sources Source: Nav2 documentation, “AMCL”.

Concrete examples

Hello Robot’s repository for the Stretch 4 describes a system that uses “RTAB-Map for global localization and map building” and Nav2 for planning. Its LOST status flags that the pose can no longer be trusted: “Pose is not trusted. Drive through mapped space or set a new 2D Pose Estimate.” For humanoids such as the Unitree G1, a third-party repository (not from the manufacturer) offers lidar localization against a user-supplied point cloud map, with no mapping solution included: “We do not provide specific mapping solutions”.

What it changes for you

In Nav2, the map → odom transform is provided by a positioning system (localization, mapping or SLAM), and the documentation lists it among “the primary requirements of the Nav2 system”. With AMCL, set_initial_pose defaults to false: the module then waits for an initial_pose message instead of reading the initial pose parameters.

Not to be confused with

SLAM carries out localization and mapping simultaneously; Nav2 presents SLAM Toolbox as “a set of tools and capabilities for 2D Simultaneous Localization and Mapping”. When SLAM Toolbox supplies the map and the map → odom transform, the documentation says to set use_localization:=False so that neither nav2_amcl nor nav2_map_server is launched.

Additional material checked on .

Related terms

References

Back to the glossary