Beluga AMCL

Beluga AMCL is a ROS node based on the Beluga library that aims to be fully compatible with the Navigation 2 AMCL node.
The compatibility between beluga_amcl and its longstanding counterparts in the ROS ecosystem provides a simple migration path for projects that want to be able to integrate the power and modularity of the Beluga library in an existing nav2_amcl-based project.

Table of Contents

• Beluga AMCL
  • Table of Contents
  • ROS Interface
    • Parameters
    • Subscribed Topics
    • Published Topics
    • Published Transforms
    • Advertised Services
  • Performance
  • Next Steps

ROS Interface

Parameters

Beluga AMCL currently supports the majority of ROS parameters used in Navigation 2 AMCL.
See Beluga AMCL documentation for further reference.

Subscribed Topics

The subscribed topic names can be changed with the parameters map_topic, scan_topic, and initial_pose_topic.

Alternatively, and instead of the scan_topic, one may set the point_cloud_topic. Point clouds are assumed to be contained in a z = constant plane in the base frame of reference. It is further assumed this plane is the same plane where the map and pose estimates are defined. If this is not the case, Beluga AMCL will misbehave. It is on the user to filter point clouds and make sure these assumption hold.

Published Topics

Published Transforms

The frame names can be changed with the parameters global_frame_id, odom_frame_id and base_frame_id. Defaults are map, odom and base.

Advertised Services

Performance

Performance reports are periodically generated and uploaded to track performance improvements and regressions. These reports are generated using a set of scripts in the beluga_benchmark package which can be used to compare the performance of beluga_amcl against that of nav2_amcl using a synthetic dataset.

The following plot displays the RSS (Resident Set Size), CPU usage, APE (Absolute Pose Error) and processing latency statistics for both beluga_amcl and nav2_amcl, with particle sizes ranging between 250 and 200000 and sensor model likelihood field.

The following plot displays the RSS (Resident Set Size), CPU usage, APE (Absolute Pose Error) and processing latency statistics for both beluga_amcl and nav2_amcl, with particle sizes ranging between 250 and 200000 and sensor model beam.

Further details can be found in the reports folder here.

Next Steps

• See example launch files showing how to run Beluga-based nodes.
• See available benchmarks for scripts and comparison with other AMCL implementations.

Changelog for package beluga_amcl

2.0.2 (2024-06-18)

• Expand grid concepts to n dimensions (#396)
• Change NDT sensor model default params (#397)
• Contributors: Ramiro Serra

2.0.1 (2024-05-24)

2.0.0 (2024-05-21)

• Please clang-tidy on Ubuntu Noble (#379)
• Add ROS 2 Jazzy to CI/CD and dev workflows (#378)
• Restore particle cloud visualization using markers (#377)
• Enable beluga_amcl ROS 2 nodes' autostart (#376)
• Use [geometry_msgs::msg::PoseArray]{.title-ref} instead of [nav2_msgs::msg::ParticleCloud]{.title-ref} (#372)
• Fix tf broadcasting (#375)
• Remove ament_index_cpp dependency (#374)
• Improve ROS version detection logic (#366)
• Add example for NDT sensor model in beluga_example (#360)
• Add NDT AMCL node and tests. (#347)
• Unify Beluga documentation (#346)
• Add death tests and build in debug mode (#323)
• Migrate [beluga_amcl]{.title-ref} to the new API (#328)
• Relocate [make_random_state()]{.title-ref} method (#324)
• Make test matchers and printers available to users (#300)
• Motion and sensor models are not mixins (#291)
• Integrate views with mixins (#284)
• Transform points in laser frame to robot frame (#276)
• Generalize planar laser scan support (#271)
• Add new discrete landmark and bearing sensor models to the beluga library (#268)
• Add [beluga_ros]{.title-ref} package (#270)
• Make particle cloud message weights represent pdf (#260)
• Add updated performance report (#255)
• Fix missing control to force a filter update on pose updates (#248)
• Change default AMCL params (#244)
• Format pre-commit hook configuration (#245)
• Add cmake-format to pre-commit hooks (#243)
• Rename motion models in Beluga AMCL for ROS 1 (#242)
• Replace ament_copyright with local script (#240)
• Remove ament_flake8 from pre-commit hooks (#237)
• Remove uncrustify and cpplint from pre-commit hooks (#236)
• Refactor resampling policies into filter update control (#233)
• Add ROS 2 Rolling CI infrastructure (#222)
• Add ROS Noetic support to [beluga_amcl]{.title-ref} (#218)
• Fix flaky transform test (#220)
• Fix map to odom transform calculation (#212)
• Add new report after recent performance updates (#208)
• Fix small issues with plotter and update MessageFilter tolerance before generating new report
• Disable branch and function coverage (#205)
• Change default threading mode to the less cpu-consuming one (#203)
• Optimize beam sensor model runtime performance (#200)
• Add AMCL parameter reference page (#196)
• Add tests for the AMCL node

File truncated at 100 lines see the full file