dynamic_robot_localization

Supports extensive customization via YAML files for localization stages, enabling adaptation from robot self-localization to object pose estimation, as detailed in the use cases examples.

Works with any sensor providing point clouds, including laser scanners and RGB-D cameras, and accepts maps from CAD files, point clouds, or ROS topics, ensuring versatility.

Integrates feature matching (e.g., SIFT, ISS3D) for initial pose and multiple ICP/NDT variants for tracking, with validators to filter implausible estimates, enhancing reliability in dynamic environments.

Couples with OctoMap for probabilistic point cloud integration, allowing the system to handle environments where objects appear or disappear, as specified in the dynamic map update section.