Revision history [back]

First, see my answer to this question, which I think is similar to yours.

Second, are you taking in landmark updates to provide a (potentially discontinuous) pose estimate, or are you using GPS? navsat_transform_node is also necessary if you are working with GPS data.

A typical use case would be:

1. ekf_localization_node or ukf_localization_node instance one: fuse continuous measurement source, e.g., odometry and IMU data. The world_frame for this instance is the same value as your odom_frame; this is typically odom. Alternatively, some robots provide an odometry message and the odom->base_link transform. This is typically less accurate than a solution with fused data sources, but it may fit you needs.
2. ekf_localization_node or ukf_localization_node instance two: fuse the same data as in instance one, but also fuse discontinuous sources like GPS or position updates from landmark observations. The world_frame for this instance is the same value as your map_frame; this is typically map. This instance is really only necessary if you have discontinuous measurement sources.

To more directly answer your question, for navsat_transform_node, the frame_id of the /odometry/gps message that gets created is depedent on the frame_id of the odometry message that gets fed to it.

First, see my answer to this question, which I think is similar to yours.

Second, are you taking in landmark updates to provide a (potentially discontinuous) pose estimate, or are you using GPS? navsat_transform_node is also only necessary if you are working with GPS data.

A typical use case would be:

1. ekf_localization_node or ukf_localization_node instance one: fuse continuous measurement source, e.g., odometry and IMU data. The world_frame for this instance is the same value as your odom_frame; this is typically odom. Alternatively, some robots provide an odometry message and the odom->base_link transform. This is typically less accurate than a solution with fused data sources, but it may fit you needs.
2. ekf_localization_node or ukf_localization_node instance two: fuse the same data as in instance one, but also fuse discontinuous sources like GPS or position updates from landmark observations. The world_frame for this instance is the same value as your map_frame; this is typically map. This instance is really only necessary if you have discontinuous measurement sources.

To more directly answer your question, for navsat_transform_node, the frame_id of the /odometry/gps message that gets created is depedent on the frame_id of the odometry message that gets fed to it.