Whether there is a bug in the frame_id of the perception module output objects

[cyn-liu]

Checklist

• [X] I've read the contribution guidelines.
• [X] I've searched other issues and no duplicate issues were found.
• [X] I'm convinced that this is not my fault but a bug.

Description

I set perception_mode=lidar and use lidar_centerpoint for object detection, but I find that my detection box does not show up on the lidar target. I debug and found that the frame_id of the output topic /perception/object_recognition/detection/centerpoint/objects is velodyne_top, I think this is unreasonable, it should be map or base_link, which is configured by param world_fram_id in the file centerpoint.param.yaml.

_I have correctly configured the extrinsic parameters from velodyne_top to base_link, and NDT localization is successful._

Expected behavior

The output topic /perception/object_recognition/objects of the perception module is drawn on the point cloud object.

Actual behavior

The output topic /perception/object_recognition/objects of the perception module is not drawn on the point cloud object.

_I don't know if frame_id=velodyne_top is a bug or not, but I can't find a reason why the detection bounding box is not on the lidar object._

/perception/object_recognition/detection/centerpoint/objects

/perception/object_recognition/detection/clustering/objects

/perception/object_recognition/detection/temporary_merged_objects

/perception/object_recognition/detection/objects_before_filter

/perception/object_recognition/detection/objects

/perception/object_recognition/objects

Steps to reproduce

1. modify input/pointcloud topic
2. build Autoware
3. launch Autoware

Versions

OS: amd ROS2: humle Autoware: The latest autoware.universe pulled from the autoware repository

Possible causes

No response

Additional context

No response

[cyn-liu]

lidar2ego external parameters and localization is correct:

The display result of topic /perception/object_recognition/objects in rviz2

[vividf]

@cyn-liu Thanks for your report! I could reproduce your result and will fix the bug soon.

[vividf]

@cyn-liu The issue can be solved with PR https://github.com/autowarefoundation/autoware.universe/pull/8674. Thanks for your report!