lidarslam_ros2: scanmatcher_node-1 process has died
Hi all.
Currently I’m using ROS2 humble, and using LS Lidar C32. I have recorded a pointcloud in a ros bag and try to play it back in our office.
First I run lidarslam. It run as normal
ros2 launch lidarslam lidarslam.launch.py
Next, I run RVIZ2 to visualise the path and modified_path. Also, have no issue.
After running lidarslam and rviz2, I play the ros2 bag. The way I play the ros bag is as following:
ros2 bag play file.bag
After few seconds, the vehicle is moving and now we can see the path.
Suddenly, there is an error occured in lidarslam. The error is as following:
[scanmatcher_node-1] [pcl::VoxelGrid::applyFilter] Leaf size is too small for the input dataset. Integer indices would overflow.
[ERROR] [scanmatcher_node-1]: process has died [pid 121510, exit code -8, cmd '/home/reka/ros2_ws/install/scanmatcher/lib/scanmatcher/scanmatcher_node --ros-args --params-file /home/reka/ros2_ws/install/lidarslam/share/lidarslam/param/lidarslam.yaml -r /input_cloud:=/c32/lslidar_point_cloud'].
Now, the lidarslam is crash. and the path is not moving anymore. Can I know what the reason behind this and how to fix it?
I tried to fix it by increasing the leaf size in params.yaml, but seems not working.
About this issue
- Original URL
- State: open
- Created 3 years ago
- Reactions: 1
- Comments: 15 (7 by maintainers)
The parameter is documented here: https://github.com/rsasaki0109/lidarslam_ros2/blob/develop/README.md?plain=1#L85 You can set the value in the ROS parameter file that is configuring the scan matching node, as in here: https://github.com/rsasaki0109/lidarslam_ros2/blob/develop/lidarslam/param/lidarslam.yaml#L11
I had the same issue for my custom recordings with an OS1-64. I just needed to adjust the parameters vg_size_for_input and vg_size_for_map. In my case, increasing vg_size_for_map to 0.5 solved the crash.
I am facing the same problem, whenever I play rosbag2 play bag file command the bag is getting closed and rviz is not showing any map
I apologize for the inconvenience, but due to recent busyness and the complexity of the issue, I would appreciate some time to address it.
I apologize, it slipped my mind. At a glance, it seems that the voxel grid filter inside updateMap, which is running in parallel, might be the issue. https://github.com/rsasaki0109/lidarslam_ros2/blob/humble/scanmatcher/src/scanmatcher_component.cpp#L422
Due to time constraints, I’ll check again on a weekday evening or weekend.
@rsasaki0109 Here is my YAML file.