hku-mars / LTAOM

Apache License 2.0
295 stars 34 forks source link

Runtime Error `[sample_nodelet_manager-2] process has died` #8

Open satyajitghana opened 2 months ago

satyajitghana commented 2 months ago

I am trying to run on my bag files,

bash run_nodelet_avia.sh
bash run_loop_optimization.sh

For some reason the fastlio node dies after a few seconds, what could be the reason?

Key Frame:0, cloud size:78
[ mapping ]: time: IMU + Map + Input Downsample: 0.000871 ave match: 0.001252 ave solve: 0.000230  ave ICP: 0.000346  map incre: 0.000062 ave total: 0.002948 icp: 0.001492 construct H: 0.000087 
[sample_nodelet_manager-2] process has died [pid 2336101, exit code -11, cmd /opt/ros/noetic/lib/nodelet/nodelet manager __name:=sample_nodelet_manager __log:=/home/satyajit/.ros/log/81ca883e-f566-11ee-98c5-7907c0d462a3/sample_nodelet_manager-2.log].
log file: /home/satyajit/.ros/log/81ca883e-f566-11ee-98c5-7907c0d462a3/sample_nodelet_manager-2*.log
[ INFO] [1712553145.779359595]: Bond broken, exiting
[ INFO] [1712553145.875414030]: Bond broken, exiting
[loop_detection_plugin_loader-4] process has finished cleanly
log file: /home/satyajit/.ros/log/81ca883e-f566-11ee-98c5-7907c0d462a3/loop_detection_plugin_loader-4*.log
[fastlio_plugin_loader-3] process has finished cleanly
log file: /home/satyajit/.ros/log/81ca883e-f566-11ee-98c5-7907c0d462a3/fastlio_plugin_loader-3*.log
^C[rviz-5] killing on exit
[rosout-1] killing on exit

Is there something wrong in the modified FASTLIO in LTAOM?

zuhaozou2017 commented 2 months ago

It seems nothing wrong with FASTLIO, but loop detection. You can try disabling loop detection in launch file.

First line: "Key Frame:0, cloud size:78". I guess the problem is that cloud points are too few for loop detection triggering runtime error.

satyajitghana commented 2 months ago

I am using a MID360

satyajitghana commented 2 months ago

in the launch config file point filter num was 5, and downsample was 0.3, I've changed it to point filter 1, and downsample 0.05, but still the cloud_size is 60-70 points only.

zMkOrg commented 3 weeks ago

I am using a MID360

When i use Mid360 to output /livox/lidar topic, the msgs's format are by default livox_ros_driver2/CustomMsg,I got the same problem as yours, When i change MID 360 /livox/lidar topc msg type to sensor_msgs/PointCloud2,the problem is sovled (I guess)

zMkOrg commented 2 weeks ago

in the launch config file point filter num was 5, and downsample was 0.3, I've changed it to point filter 1, and downsample 0.05, but still the cloud_size is 60-70 points only.

Also , you can use mid360 with livox_ros_driver2/CustomMsg format, just need to make some changes in preprocess.cpp at line 147 and line 160 as hinted