Hello, I am facing a problem regarding memory corruption. I have 2 VLP-16 pointcloud dataset (one is in front tilted with an angle of 45 degree and another one is at the centre verticle to the ground, data from both of them were merged and transformed with respect to base_link) along with imu dataset in a bag file of a rover which roves in an indoor environment for around 1500s. Special things to mention:
The IMU is producing data at a rate of 50 Hz.
The 2 merged lidar is producing data at a rate of 10 Hz.
There is no camera in the system.
Both the data of lidar and IMU are transformed with respect to baselink of the rover. So, I set the estimate_extrinsic parameter to 0 and opt_exrtinsic to 0 and gave the transform from IMU frame to lidar frame as identity rotation and translation.
Now here is what the data started to produce after launching the roslaunch lio test_indoor.launch and roslaunch lio map_4D_indoor.launch & the bag file.
After 23s, the odometry and the map are visible & the terminal shows 'imu initialized' (Before 23s, it was throwing an error saying 'imu excitation was not enough' which is reasonable as the rover started to move after 23s).
So when it started moving (after 23s), everything was going good. But as it continued to move on, it started to lag behind the realtime data. At about 173s, it threw an error malloc(): memory corruption error (on the terminal where test_indoor.launch file was launched) and the odometry and mapping topics stopped producing any data.
I reran the package and monitored the odometry and mapping topic publish rate from the beginning. The rate consistently decreases (for odometry topic, it started from 5 Hz but as time went, it decreased to 2.5 Hz,...,1.5 Hz... and so on and for the map topic, it started from 2 Hz but as with the time, it decreased to 1.0 Hz...,0.5 Hz,...,0.3 Hz and so on.)
I also monitored the RAM memory usage. It drastically increased along with swap memory and cpu processors work cycle.
Note: I have changed the frame_ids accordingly and after running the launch files, there is no error on the terminal regarding tf mismatch or something similar. Initially, it starts to produce great map but lags behind as it continued to explore the other rooms inside the campus with lagging and eventually die after sometime..
The exact error was:
malloc(): memory corruption
[lio_processor - 1] process has dies [pid 24315, exit code - 6 /home/arghya/costar_ws/devel/lib/lio/lio_processor_node __name:=lio_processor __log:=/homr/arghya/.ros/log/108f743e-0034-11eb-9c75-4485007e1d87/lio_processor-1.log]. log file: /home/arghya/.ros/log/108f743e-0034-11eb-9c75-4485007e1d87/lio_processor-1*.log
Hello, I am facing a problem regarding memory corruption. I have 2 VLP-16 pointcloud dataset (one is in front tilted with an angle of 45 degree and another one is at the centre verticle to the ground, data from both of them were merged and transformed with respect to base_link) along with imu dataset in a bag file of a rover which roves in an indoor environment for around 1500s. Special things to mention:
estimate_extrinsicparameter to0andopt_exrtinsicto0and gave the transform from IMU frame to lidar frame as identity rotation and translation.Now here is what the data started to produce after launching the
roslaunch lio test_indoor.launchandroslaunch lio map_4D_indoor.launch& the bag file.malloc(): memory corruption error(on the terminal where test_indoor.launch file was launched) and the odometry and mapping topics stopped producing any data.Note: I have changed the frame_ids accordingly and after running the launch files, there is no error on the terminal regarding tf mismatch or something similar. Initially, it starts to produce great map but lags behind as it continued to explore the other rooms inside the campus with lagging and eventually die after sometime..
The exact error was:
Is there any solution to this problem ?