report MemoryError when playback bagfile

[yebin999]

Hi,
Thanks for your excellent work.

But it didn't seem play back the bag file such "ALoop_fast_disturbed_surround.bag.baiduyun.p.downloading" successfully in my machine.
below is my env and command:
ros version: melodic
command: rosbag info ALoop_fast_disturbed_surround.bag.baiduyun.p.downloading

and then it reported MemoryError.

did you meet this issue or could you help give ideas for this issue?

Thanks!

[jarvishou829]

Hi, Thanks for your excellent work.

But it didn't seem play back the bag file such "ALoop_fast_disturbed_surround.bag.baiduyun.p.downloading" successfully in my machine. below is my env and command: ros version: melodic command: rosbag info ALoop_fast_disturbed_surround.bag.baiduyun.p.downloading

and then it reported MemoryError.

did you meet this issue or could you help give ideas for this issue?

Thanks!

It seems that the download has not been finished. The bag file should not end with '.baiduyun.p.downloading'. Once finished, the file name should be 'ALoop_fast_disturbed_surround.bag' exactly.

[yebin999]

Hi, Thanks for your excellent work.
But it didn't seem play back the bag file such "ALoop_fast_disturbed_surround.bag.baiduyun.p.downloading" successfully in my machine. below is my env and command: ros version: melodic command: rosbag info ALoop_fast_disturbed_surround.bag.baiduyun.p.downloading
and then it reported MemoryError.
did you meet this issue or could you help give ideas for this issue?
Thanks!

It seems that the download has not been finished. The bag file should not end with '.baiduyun.p.downloading'. Once finished, the file name should be 'ALoop_fast_disturbed_surround.bag' exactly.

oh, it's my mistake, thanks a lot again!

[yebin999]

hi
i met another issue is the fisheye data in the record bag didn't seem to be subscribed by rviz successfully. the msg type of camera data is sensor_msg/CompressedImage, but the rviz seems receive camera data using raw(i guess may be sensor_msg/Image).
did you have any related work about visualisation in rviz to give me some idea?

Thanks!

[jarvishou829]

hi i met another issue is the fisheye data in the record bag didn't seem to be subscribed by rviz successfully. the msg type of camera data is sensor_msg/CompressedImage, but the rviz seems receive camera data using raw(i guess may be sensor_msg/Image). did you have any related work about visualisation in rviz to give me some idea?

Thanks!

Sorry for the late reply. A simple solution is to write a ROS node to listen to the CompressedImage topic and publish the Image topic. If you just want to visualize the images, you can read out and save the images.

[yebin999]

hi i met another issue is the fisheye data in the record bag didn't seem to be subscribed by rviz successfully. the msg type of camera data is sensor_msg/CompressedImage, but the rviz seems receive camera data using raw(i guess may be sensor_msg/Image). did you have any related work about visualisation in rviz to give me some idea?
Thanks!

Sorry for the late reply. A simple solution is to write a ROS node to listen to the CompressedImage topic and publish the Image topic. If you just want to visualize the images, you can read out and save the images.

thanks for your help!

[yebin999]

Hi,
sorry for disturbing you again, would you mind sending me a rough version code about IPM if you still have?

many thanks!

[yebin999]

Hi, sorry for disturbing you again, would you mind sending me a rough version code about IPM if you still have?

many thanks!

my email is [email protected]

[jarvishou829]

Hi, sorry for disturbing you again, would you mind sending me a rough version code about IPM if you still have?

many thanks!

Sorry that no existing IPM code is left. However, it can be realized by projecting images according to the extrinsic. Just define an empty image ground plane, project the surrounding images separately on the plane, calculate the exact region of each surrounding image on the ground plane accordingly, and clip and mix them. Maybe you can get help from a chat AI.

[yebin999]

Hi, sorry for disturbing you again, would you mind sending me a rough version code about IPM if you still have?
many thanks!

Sorry that no existing IPM code is left. However, it can be realized by projecting images according to the extrinsic. Just define an empty image ground plane, project the surrounding images separately on the plane, calculate the exact region of each surrounding image on the ground plane accordingly, and clip and mix them. Maybe you can get help from a chat AI.

thanks for your reply in short time, yes, this way is possible if the distortion parameter of cameras in simulate enviroment are clear, but i am not sure whether the params i got from the sample in herehttps://www.svlsimulator.com/docs/simulation-content/sensors-list/#color-camera are right
"DistortionParameters": [ -0.25349, 0.11868, 0, 0 ],
could you help me double check them? i am not sure whether you still remember them, hhha...
anyway, many thanks!

[yebin999]

https://www.svlsimulator.com/docs/simulation-content/sensors-list/#color-camera

sorry, that link seems can't click directly in here, maybe need copy to browser and open

[jarvishou829]

Hi, sorry for disturbing you again, would you mind sending me a rough version code about IPM if you still have?
many thanks!

Sorry that no existing IPM code is left. However, it can be realized by projecting images according to the extrinsic. Just define an empty image ground plane, project the surrounding images separately on the plane, calculate the exact region of each surrounding image on the ground plane accordingly, and clip and mix them. Maybe you can get help from a chat AI.

thanks for your reply in short time, yes, this way is possible if the distortion parameter of cameras in simulate enviroment are clear, but i am not sure whether the params i got from the sample in herehttps://www.svlsimulator.com/docs/simulation-content/sensors-list/#color-camera are right "DistortionParameters": [ -0.25349, 0.11868, 0, 0 ], could you help me double check them? i am not sure whether you still remember them, hhha... anyway, many thanks!

All parameters can be found in folder /SUPS/bag_file/config in the link.

[yebin999]

Hi, sorry for disturbing you again, would you mind sending me a rough version code about IPM if you still have?
many thanks!

Sorry that no existing IPM code is left. However, it can be realized by projecting images according to the extrinsic. Just define an empty image ground plane, project the surrounding images separately on the plane, calculate the exact region of each surrounding image on the ground plane accordingly, and clip and mix them. Maybe you can get help from a chat AI.

thanks for your reply in short time, yes, this way is possible if the distortion parameter of cameras in simulate enviroment are clear, but i am not sure whether the params i got from the sample in herehttps://www.svlsimulator.com/docs/simulation-content/sensors-list/#color-camera are right "DistortionParameters": [ -0.25349, 0.11868, 0, 0 ], could you help me double check them? i am not sure whether you still remember them, hhha... anyway, many thanks!

All parameters can be found in folder /SUPS/bag_file/config in the link.

I checked those config files in SUPS/bag_file/config. In the calibration_Surround.json shows the camera type is fisheye and Distorted is true but the DistortionParameters list is empty, and i didn't find the DistortionParameters list has value in other files as well, so the params seem be still unclear...

[leijobs]

@jarvishou829 Hello, can you provide the homography matrix from fishseye image to BEV image? Thanks, I can't found out the right one, or can you provide the surround view image correspond to BEV image named with 1647xx.xx timestmap?

[xu-cpu]

Hello author, may I ask if the true value trajectory of the dataset can be provided? For example, when comparing algorithms such as lio-sam in the paper, it seems that the reference trajectories used are not in the folder. Can you share these reference trajectories? [email protected], thank you very much indeed.