Frame delayed at time

[GHz12138]

[joshi-bharat]

I will have a look at it tomorrow

[GHz12138]

I have two questions which I hope you can answer. 1.Svin stereorig v1.launch file whether to change stereorig /stereo rig config.yaml to stereorig v1 config.yaml 2.Why do dropping frames and frame delays occur？ | | ***@***.*** | | ***@***.*** |

…

---- Replied Message ---- | From | Bharat ***@***.***> | | Date | 08/14/2023 20:40 | | To | ***@***.***> | | Cc | ***@***.***>***@***.***> | | Subject | Re: [AutonomousFieldRoboticsLab/SVIn] Frame delayed at time (Issue #18) | I will have a look at it tomorrow — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[Unigrass]

I'm having the same issue when running on the bus and cemetery dataset. "Dropping frame" and "frame delay" error appeared. In terms of the cemetery dataset, the trajectory can run in a whole loop without long time stoppage, but the trajectory plot froze when trying to run the second loop in rviz. In terms of the bus dataset, it just ceased and stopping working.

[Unigrass]

I have several questions which I hope u can answer.
When I tried to run the algorithm on the cave and bus dataset, the following error appeared in rviz.

It seems that information about the sonar sensor could not be successfully received. I tried for a few days but still could not debug it. When I tick or untick the "Sonar" icon, does it enable and disable the sonar sensor in SVin2, or does it just display the sonar features on the trajectory plot without the function of turning it on and off? Also, I would like to test the sonar performance only. Do I just need to tick the "Sonar" icon in rviz and untick everything else, or do I have to go to the code and extract the sonar part?

[Unigrass]

Also, my team could not reproduce the result of this algorithm. The trajectory plot just froze and stopped working halfway. I think the reason is that there's tracking loss caused by lack of features, and the algorithm doesn't have a re-initialization process (didn't see that in the paper). How could I solve that? If I need to adjust some parameters to reproduce the result, could you provide me the parameters used?

[joshi-bharat]

Sorry that you guys are having issues. I have some knowledge of these issues. I will add documentation on how to fix these. Give me a day or two.

[Unigrass]

Sorry that you guys are having issues. I have some knowledge of these issues. I will add documentation on how to fix these. Give me a day or two.

I'd like to ask when you will be able to post the solution, as our project is currently stuck in this section. Thanks for taking time to fix it.

[xuduoudo0505]

i meet the same problem,Have you solved the problem?

[GHz12138]

Hello, thank you for your wonderful sharing. Recently, I extracted the original picture information from the rosbag format files of the cave data set and cemetery data set you shared before, and found that the occurrence frequency of the left and right images was unstable, resulting in a large difference in time stamps, and there was a pause in the middle, resulting in a different number of pictures from the left and right cameras. In addition, the original rosbag information had duplicate messages. For example, there are 13794 messages retrieved by rosabag info in the cave data set, but there are actually 60 duplicate picture messages, and only 13734 pictures extracted. These problems may be the cause of frame loss and delay in the algorithm. Do you have any solutions?

/slave2/image_raw/cam_info 13794 msgs : sensor_msgs/CameraInfo
/slave2/image_raw/compressed 13794 msgs : sensor_msgs/CompressedImage