RTAB-Map Post-processing error with g2o

[Satish Akula]

Hi,

I created an RTAB-Map DB file with 3 intel real sense(Multi-camera) cameras by using vis/estimation type parameter as 0 as below

 

but while I'm trying to apply Sparse Bundle Adjustment(SBA) with g20 getting below error

[ERROR] (2020-04-02 17:05:42.677) Optimizer.cpp:494::optimizeBA() Missing calibration for node 1

[matlabbe]

Hi,

can you share that database? I don't have a multi cameras setup right now. One thing is to make sure all data is downloaded in the UI, with Edit->Download all clouds.

cheers,
Mathieu

[Satish Akula]

Hi Mathieu,

Sorry for the late response.
You can download the database file from the below link.
https://drive.google.com/open?id=1Hva2Swh3obysep9hfA_Jyo3LdrdbiO6m

[matlabbe]

Hi,

I updated the error message. It is because SBA for multicam is not implemented.

About the database, not sure what the scans mean. Also, the whole map is tilted, you may have to adjust the TF between the base of the robot and the cameras.

cheers,
Mathieu

[Satish Akula]

Hi Mathieu,

Many thanks for the quick reply.
Post-processing multi-camera database file with SBA is not possible, right?
what you mean by "not sure what the scans mean"?
its a handheld scanning only, captured the space by holding the camera setup in hand not by using the robot.


Thanks,
Satish

[matlabbe]

There are laser scans recorded in the database. They seem to be fake laser scans created from the depth images. Yes, SBA post processing won't work for multi-camera. If it was captured by hand, if the map is tilted it is because the cameras started tilted towards the ground.

[Satish Akula]

Hi Mathieu,

I'm not using any package like depthimage_to_laserscan to convert depth to laser then how depth image is generating fake laser scans.
Is those fake laser scans cause any problem to the output, how can I remove those?


Thanks,
Satish

[matlabbe]

If you used this example, set gen_scan to false or remove that parameter.

They were not used for localization, but the occupancy grid was created wrongly using them.

[Satish Akula]

Hi Mathieu,

Thanks for the reply,
Yes, I'm using the gen_scan parameter. I will remove it and check the output.

Once again thanks for the support.


Thanks,
Satish