Till Beemelmanns

Thanks, alright. I will further improve this path conversion by taking only the last parts of the original path.

We tested both, this branch and this attempt https://github.com/ouster-lidar/ouster-ros/pull/302 to solve the problem with missing parts of the point cloud. We set the recv buffer size to the following values...

We could observe the same problem with four cameras connected to one computer. We can observe inconsistent start-ups. If the time between shutting down the camera drivers and restarting the...

1. Killing and relaunching just single camera seems to be stable. The problem seems to occur then launching multiple cameras at the same time. However, some weeks ago we had...

Thanks for the swift response! Yes, we will try to build the driver from source in our docker setup or try to copy `libcamera_driver.so` to the correct place. I will...

@Marius-N-R No we did not have problems running `driver_node.launch.py` using docker. However, we use a docker-compose setup with several cameras and we can define the `devices` there: ``` services: camera-driver-1:...

I trained PETR with the current code and with the suggestions in this PR. Both versions end up at the same metrics with only marginal differences: ``` NuScenes metric/pred_instances_3d_NuScenes/NDS: 0.4321...

Can you please provide more information about your setup. Otherwise, it is not possible to resolve this problem. Are you using the docker setup ? Which Tensorflow Version are you...

Can you double check that the dataset loaded correctly and placed at the correct location? And double check that all pathes are correct. The error usually arises when the dataloader...

Closed due to inactivity. Feel free to open issue again if you still have problems.