The generated pcd point cloud map cannot be read because the capacity is too large

[mimimidododo]

I recorded the driving data using lidar and saved it as a rosbag. Although I used Autoware to successfully generate a pcd point cloud map using the data recorded by rosbag, Autoware cannot read it because the pcd file is too large (the size of the pcd point cloud map is about 800M). (The progress bar for reading the pcd point cloud map is always 0) Can anyone tell me what to do in this situation?

I found in the sample data downloaded from the Internet that the rosbag data can be divided into several parts to generate multiple pcd point cloud maps with relatively small capacity, and then select all the generated maps to read together when reading, these The pcd point cloud map with relatively small capacity will be automatically merged to generate a large pcd point cloud map. But I don't know how this is done.

[angry-crab]

A quick solution is to downsample the map file.

sudo apt update
sudo apt install pcl-tools
pcl_voxel_grid intput.pcd output.pcd -leaf 0.1, 0.1, 0.1

Adjust the leaf parameter accordingly. Hope this would help.

[mimimidododo]

I have tried it. The pcd map can be displayed normally. thank you very much.
Allow me to ask one more question. Is it really impossible to generate multiple pcd maps separately like the example data, and then import all the generated pcd maps and then synthesize a large pcd map?

[angry-crab]

It sounds like you are trying to use Autoware as a SLAM package, which is actually beyond our intention. However, as long as you have pcd frames and their corresponding pose infomation, you would be able to generate a map. In your case, I'm not sure what is the purpose of generating separate maps and merging them all together in the end. At least Autoware.Universe does not try to do so. Back to your question, if we were given segmented maps and their poses in the global coordinate, all we need to do is just transform each map and put them all together.

[mimimidododo]

Thank you so much for anwser. Because I see maps separated like this in the sample data. So I'm just curious how the example data does it. Because it takes too much time to generate a large map with a rosbag with a capacity of about 10GB.