wh200720041 / iscloam

Intensity Scan Context based full SLAM implementation for autonomous driving. ICRA 2020
Other
578 stars 134 forks source link

about memory #15

Open HeXu1 opened 4 years ago

HeXu1 commented 4 years ago

hi,thanks for your great work.

what's the memory consumed mainly? it seems my 16G memory almost be used,but the robot doesn't move so far.

hope for your reply,thanks

wh200720041 commented 4 years ago

Hi @HeXu1

The memory is mainly consumed by input laser cloud and mapping. The loop closure detection checks for all input frames instead of key frames. Hence all historical point clouds are stored into memory.

A new version is updated and I did some modification to reduce the point cloud size when stored into database.

If you are trying to use it for long term localization, maybe you should consider to store key frames only for loop closure detection

HeXu1 commented 4 years ago

Hi,I will test it. thanks!