Open mattwilliamson opened 5 months ago
I am running same on RPI4, and I measured the cpu usage to be 10-12%. I also compared it with a previous driver, and found that it is about the same. So this must be a problem with your jetson or the docker image way of passing the serial data.
On Nvidia Orin Nano, ldlidar node uses 100% of one core. This seems high for me since it's just parsing out serial input and publishing. Comparatively, nav2 uses less. Is there any way to reduce it? Lower publish rate or something? Maybe we should try to use static memory or maybe some external library like CUDA or something to speed it up?
I'm just using the stock launch file and parameters.