Hi, I’m trying to use the Jetson Orin-based computer with 6 Leopard Imaging Owl cameras. To receive video streams from cameras I use isaac_ros_argus_camera package. The test pipeline for each camera is very simple and consists of two composable nodes (argus_camera driver + rectification) combined with a composable node container (see the launch file attached). When I run this pipeline the CPU load is like 120% per camera (7 out 12 CPU cores are fully loaded), see the screenshot below.
Is it a normal CPU load for such a pipeline or I’m doing something wrong? I’m expecting that CPU load should be much lower because, as far as I understand, NITROS transports should be used by default for such pipelines.
I noticed that you are using the old Isaac ROS 2.1 setup. I recommend updating to Jetpack 6 and using Isaac ROS 3.0, as we have made major updates to NITROS and all Isaac ROS packages.
Regarding the performance with Hawk camera I report this note:
Argus Camera outputs sensor_msgs/Image at the sensor data rate, subject to performance capabilities of the Jetson platform(s). For example, a Hawk camera configured for 30fps (frames per second) stereo 1920x1080 will output time-synchronized left and right camera frames sensor_msgs/Image at 30fps.
We can’t switch to Jetpack 6 since we’re using an Orin-based system and corresponding BSP from ConnectTech based on Jetpack 5.1.2 (Rugue carrier + video capture board).
We tried to build an updated version of the argus_camera driver from Isaac 3.0 on the Jetpack 5.1.2-based system but the following error occurred:
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: nvidia-container-runtime did not terminate successfully: exit status 1: time=“2024-06-13T10:18:55+03:00” level=error msg="failed to create NVIDIA Container Runtime: failed to construct OCI spec modifier: requirements not met: unsatisfied condition: cuda>=12.2 (cuda=11.4)
Isaac ROS 3 is designed for Jetpack 6, and the error you are facing is related to your rebuild for Jetpack 5.1.2
I’m forwarding your issue to the engineering to find a way to support you.
