ros_deep_learning package not found
MiguelOsun opened this issue · 13 comments
Hi, thank you for all your work.
I am using a Jetson Nano with L4T R32.7.1. so when I run your docker/run.sh -ros=foxy. The problem is that, when I tried to test with roslaunch to launch video_viewer, it said there was no "ros_deep_learning" package. I followed every step in the repo, so I dont really know if it is because I am doing something wrong or it is a problem related to the container itself.
Check my issue : https://github.com/dusty-nv/ros_deep_learning/issues/115.
I used same Nano with same R32.7.1 like you. The container for that version has not been updated yet so there is no pre-installed ros deep learning package. My suggestion is to build from source like I did with the melodic branch version because I think you also used Ubuntu 18.04 for your Nano.
Hi! As you said, I am using Ubuntu 18.04 too. I will try building from source with Melodic. Thank you for your help.
@MiguelOsun I have updated the container images for L4T R32.7.1 to include ros_deep_learning package, see #115 (comment)
Hi! Thank you for your help! I am new to Docker Hub, and I have started recently to work with containers. Does this repo use this updated container images, or do i have to update them?
If you had already tried running the container, you should pull it again to fetch the updated one - for example: sudo docker pull dustynv/ros:noetic-pytorch-l4t-r32.7.1
Hi,
as you said, i pulled the image again and i dont have that problem anymore, but when i try to launch any node of the ros_deep_learning package i have this issue:
[video_output-2] /ros_deep_learning/install/ros_deep_learning/lib/ros_deep_learning/video_output: error while loading shared libraries: libnppicc.so.10: cannot open shared object file: No such file or directory
[video_source-1] /ros_deep_learning/install/ros_deep_learning/lib/ros_deep_learning/video_source: error while loading shared libraries: libnppicc.so.10: cannot open shared object file: No such file or directory
Is this because i have another version of CUDA installed?
Hello, I have the same problem with a Xavier AGX. I have a fresh installation with Jetpack 5.1.1 which includes CUDA 11 and have the same error.
[video_output-2] /ros_deep_learning/install/ros_deep_learning/lib/ros_deep_learning/video_output: error while loading shared libraries: libnppicc.so.10: cannot open shared object file: No such file or directory
[video_source-1] /ros_deep_learning/install/ros_deep_learning/lib/ros_deep_learning/video_source: error while loading shared libraries: libnppicc.so.10: cannot open shared object file: No such file or directory
But if I want to downgrade to CUDA 10 I have to install Jetpack 4.6 which is only for Ubuntu 18.
It is possible to modify any instruction to try to use the CUDA 11 drivers?
Thank you in advance.
Jorge
Hi @grafoteka, which version of the container are you running?
Hi @grafoteka, which version of the container are you running?
Hi @dusty-nv, I'm using the last version that you update a few days ago. Galactic-pytorch-l4t-r32.7.1
https://hub.docker.com/layers/dustynv/ros/galactic-pytorch-l4t-r32.7.1/images/sha256-1fe64222761f7bbc438592d0b22bfa09ec51c92f7f160e71f5bb6ae1cb46369e?context=explore
Hi, as you said, i pulled the image again and i dont have that problem anymore, but when i try to launch any node of the ros_deep_learning package i have this issue:
[video_output-2] /ros_deep_learning/install/ros_deep_learning/lib/ros_deep_learning/video_output: error while loading shared libraries: libnppicc.so.10: cannot open shared object file: No such file or directory [video_source-1] /ros_deep_learning/install/ros_deep_learning/lib/ros_deep_learning/video_source: error while loading shared libraries: libnppicc.so.10: cannot open shared object file: No such file or directory
Is this because i have another version of CUDA installed?
@MiguelOsun Which CUDA version are you using? Im using CUDA 10.2
hi,
im trying to use dustynv/ros:foxy-ros-base-l4t-r32.5.0 container and it not include ros_deep_learning package.
can you please update the container?
im trying to use dustynv/ros:foxy-ros-base-l4t-r32.5.0 container and it not include ros_deep_learning package.
@saarCogni only the ros:*-pytorch-* images have ros_deep_learning package in them. dustynv/ros:foxy-pytorch-l4t-r32.7.1 has been updated to have ros_deep_learning in it, but I don't have systems still running older versions of JetPack 4. So you would either need to rebuild the ros:foxy-pytorch container for L4T R32.5.0, or install ros_deep_learning into it yourself (which you can do by following this script: https://github.com/dusty-nv/jetson-containers/blob/master/scripts/ros_deep_learning.sh), or upgrade your JetPack to L4T R32.7.
@dusty-nv
hi,
when im running the docker sudo docker run --runtime nvidia -it --rm --network host dustynv/ros:foxy-pytorch-l4t-r32.5.0 bash
on NVIDIA Jetson Xavier with the following script: ros_deep_learning.sh
this is what i get:
`root@algokobuki-desktop:/# ./ros_deep_learning.sh
ROS_DISTRO = foxy
sourcing /opt/ros/foxy/install/setup.bash
ROS_ROOT /opt/ros/foxy
ROS_DISTRO foxy
Cloning into 'ros_deep_learning'...
remote: Enumerating objects: 39, done.
remote: Counting objects: 100% (39/39), done.
remote: Compressing objects: 100% (29/29), done.
remote: Total 39 (delta 12), reused 18 (delta 7), pack-reused 0
Unpacking objects: 100% (39/39), done.
Building ros_deep_learning package for ROS2 foxy
Starting >>> ros_deep_learning
[Processing: ros_deep_learning]
--- stderr: ros_deep_learning
detected ROS_DISTRO=ROS_FOXY
CMake Warning at CMakeLists.txt:19 (find_package):
By not providing "FindVPI.cmake" in CMAKE_MODULE_PATH this project has
asked CMake to find a package configuration file provided by "VPI", but
CMake did not find one.
Could not find a package configuration file provided by "VPI" (requested
version 2.0) with any of the following names:
VPIConfig.cmake
vpi-config.cmake
Add the installation prefix of "VPI" to CMAKE_PREFIX_PATH or set "VPI_DIR"
to a directory containing one of the above files. If "VPI" provides a
separate development package or SDK, be sure it has been installed.
detected ROS2 (ament_cmake)
CMake Warning:
Manually-specified variables were not used by the project:
CATKIN_INSTALL_INTO_PREFIX_ROOT
make[2]: *** No rule to make target '/usr/local/cuda/lib64/libnppicc.so', needed by 'video_output'. Stop.
make[2]: *** Waiting for unfinished jobs....
make[1]: *** [CMakeFiles/video_output.dir/all] Error 2
make: *** [all] Error 2
Failed <<< ros_deep_learning [38.9s, exited with code 2]
Summary: 0 packages finished [39.5s]
1 package failed: ros_deep_learning
1 package had stderr output: ros_deep_learning`
thanks for the help