Welcome to Jetson Orin NANO Developer Kit repository
01.Build Isaac ROS environment
Isaac ROS Environment Setup (Jetson Orin Nano)
Isaac ROS Environment Setup (Jetson Orin Nano)1. Scope of Application2. Obtaining and Checking FilesDisk Space Requirements3. Importing the Docker Image4. Extracting the Course Workspace5. Starting the Container5.1 Preparing Fast DDS and the Terminal Environment5.2 Creating the Startup Script6. Checking the ROS Environment7. Model Preparation Rules8. Optional: Deleting the Archives After Deployment Is Confirmed9. Usage Notes10. Next Steps
1. Scope of Application
This tutorial is intended for deploying the Isaac ROS environment and the case workspace provided by the course on a Jetson Orin Nano 8GB. Before deployment, confirm that the device matches the table below; for other Jetson models or JetPack versions, image compatibility must be confirmed first.
Note: The remaining space of the factory image system is insufficient to set up the Isaac ROS environment. Use the dedicated Isaac image provided in the materials (the environment is already configured), or use another third-party system image with sufficient space to set up the environment. For a third-party system image, refer to the following tutorials: 【Chapter 2 Mainboard Basics: 02. Writing the Jetson Clean System】 -> 【Chapter 2 Mainboard Basics: 03. Installing the Jetson Environment】 -> 【Chapter 8 Docker Basics: 02. Docker Installation】
| Item | Requirement |
|---|---|
| Mainboard | Jetson Orin Nano 8GB, aarch64 |
| JetPack / L4T | JetPack 7.2.1 / L4T R39.2.1 |
| System | Ubuntu 24.04 |
| Docker environment | Installed and runnable by the current user |
| NVIDIA Container Toolkit | Installed |
| Disk space | Must hold the imported image, the extracted workspace, and runtime headroom at the same time; verify with the formula in this section |
The environment consists of two parts: the Docker image provides ROS 2 Jazzy and the Isaac ROS software; the workspace archive provides the course models, data, and case files. The cases can only be run after both parts are deployed.
2. Obtaining and Checking Files
Disk Space Requirements
Before importing, the partition holding the Docker data directory and the partition holding the workspace target should satisfy the following respectively:
If both are on the same partition: the remaining space must be at least 65.3GB + the extracted workspace size + 20GB runtime headroom.
If the two are on different partitions: reserve at least 85GB on the partition holding the Docker data directory; reserve at least the extracted workspace size + 10GB on the partition holding the workspace
After confirming that the space is sufficient, place the following two files in the /home/jetson/ directory of the Jetson:
| File | Purpose |
|---|---|
my-isaac-ros-slam-4.6-all.tar.gz | Isaac ROS Docker image, the archive is about 24.3GB; the Docker image after import is about 65.3GB |
isaac_ros_workspaces.tar.gz | Case workspace, models, and data; the archive size and the extracted size vary with the delivered content, subject to the actual files and the delivery list |
If the files are on another computer, upload them from the terminal of that computer; replace <Jetson_IP> with the actual IP of the Jetson:
scp my-isaac-ros-slam-4.6-all.tar.gz jetson@<Jetson_IP>:/home/jetson/scp isaac_ros_workspaces.tar.gz jetson@<Jetson_IP>:/home/jetson/Then confirm the files, the system, and the disk space on the Jetson:
xxxxxxxxxxuname -mcat /etc/nv_tegra_releasedocker --versiondocker info --format 'DockerRootDir={{.DockerRootDir}} DefaultRuntime={{.DefaultRuntime}} Runtimes={{json .Runtimes}}'df -h /var/lib/docker /home/jetsonls -lh /home/jetson/my-isaac-ros-slam-4.6-all.tar.gz /home/jetson/isaac_ros_workspaces.tar.gzThe device should be aarch64, and the L4T version should be R39.2.1. The Docker runtime list should contain nvidia.
3. Importing the Docker Image
Execute on the Jetson host:
xxxxxxxxxxdocker load -i /home/jetson/my-isaac-ros-slam-4.6-all.tar.gzdocker image inspect my-isaac-ros-slam:4.6-all --format 'OS={{.Os}} Arch={{.Architecture}}'docker images | grep my-isaacAfter the import is complete, the image my-isaac-ros-slam:4.6-all should be visible, with the architecture arm64. Loading the image takes some time; do not close the terminal or delete the archive in the middle.
Expected output:
xxxxxxxxxxOS=linux Arch=arm64my-isaac-ros-slam:4.6-all xxxxxxx about 65.3GB
4. Extracting the Course Workspace
First view the directory structure in the archive:
xxxxxxxxxxtar -tzf /home/jetson/isaac_ros_workspaces.tar.gz | head -n 30The archive should contain the complete workspaces/isaac_ros-dev/ workspace, including at least:
isaac_ros_assets/: course models and rosbag datalaunch/: case launch filesrviz/: RViz configurationscripts/and the case helper scripts in the root of the workspacefastdds_no_shm.xml: the Fast DDS configuration used by the course
If the list is missing the directories or files above, obtain the complete workspace archive from the course provider first. Select the corresponding command according to the result of tar -tzf. If the paths in the archive start with workspaces/isaac_ros-dev/, extract to /home/jetson:
xxxxxxxxxxmkdir -p /home/jetson/workspacestar -xzf /home/jetson/isaac_ros_workspaces.tar.gz -C /home/jetsontest -d /home/jetson/workspaces/isaac_ros-dev/isaac_ros_assetstest -d /home/jetson/workspaces/isaac_ros-dev/launchtest -d /home/jetson/workspaces/isaac_ros-dev/rvizIf the paths in the archive start directly with isaac_ros-dev/, extract to /home/jetson/workspaces and check:
xxxxxxxxxxmkdir -p /home/jetson/workspacestar -xzf /home/jetson/isaac_ros_workspaces.tar.gz -C /home/jetson/workspacestest -d /home/jetson/workspaces/isaac_ros-dev/isaac_ros_assetstest -d /home/jetson/workspaces/isaac_ros-dev/launchtest -d /home/jetson/workspaces/isaac_ros-dev/rvizBefore extracting, confirm that the target directory does not exist or contains no files that need to be kept; if a check fails, verify the directory structure of the archive first and do not continue starting the container.
5. Starting the Container
5.1 Preparing Fast DDS and the Terminal Environment
The workspace is located in a host-mounted directory, so a configuration file placed in the workspace is retained across container restarts. First check whether the workspace already has fastdds_no_shm.xml; if not, create it on the Jetson host:
xxxxxxxxxxcat > /home/jetson/workspaces/isaac_ros-dev/fastdds_no_shm.xml <<'EOF'<?xml version="1.0" encoding="UTF-8" ?><profiles xmlns="http://www.eprosima.com/XMLSchemas/fastRTPS_Profiles"> <transport_descriptors> <transport_descriptor> <transport_id>udp_only</transport_id> <type>UDPv4</type> </transport_descriptor> </transport_descriptors> <participant profile_name="udp_only_participant" is_default_profile="true"> <rtps> <userTransports> <transport_id>udp_only</transport_id> </userTransports> <useBuiltinTransports>false</useBuiltinTransports> </rtps> </participant></profiles>EOFThe Fast DDS configuration uses the UDP transport and is the configuration of the container environment of this course. The startup script generates /tmp/.bashrc each time the container starts, which only contains the ROS and workspace environment settings; there is no need to create another environment file on the host, and the user configuration in the image is not modified.
5.2 Creating the Startup Script
On first use, create the startup script on the Jetson host:
xcat > /home/jetson/run_isaac.sh <<'EOF'set -e
ISAAC_ROS_WS="$HOME/workspaces/isaac_ros-dev"
if [[ ! -d "$ISAAC_ROS_WS" ]]; then echo "The workspace does not exist: $ISAAC_ROS_WS" >&2 exit 1fiif [[ ! -f "$ISAAC_ROS_WS/fastdds_no_shm.xml" ]]; then echo "The Fast DDS configuration file does not exist: $ISAAC_ROS_WS/fastdds_no_shm.xml" >&2 exit 1fi
docker run --rm -it \ --runtime=nvidia --gpus all \ --network host --ipc host --privileged \ -u 1000:1000 -e HOME=/tmp \ -e DISPLAY="$DISPLAY" -e QT_X11_NO_MITSHM=1 \ -v /tmp/.X11-unix:/tmp/.X11-unix:rw \ -v "$ISAAC_ROS_WS:/workspaces/isaac_ros-dev" \ --entrypoint /bin/bash \ my-isaac-ros-slam:4.6-all \ -c 'cat > /tmp/.bashrc <<'"'"'BASHRC_EOF'"'"'source /opt/ros/jazzy/setup.bashexport ISAAC_ROS_WS=/workspaces/isaac_ros-devexport FASTRTPS_DEFAULT_PROFILES_FILE=/workspaces/isaac_ros-dev/fastdds_no_shm.xmlBASHRC_EOFexec /bin/bash --rcfile /tmp/.bashrc -i'EOFxxxxxxxxxxchmod +x /home/jetson/run_isaac.shThe script is written for the jetson user with UID/GID 1000:1000; other users should run id first and then modify the -u parameter according to the actual result. When the container starts, the ROS and workspace environment is written to /tmp/.bashrc, and an interactive Bash is started with that file through --rcfile. --privileged is part of the current device configuration of the course and is not recommended to be applied directly to containers for other purposes.
Graphical cases must be started from the Jetson desktop terminal. First allow the local Docker container to access the current desktop, then run the startup script:
xxxxxxxxxxxhost +local:dockercd /home/jetson./run_isaac.shAfter all graphical programs are closed, the authorization can be revoked:
xxxxxxxxxxxhost -local:dockerDo not use xhost + to disable all graphical access control, and do not modify the global Docker runtime by yourself. If an error is reported at startup, record the complete error message and check the image, the workspace, and the desktop session configuration.
After the container starts, open another Jetson terminal to view the container name:
xxxxxxxxxxdocker ps --format 'table {{.Names}}\t{{.Image}}\t{{.Status}}'When an additional container terminal is needed, execute on the Jetson host:
xxxxxxxxxxdocker exec -it <container name> bashReplace <container name> with the name shown by docker ps.
6. Checking the ROS Environment
When entering the container with the startup script of this tutorial, the ROS and Fast DDS environment is loaded automatically by /tmp/.bashrc. Execute the following commands inside the container to check:
xxxxxxxxxxecho "$ROS_DISTRO"echo "$ISAAC_ROS_WS"echo "$FASTRTPS_DEFAULT_PROFILES_FILE"test -f "$FASTRTPS_DEFAULT_PROFILES_FILE"test -d "$ISAAC_ROS_WS/isaac_ros_assets"test -d "$ISAAC_ROS_WS/launch"The expected ROS distribution is jazzy, the workspace is /workspaces/isaac_ros-dev, and the Fast DDS configuration file exists. Then check the core package:
xxxxxxxxxxros2 pkg prefix isaac_ros_visual_slamThe expected package path is located in /opt/ros/jazzy. If the container is started in another way, source /opt/ros/jazzy/setup.bash must be executed manually in each new terminal, and ISAAC_ROS_WS and FASTRTPS_DEFAULT_PROFILES_FILE must be set.
7. Model Preparation Rules
Case 5 uses the TensorRT plan in the optimized_deployable_shuffleseg_unet_amr_v1.0 directory, which is loaded directly by the launch file; case 7 uses the Triton model peoplenet. Cases 1–4, 6, 8, and 9 do not need to be switched in this way. Before switching, stop the current case first, check the two directories inside the container, and then execute as needed:
xxxxxxxxxxcd /workspaces/isaac_ros-dev/isaac_ros_assetsls models/triton/ models_pending/
# Move all current models under models/triton/ back to the pending directorymv models/triton/* models_pending/
# Replace <model name> with the corresponding directory name abovemv models_pending/<model name> models/triton/ls models/triton/mv models/triton/* moves all models in the active directory away, so execute it only after confirming that no other model needs to be retained; if the target model is already in the active directory, no switch is required. Before moving, confirm that there is no directory with the same name in pending; if the command reports an error, check the directories first and do not hide the error with 2>/dev/null. The CenterPose FP32 plan of case 8 is specified by the startup script and is not switched here.
8. Optional: Deleting the Archives After Deployment Is Confirmed
After confirming that the image has been imported successfully, the workspace has been extracted, and the container and cases can start normally, the two archives can be deleted if the installation packages do not need to be retained. This operation does not delete the imported Docker image or the extracted workspace files; keep the archives while the deployment has not been confirmed successful.
xxxxxxxxxxrm /home/jetson/my-isaac-ros-slam-4.6-all.tar.gzrm /home/jetson/isaac_ros_workspaces.tar.gz9. Usage Notes
Run only one high-load case at a time; start the next one after the current case is finished.
Start and stop the nodes of each case according to its operation instructions, and exit normally with
Ctrl+Cin the corresponding terminal; do not terminate processes in batches withclean_rosor a broadpkill.After the container exits, the container itself is deleted automatically; the Docker image and the files in
/home/jetson/workspaces/isaac_ros-dev/on the host are retained.If
groups: cannot find name for group IDis displayed but the container, the ROS nodes, and the case checks are normal, this prompt usually does not affect operation.If RViz or rqt cannot display, confirm that it was started from the Jetson desktop terminal and check whether the desktop graphical authorization has been set.
If Docker, image import, or workspace checks fail, record the complete error message and contact the course provider; do not delete the delivery archives or modify the system-level Docker configuration.
10. Next Steps
After the environment check passes, run the course content by referring to the operation instructions of cases 1–9 in the directory of this chapter.