The official NASA Lunabotics GitHub repository for University of Minnesota Robotics.
graph LR
subgraph A[Operator Laptop]
B[RQT Camera Frontend]
J[joy_node]
end
subgraph C[Robot]
subgraph F[Nvidia Jetson AGX Orin]
G[motor_control]
H[GStreamer NVENC AV1 Encoding]
I[isaac_ros_nvblox]
L[ros2socketcan_bridge]
M[Nav2]
N[Subsystem ROS 2 Nodes]
O[rovr_control]
P[ZED ROS 2 Wrapper]
end
D[Arduino Microcontroller]
K[Limit Switches and Absolute Encoders]
E[VESC Motor Controllers]
end
K --> D
O <-- Serial Bus --> D
H -- WiFi Connection --> B
L <-- CAN Bus --> E
P --> I
I -- Cost Map --> M
M --> O
J -- /joy --> O
M -- /cmd_vel --> N
G -- /CAN/can0/transmit --> L
L -- /CAN/can0/receive --> G
O -- /cmd_vel --> N
O -- ROS 2 Services --> N
N -- ROS 2 Services --> G
How to Run Inside the Dev Container On Windows/Mac
Open vscode and install the "Dev Containers" extension. Then, with vscode open, press ctrl+shift+p to open the vscode command palette and type "Clone Repository in Container Volume". Select the "Dev Containers: Clone Repository in Container Volume" option, then select "Clone a repository from GitHub in a Container Volume". Search for and select our Lunabotics repository (the repository named "Lunabotics"). If you are cloning the repository directly into the container volume, you do NOT need to clone the repo locally, it will be automatically cloned into the repo.
After opening the container, you can run the following command in the Command Palette (Ctrl + Shift + P) to build the project:
Tasks: Run Build Task
Or, if your machine does not have an Nvidia GPU or you haven't set it up with container-toolkit, run the following commands instead in the Command Palette (Ctrl + Shift + P):
Tasks: Configure Default Build Task
Build No GPU Tasks
Optionally, traditional "colcon build" commands can be run in the vscode terminal instead of using the Command Palette commands above.
Updating Dev Container For Windows/Mac
If you ever need to rebuild the remote container image, first update the x86_64 and aarch64 images:
cd ~/Lunabotics/src/isaac_ros/isaac_ros_common/docker
docker build --build-arg="BASE_IMAGE=nvcr.io/nvidia/isaac/ros:x86_64-ros2_humble_bcf535ea3b9d16a854aaeb1701ab5a86" -f Dockerfile.user -t umnrobotics/isaac_ros:x86_64.ros2_humble.user .
docker build --build-arg="BASE_IMAGE=umnrobotics/isaac_ros:x86_64.ros2_humble.user" -f Dockerfile.umn -t umnrobotics/isaac_ros:x86_64.ros2_humble.user.umn .
docker push umnrobotics/isaac_ros:x86_64.ros2_humble.user.umn
docker build --build-arg="BASE_IMAGE=nvcr.io/nvidia/isaac/ros:aarch64-ros2_humble_b7e1ed6c02a6fa3c1c7392479291c035" -f Dockerfile.user -t umnrobotics/isaac_ros:arm64.ros2_humble.user --platform "arm64" .
docker build --build-arg="BASE_IMAGE=umnrobotics/isaac_ros:arm64.ros2_humble.user" -f Dockerfile.umn -t umnrobotics/isaac_ros:arm64.ros2_humble.user.umn --platform "arm64" .
docker push umnrobotics/isaac_ros:arm64.ros2_humble.user.umn
Then, run the following command with the devcontainer cli installed:
docker pull umnrobotics/isaac_ros:x86_64.ros2_humble.user.umn
docker pull umnrobotics/isaac_ros:arm64.ros2_humble.user.umn --platform "arm64"
docker manifest rm umnrobotics/isaac_ros:latest
docker manifest create umnrobotics/isaac_ros:latest --amend umnrobotics/isaac_ros:arm64.ros2_humble.user.umn --amend umnrobotics/isaac_ros:x86_64.ros2_humble.user.umn
docker manifest push umnrobotics/isaac_ros:latest
docker buildx create --use
devcontainer build --push true --workspace-folder . --platform="linux/amd64,linux/arm64" --image-name "umnrobotics/ros:isaac_ros_devcontainer"
How to Run Inside ISAAC ROS Container/Dev Container On Linux/Jetson
First, do the following before running run_dev.sh:
printf "CONFIG_IMAGE_KEY=ros2_humble.realsense.deepstream.user.zed.umn \n" > ~/Lunabotics/src/isaac_ros/isaac_ros_common/scripts/.isaac_ros_common-config
To use Gazebo in the ISAAC ROS container, do this instead:
printf "CONFIG_IMAGE_KEY=ros2_humble.realsense.deepstream.user.zed.umn.gazebo \n" > ~/Lunabotics/src/isaac_ros/isaac_ros_common/scripts/.isaac_ros_common-config
Regular Container
To make it so zed modules won't rerun every time you start the container, do this:
echo "-v /usr/local/zed/resources:/usr/local/zed/resources -v /ssd:/ssd" > ~/Lunabotics/src/isaac_ros/isaac_ros_common/scripts/.isaac_ros_dev-dockerargs
Then run this command:
cd ~/Lunabotics/src/isaac_ros/isaac_ros_common/docker
../scripts/run_dev.sh ~/Lunabotics
Alternatively, you can run this command, which will execute the printf, the echo, and the run_dev.sh script:
./scripts/enter_isaac_ros_container.sh
Dev Container
First run this command to build the container:
../scripts/build_devcontainer_image.sh
Then use Command Palette (Ctrl + Shift + P) to open devcontainer
>Dev Containers: Reopen in Container
When cloning this repository manually on Linux, you must run git submodule update --init --recursive inside the workspace folder to recursively pull all git submodules before building the project.
Use colcon build --symlink-install when building so that Python nodes do not need to be rebuilt every time.
Additionally, use these commands to avoid building unnecessary packages when testing:
colcon build --symlink-install --packages-up-to <package name>
or colcon build --symlink-install --packages-select <package name>
Make sure to source install/setup.bash in every new terminal.
Run rosdep install -i --from-path src --rosdistro $ROS_DISTRO -y --skip-keys "nvblox negotiated" to install package dependencies.
Run ros2 run <package_name> <node_name> to run just the specified node.
Run ros2 launch <package_name> <launch file name> to launch a launch file.
Run rm -rf build install log to clean your workspace if you need to build completely from scratch.
Run docker system prune -a --volumes to remove all Docker images, containers, and volumes from your machine. This is useful when the ISAAC ROS container gets messed up and you want to rebuild the container from scratch.
To normalize line endings in git, use the command:
git config --global core.autocrlf true
Gazebo Installation & Resources
Install Gazebo Fortress by running: sudo apt-get install ros-humble-ros-gz
More info here. Remember that we are using ROS 2 Humble.
Instructions for building the ROS bridge (ros_gz) can be found here.
Information about ROS types -> gazebo types can be found here
Running the Gazebo Simulation
To run the Gazebo simulation:
colcon build --symlink-install --packages-up-to ros_gz_launch
source install/setup.bash
ros2 launch ros_gz_launch UCF_field.launch.py
Then to control the robot, you will need to run:
ros2 run teleop_twist_keyboard teleop_twist_keyboard
in another terminal to control the robot with your keyboard.
Alternatively, you can run these nodes:
ros2 run joy joy_node
ros2 run rovr_control main_control_node
to control the robot using a gamepad and our button bindings assigned in the main_control_node.
ros2 run joy joy_node --ros-args --params-file config/joy_node.yaml
Follow this tutorial to set up Apriltag detection on your machine.
VESC CAN Status Frames Spreadsheet
How to load the CAN modules at startup on Nvidia Jetson
1: Put the following in a .conf file in /modules-load.d/
#Setting up the CAN bus
can
can_raw
mttcan
#eof
2: Find the file /etc/modprobe.d/denylist-mttcan.conf and either delete it or comment out the one line in it (The filename might be .../blacklist-mttcan.conf)
3: Make a script called "can_startup.sh" in the root directory for the system, with the following contents:
#! /usr/bin/sh
sudo ip link set can0 up type can bitrate 500000
sudo ip link set can1 up type can bitrate 500000
4: Run the command "sudo crontab -e" and put this line in the file that appears:
@reboot sleep 5 && echo 'robot' | sudo -S sh /
can_startup.sh 2>&1 | logger -t mycmd
And that should work. If it doesn't and you need to read the output of the crontab, use this command:
sudo grep 'mycmd' /var/log/syslog
Gstreamer Server/Client Instructions
To start gstreamer client make sure to add the deepstream layer to the docker layers
To start the gstreamer client run the following commands:
colcon build --symlink-install --packages-up-to gstreamer
source install/setup.bash
rqt --force-discoverTo start the gstreamer server run the following commands:
colcon build --symlink-install --packages-up-to gstreamer
source install/setup.bash
ros2 run gstreamer server_nodeFollow these instructions.
Follow these instructions.
sudo service docker stop
sudo mv /var/lib/docker /hd/docker
sudo ln -s /hd/docker /var/lib/docker # Create a symbolic link
sudo service docker start