MFI AMR Workspace Setup and Usage

Note: Before you begin, verify that you have sufficient storage space available on your device (At least 30 GB).

On Jetson Xavier platforms, an external drive is required to have enough storage space. Follow the instructions in Jetson Xavier AGX SD card Setup to configure the storage

Setup

1. Configure nvidia-container-runtime as the default runtime for Docker.

   Using your text editor of choice, add the following items to /etc/docker/daemon.json.

   {
       ...
       "runtimes": {
           "nvidia": {
               "path": "nvidia-container-runtime",
               "runtimeArgs": []
           }
       },
       "default-runtime": "nvidia"
       ...
   }

2. Then, restart Docker:

   sudo systemctl daemon-reload && sudo systemctl restart docker

3. Install Git LFS in order to pull down all large files:

   sudo apt-get install git-lfs

   git lfs install

4. Navigate to src

   Note: Remember to replace <path-to-mfi_amr_ws> with the correct path in the following command.

   cd <path-to-mfi_amr_ws>/src

5. Install vcstool -

   curl -s https://packagecloud.io/install/repositories/dirk-thomas/vcstool/script.deb.sh | sudo bash
   sudo apt-get update
   sudo apt-get install python3-vcstool

6. Clone all required repos using vcstool-

   1. isaac_ros_common
   2. realsense (4.51.1)

   vcs import < ros2.repos

7. Clone the librealsense repo setup udev rules. Remove any connected relasense cameras when prompted:

    cd /tmp && \
    git clone https://github.com/IntelRealSense/librealsense && \
    cd librealsense

   sudo ./scripts/setup_udev_rules.sh

8. Configure the container created by isaac_ros_common/scripts/run_dev.sh to include librealsense. Create the .isaac_ros_common-config file in the isaac_ros_common/scripts directory:

   Note: Remember to replace the two instances of <path-to-mfi_amr_ws> with the correct path in the following command.

   cd <path-to-mfi_amr_ws>/src/isaac_ros_common/scripts && \
   touch .isaac_ros_common-config && \
   echo CONFIG_IMAGE_KEY=foxy.realsense_custom.mfi_amr > .isaac_ros_common-config && \
   echo CONFIG_DOCKER_SEARCH_DIRS="(<path-to-mfi_amr_ws>/docker)" >> .isaac_ros_common-config

9. Create alias to launch the docker container

Note: Remember to replace the two instances of <path-to-mfi_amr_ws> with the correct path.

Edit the /etc/bash.bashrc file using sudo nano /etc/bash.bashrc and add an entry as follows - .

alias ros2_foxy_docker="cd <path-to-mfi_amr_ws>/src/isaac_ros_common && ./scripts/run_dev.sh <path-to-mfi_amr_ws>"

Then, run this command to source the bashrc

source /etc/bash.bashrc

12. Launch the docker container using the run_dev.sh script. If you run this command twice then the script will just attach a new shell to the existing container.

ros2_foxy_docker

13. Build workspace using colcon build

colcon build --packages-skip-regex isaac_ros*

Then, run this command -

source install/setup.bash

Testing Installation

14. Testing Velodyne VLP16
    Run the following command in the container -

ros2 launch velodyne velodyne-all-nodes-VLP16-launch.py

Run the following command to check the frequency of the point cloud topic. You should see an average rate of 10 if everything goes well.

ros2 topic hz /velodyne_points

15. Testing realsense installation:

Note: Plug in your realsense camera before launching the docker container.

ros2 launch realsense2_camera rs_launch.py

Run the following command to check the frequency of the rgb and depth image topic. You should see an average rate greater than 10 if everything goes well.

ros2 topic hz /camera/color/image_raw

ros2 topic hz /camera/depth/image_rect_raw

Jetson Xavier AGX SD card Setup

To automount the SD card, an entry with a reference to the device path or UUID of the device has to be added in fstab ( File System Table ) file.

Run this command to identify the attached device and note its UUID.

sudo blkid

Edit the fstab file by executing the following command in the terminal. Add an entry similar to the one below after changing the UUID.

sudo nano /etc/fstab

UUID=<put-UUID-here> /mnt  ext4    defaults    0   0

VSCode Setup

Prerequisites

You should already have Docker and VSCode with the remote containers plugin installed on your system.

• docker
• vscode
• vscode remote containers plugin

Steps to connect to container -

1. Launch the container using ros2_foxy_docker
2. Launch VS code and click on Open a Remote Window towards the bottom left corner
3. Choose Attach to a Running Container
4. Select the appropriate container (isaac_ros_dev*)
5. Set working directory to /workspaces/isaac_ros-dev in VS Code

Development Workflow

Using open-source ROS2 packages

If you require installing any packages add them to the Dockerfile. If you require building them then add them to ros2.repos and use vcstool to manage them

Updating Docker

• Every time we execute the ros2_foxy_docker, it will automatically build the Docker image if any modifications are detected in the dockerfile.
• If there is a need to install external dependencies, by adding those to the Dockerfile (docker/Dockerfile.mfi_amr) and pushing the updates to github, we can ensure that these dependencies are accessible to all development environments.
• You can follow the template present in Dockerfile as a starting point to add any dependencies.

Git workflow

We can use tree types of branches - master, devel, feat, hotfix as follows (Reference)

• A devel branch is created from master
• feat branches are created from devel
• When a feat is complete it is merged into the devel branch
• After milestone demonstrations,devel is merged into master
• If an issue in master is detected a hotfix branch is created from master
• Once the hotfix is complete it is merged to both devel and master

Good practices related to commits -

• Include the ROS2 package names in the commit message e.g. "[mapping] some change"
• If possible, include only changes to a single package in a commit
• Break down large changes into smaller commits