When setting up the Jetson Nano, you will need a monitor, keyboard, mouse, and internet connection provided directly to the Nano. We recommend performing this setup before installing the Rambutan in your electronics box. This will give you room to plug in HDMI and USB cables for the setup process. If you don't have the ability to provide an ethernet connection to the internet (i.e. you only have WiFi), you will need a USB WiFi adapter. Some adapters do not work out of the box with the Nano as they lack compatible drivers. Be sure to check that the adapter you intend to use is supported before making any purchases.
First, you will need to create a bootable SD card and set up your Jetson Nano, following the guide found here: https://developer.nvidia.com/embedded/learn/get-started-jetson-nano-devkit#intro
As an additional resource, which includes some more detail, you can read: https://www.pyimagesearch.com/2020/03/25/how-to-configure-your-nvidia-jetson-nano-for-computer-vision-and-deep-learning/
NOTE: At the time of writing this guide, the latest version available was 32.4.2, which includes Ubuntu 18.04 and CUDA 10.2. The rest of this guide will assume the use of these versions.
After finishing the hardware and SD image setup process, make sure you have an internet connection, power on the Nano and walk through the following commands in a terminal:
# First, update pre-installed software
sudo apt-get update
sudo apt-get upgrade
# Next we will install some basic tools and dependencies that we will need:
sudo apt-get update -y
sudo apt-get install -y --no-install-recommends \
apt-transport-https \
apt-utils \
build-essential \
cmake \
curl \
git \
gnupg2 \
less \
libv4l-dev \
libxml2-dev \
libxslt1-dev \
lsb-release \
nano \
openssh-server \
python3-pip \
sudo \
supervisor \
udev \
vim \
wget
# Install ROS2 Eloquent:
# Note: We have to use Eloquent for now, because Foxy is only supported on Ubuntu 20.04, which isn't released for the Nano at the time of writing this guide.
# Set up apt sources list
curl -s https://raw.githubusercontent.com/ros/rosdistro/master/ros.asc | sudo apt-key add -
sudo sh -c 'echo "deb [arch=$(dpkg --print-architecture)] http://packages.ros.org/ros2/ubuntu $(lsb_release -cs) main" > /etc/apt/sources.list.d/ros2-latest.list'
# Install ROS2 Eloquent and some additional packages
sudo apt-get update -y
sudo apt-get install -y --no-install-recommends \
ros-eloquent-ros-base \
python3-argcomplete \
python3-colcon-common-extensions \
python3-rosdep \
python3-vcstool \
python3-wheel
# Install packages required to run mavproxy and status_light services
sudo -H pip3 install mavproxy Jetson.GPIO bluerobotics-ping
# Install Conan C++ package manager (used to build software)
pip3 install conan==1.26.1
# Add user and usr/local bin folders to PATH, so you can run conan
echo "PATH=\"\${PATH}:/home/${USER}/.local/bin:/usr/local/bin\"" >> ~/.bashrc
source ~/.bashrc
## ----------------------
# Set up various permissions and rules to make sure all files, cameras, and serial ports can be accessed
# Change permissions on /opt so you can write to it as a non-root user
sudo chown -R ${USER}:${USER} /opt
# First add your user to necessary groups
sudo usermod -a -G video ${USER}
sudo usermod -a -G tty ${USER}
sudo usermod -a -G dialout ${USER}
# Disable console on /dev/ttyTHS1 (pre-enabled on stock Nano images)
sudo systemctl stop nvgetty
sudo systemctl disable nvgetty
# Add udev rule to allow non-root access on /dev/ttyTHS* serial devices
sudo sh -c "echo 'KERNEL==\"ttyTHS*\", MODE=\"0666\"' > /etc/udev/rules.d/55-tegraserial.rules"
## -----------------------
# Download the ZED SDK that matches the Jetpack 4.4, CUDA 10.2 release on the Nano
curl -O https://stereolabs.sfo2.cdn.digitaloceanspaces.com/zedsdk/3.2/ZED_SDK_Tegra_JP44_v3.2.0.run
# Make the installer executable
chmod +x ZED_SDK_Tegra_JP44_v3.2.0.run
# Run and follow the instructions. Enter 'y' for all y/n prompts and leave values as their default
./ZED_SDK_Tegra_JP44_v3.2.0.run
# The output should look something like:
# Verifying archive integrity... 100% All good.
# Uncompressing 'ZED camera SDK by Stereolabs' 100%
# Installing...
# To continue you have to accept the EULA. Accept [Y/n] ?y
# Installing...
# Installation path: /usr/local/zed
# [sudo] password for youruser:
# Checking CUDA version...
# OK: Found CUDA 10.2
# Do you want to install the static version of the ZED SDK [Y/n] ?y
# Do you want to install the Object Detection module (recommended), cuDNN and TensorRT will be installed [Y/n] ?y
# Install samples (recommended) [Y/n] ?y
# Installation path: /usr/local/zed/samples/
# Do you want to auto-install dependencies (recommended) ? following packet will be installed via the package manager : libjpeg-turbo8 libturbojpeg libusb-1.0 libopenblas-dev libv4l-0 curl unzip libpng-dev libglew-dev freeglut3-dev qt5-default libqt5opengl5 libqt5svg5 [Y/n] ?y
# ...
# Dependencies installation complete
# ZED SDK installation complete !
# After successfully installing the ZED SDK, *WITH* the camera connected, run the following, which will open a GUI and also download the calibration file for your specific camera:
/usr/local/zed/tools/ZED_Explorer
# Now that everything has been installed, reboot so that certain modifications take effect
sudo reboot nowSimilar to the groundstation setup, we need to clone and build the software. NOTE: There are a few additional steps involved.
# Download the software to the /opt/openasv directory
git clone https://github.com/oceansystems/asv /opt/openasv
# Navigate to the software directory
cd /opt/openasv/vehicle
# Install systemd service files
sudo cp -a ./res/systemd/. /lib/systemd/system/
# Install conan configuration
mkdir -p ~/.conan/profiles
cp /opt/openasv/vehicle/res/conan/default ~/.conan/profiles/
# Build the conan dependencies
# NOTE: This will take some time on the first run
cd /opt/openasv/vehicle/conan-packages
./build.sh
# Build the vehicle software
# NOTE: This will take some time on the first run
cd /opt/openasv/vehicle/src
./build.sh
# Once the software is successfully built, enable the installed systemd service files for the software
sudo systemctl enable asv_bridge
sudo systemctl enable mavproxy
sudo systemctl enable recorder
sudo systemctl enable status_lightWe will want to make sure that when using a USB flash drive, the drive is always automatically mounted at boot and mounts to a consistent filesystem location, so our software can locate it.
First, format a flash drive using your preferred method. We recommend using the FAT32 format and the following instructions will assume a drive formatted to FAT32.
Once you have a USB drive formatted to FAT32, plug it into a free USB port on the Nano. Next, follow these steps:
# With your USB drive plugged in, run:
sudo blkid
# Look for your flash drive in the output of the previous command. \
# As an example, for a flash drive with a label name of "CYBIE", this is returned:
# /dev/sda1: LABEL="CYBIE" UUID="B668-998A" TYPE="vfat" PARTUUID="e1ee37e1-01"
# Create a variable with the UUID information for your device. As an example:
export USB_UUID="B668-998A"
# Copy the UUID value and add it to your /etc/fstab file in the following manner:
sudo sh -c "UUID=${USB_UUID} /media/${USER}/usb vfat rw,nofail,umask=022,uid=1000,gid=1000 0 0"
# Create the directory and change its permissions
sudo mkdir -p /media/${USER}/usb
sudo chown ${USER}:${USER} /media/${USER}/usb
# Reboot to get the above changes to take effect and then verify that the device is successfully mounted by writing a file to it and checking it on another device
sudo reboot nowIn general, you will want to configure your software to write data to a USB3 flash drive, such that you don't accidentally fill up the Nano's main filesystem. This will also allow you to easily offload the recordings, which can quickly grow to several gigabytes in size.
To configure the storage location, modify the /opt/openasv/vehicle/src/recorder/config/recorder.yaml file, which has contents that default to:
recorder:
ros__parameters:
minimum_sample_period_msecs: 2000
storage_path: "/tmp"If you configured your USB drive to mount to /media/${USER}/usb, set the storage_path variable to match that (replacing ${USER} with your actual username).
If you like, you can also modify the minimum period that recorder records samples at. By default, recorder will record samples no faster than once every 2000ms (recording is triggered on the earliest GPS update that happens after this period). By increasing the minimum period, you can control how quickly your storage disk space fills up, depending on your application and setup's needs.