Skip to content

Getting Started#

Foundational setup for deploying Autoware on Ubuntu 22.04 (AMD64/ARM64).

Deployment Overview#

graph TD
    A[Configure Host System] --> B[Pull Docker Image]
    B --> C[Enter Container]
    C --> D[Configure Environment]
    D --> E[Install Autoware Packages]
    E --> F[Verify Installation]

    style A fill:#e3f2fd
    style F fill:#c8e6c9

System Requirements#

Hardware Requirements#

Minimum Specifications#

  • CPU: 8-core x86_64 (Intel/AMD) or ARMv8 (ARM64)
  • RAM: 32 GB (16 GB for simulation-only)
  • Storage: 30 GB of free space
  • GPU: NVIDIA GPU with CUDA 11.8+ and Compute Capability 5.0+ (required for perception)
  • CPU: 16+ cores
  • RAM: 64 GB+
  • Storage: 100 GB+ SSD
  • GPU: NVIDIA RTX 3080+ (x86_64) or AGX Orin (ARM)

Network#

  • Gigabit Ethernet for sensors
  • Internet access for setup
  • Optional: Secondary interface for vehicle communication

Operating System#

  • Ubuntu 22.04 LTS
  • Kernel 5.15+
  • Real-time kernel recommended for production

Deployment Options#

Choose between two deployment approaches:

  1. Native Installation - Direct installation on the host system
  2. Containerized Deployment (Recommended) - Using Docker for isolation and reproducibility

System Preparation#

1. Install Essential Packages#

# Update package lists and upgrade system
sudo apt update && sudo apt upgrade -y

# Install essential tools
sudo apt install -y \
  build-essential \
  cmake \
  git \
  wget \
  curl \
  gnupg \
  lsb-release \
  software-properties-common \
  python3-pip \
  python3-venv \
  ansible

Containerized Deployment Workflow#

If you prefer to install Autoware on Intel/AMD-based ECUs from native source code, you may refer to this page and skip the remaining of this page.

This section describes the recommended containerized deployment approach using Docker, Ansible, and Debian packages.

1. Configure Host System#

Before deploying containers, ensure your host system has the necessary tools:

  1. Install Docker Engine: Follow the official Docker installation guide for Ubuntu 

    # After installation, add user to docker group
sudo usermod -aG docker $USER
newgrp docker

  2. Install NVIDIA Container Toolkit: Follow the official NVIDIA installation guide

  3. Note: Pre-installed with JetPack on NVIDIA Jetson devices

  4. Install NVIDIA Driver: Ensure NVIDIA drivers are installed on the host 

    # Verify NVIDIA driver installation
nvidia-smi

2. Pull Pre-configured Images#

Use pre-configured Docker images that include CUDA and necessary dependencies:

For x86_64 Systems#

docker pull ghcr.io/autowarefoundation/autoware:universe-cuda-20250414

For ARM64 Systems (Jetson)#

docker pull nvcr.io/nvidia/l4t-tensorrt:r8.6.2-devel

Note: These images include CUDA pre-installed. For custom CUDA installations, see Workflow Customization.

3. Enter Container Environment#

Launch the container with GPU support and necessary mounts:

docker run -it --name autoware-dev \
  --gpus all \
  --runtime nvidia \
  --network host \
  --privileged \
  -v ~/autoware-deployment:/workspace \
  -v /tmp/.X11-unix:/tmp/.X11-unix:rw \
  -e DISPLAY=$DISPLAY \
  <image-name> \
  /bin/bash

Replace <image-name> with the image you pulled in step 2.

4. Configure Container Environment with Ansible#

Inside the container, download and run the Ansible configuration:

git clone -b 2025.02 https://github.com/autowarefoundation/autoware.git
cd autoware
./setup-dev-env.sh

5. Install Autoware Debian Packages#

After Ansible configuration, visit the Autoware Debian package release page and install Autoware packages:

# Configure Autoware repository (example for x86_64)
# Visit the release page and download autoware-localrepo_2025.2-1_amd64.deb. Then,
sudo apt install ./autoware-localrepo_2025.2-1_amd64.deb

# Update and install Autoware
sudo apt update
sudo apt install -y autoware-full

# Source the environment
source /opt/autoware/autoware-env

For ARM64/Jetson systems, use the appropriate repository package from the release page.

6. Verification#

Verify the installation is successful:

# Check ROS 2 installation
ros2 --version

# Check Autoware packages
ros2 pkg list | grep autoware

# Verify CUDA (should show GPU information)
nvidia-smi

# Run a simple test
ros2 launch autoware_launch planning_simulator.launch.xml

If all checks pass, your containerized Autoware environment is ready for use.

Post-Installation Verification#

This section uses a shell script to verify if the Autoware is successfully deployed to the system.

1. System Check Script#

Download the system check script here and move the shell script to the working directory. Executes the verification script to verify the deployment.

chmod +x verify-installation.sh
./verify-installation.sh

2. Test Basic Functionality#

Follow the tutorial to run the planning simulation.

Next Steps#

With the base system configured, proceed to one of the following sections:

  1. x86_64-based ECU Configuration for Intel/AMD platforms
  2. ARM-based ECU Configuration for NVIDIA Jetson and the other ARM platforms