Skip to main content
Python Software Foundation 20th Year Anniversary Fundraiser  Donate today!

A tool to build ROS workspaces for various target architectures and platforms.

Project description

ROS / ROS 2 Cross Compile Tool

License Documentation Status

A tool to automate compiling ROS and ROS 2 workspaces to non-native architectures.

:construction: ros_cross_compile relies on running emulated builds using QEmu, #69 tracks progress toward enabling cross-compilation.

Supported targets

This tool supports compiling a workspace for all combinations of the following:

  • Architecture: armhf, aarch64, x86_64
  • ROS Distro
    • ROS: kinetic, melodic
    • ROS 2: dashing, eloquent, foxy
  • OS: Ubuntu, Debian

NOTE: ROS 2 supports Debian only as a Tier 3 platform. This means that there are not apt repositories available for the ROS 2 Core on this platform. Because of that, when targeting Debian for a ROS 2 workspace, you must also include the source for the core as well. It is recommended to use a release branch of ros2.repos from to do so, rather than master, so that you are not affected by development branch bugs and API changes.

Supported hosts

This tool officially supports running on the following host systems. Note that many others likely work, but these are being thoroughly tested.

  • Ubuntu 18.04 Bionic Beaver
  • OSX Mojave



This tool requires that you have already installed

  • Docker
    • Follow the instructions to add yourself to the docker group as well, so you can run containers as a non-root user
  • Python 3.5 or higher

If you are using a Linux host, you must also install QEmu (Docker for OSX performs emulation automatically):

sudo apt-get install qemu-user-static

Installing ros_cross_compile

To install the stable release,

pip3 install ros_cross_compile

If you would like the latest nightly build, you can get it from Test PyPI

pip3 install --index-url ros_cross_compile

How it works, high level

  1. Collect dependencies
    1. Create a Docker image that has rosdep
    2. Run the rosdep image against your target workspace to output a script that describes how to install its dependencies
  2. Create "sysroot image" that has everything needed for building target workspace
    1. Use a base image for the target architecture (aarch64, armhf, ...)
    2. Install build tools (compilers, cmake, colcon, etc)
    3. Run the dependency installer script collected in Step 1 (if dependency list hasn't changed since last run, this uses the Docker cache)
  3. Build
    1. Runs the "sysroot image" using QEmu emulation
    2. colcon build
  4. (Optional) Create runtime image
    1. Creates a docker image that can be used on the target platform to run the build. See "Runtime Image" section.


This package installs the ros_cross_compile command. The command's first argument is the path to your ROS workspace.

Here is a simple invocation for a standard workflow.

ros_cross_compile /path/to/my/workspace --arch aarch64 --os ubuntu --rosdistro dashing

For information on all available options, run ros_cross_compile -h. See the following sections for information on the more complex options.

Package Selection and Build Customization

To choose which packages to install dependencies for, this tool runs colcon list on your workspace. To build, it runs colcon build.

You can provide arbitrary arguments to these commands via the colcon defaults.yaml.

You can either specify the name of this file via ros_cross_compile --colcon-defaults relative/path/to/defaults.yaml, or if not specified, a file called defaults.yaml will be used if present.

For example, there are repositories checked out in your workspace that contain packages that are not needed for your application - some repos provide many packages and you may only want one! In this scenario there is a "bringup" package that acts as the entry point to your application:

# my_workspace/defaults.yaml
  # only install dependencies for source packages that my package depends on
  packages-up-to: [my_application_bringup]
  # only build up to my package
  packages-up-to: [my_application_bringup]
  # example of a boolean commandline argument
  merge-install: true

Custom rosdep script

Your ROS application may need nonstandard rosdep rules. If so, you have the option to provide a script to be run before the rosdep install command collects keys.

This script has access to the "Custom data directory" same as the "Custom setup script", see the following sections. If you need any extra files for setting up rosdep, they can be accessed via this custom data directory.

Note that:

  1. Rosdeps are always collected in an Ubuntu Bionic container, so scripts must be compatible with that

Here is an example script for an application that adds extra rosdep source lists

cp ./custom-data/rosdep-rules/raspicam-node.yaml /etc/ros/rosdep/custom-rules/raspicam-node.yaml
echo "yaml file:/etc/ros/rosdep/custom-rules/raspicam-node.yaml" > /etc/ros/rosdep/sources.list.d/22-raspicam-node.list
echo "yaml" > /etc/ros/rosdep/sources.list.d/18-aws-python.list

Tool invocation for this example:

ros_cross_compile /path/to/my/workspace --arch aarch64 --os ubuntu \
  --custom-rosdep-script /path/to/ \
  --custom-data-dir /arbitrary/local/directory

Custom setup script

Your ROS application may have build needs that aren't covered by rosdep install. If this is the case (for example you need to add extra apt repos), use the option --custom-setup-script to execute arbitrary code in the sysroot container.

The path provided may be absolute, or relative to the current directory.

Keep in mind

  • It's up to the user to determine whether the script is compatible with chosen base platform
  • Make sure to specify non-interactive versions of commands, for example apt-get install -y, or the script may hang waiting for input
  • You cannot make any assumptions about the state of the apt cache, so run apt-get update before installing packages
  • The script runs as root user in the container, so you don't need sudo

Below is an example script for an application that installs some custom Raspberry Pi libraries.

apt-get update
apt-get install -y software-properties-common

# Install Raspberry Pi library that we have not provided a rosdep rule for
add-apt-repository ppa:rpi-distro/ppa
apt-get install -y pigpio

Custom data directory

Your custom setup or rosdep script (see preceding sections) may need some data that is not otherwise accessible. For example, you need to copy some precompiled vendor binaries to a specific location, or provide custom rosdep rules files. For this use case, you can use the option --custom-data-dir to point to an arbitrary path. The sysroot build copies this directory into the build environment, where it's available for use by your custom setup script at ./custom-data/.


Custom data directory (/arbitrary/local/directory)

+-- my-data/
|   +-- something.txt

Setup Script (/path/to/

cat custom-data/something.txt

Tool invocation:

ros_cross_compile /path/to/my/workspace --arch aarch64 --os ubuntu \
  --custom-setup-script /path/to/ \
  --custom-data-dir /arbitrary/local/directory

Now, during the sysroot creation process, you should see the contents of something.txt printed during the execution of the custom script.

NOTE: for trivial text files, as in the preceding example, you could have created those files fully within the --custom-setup-script. But for large or binary data such as precompiled libraries, this feature comes to the rescue.

Runtime Image

ros_cross_compile can optionally create and tag a Docker image that contains the build output and its runtime dependencies.

The argument --runtime-tag takes a single value, which is the tag used for the output image.

ros_cross_compile $workspace --runtime-tag $OUTPUT_IMAGE

One way to deploy this image is to push it to a registry, from where it can be pulled onto a target platform

docker push $OUTPUT_IMAGE

The image contains any necessary emulation binaries to run locally if desired for smoke testing.

docker run -it $OUTPUT_IMAGE
# In the shell inside the running container, the setup is already sourced for the default entrypoint
ros2 launch my_package

Note: Currently this feature is a thin layer on top of the image used for building, so it is not a fully minimal image - it contains build tools, build dependencies, and test dependencies in addition to the necessary runtime dependencies. Future work is planned to slim down this output image to a properly minimal runtime. This work is tracked in


For a new user, this section walks you through a representative use case, step by step.

This tutorial demonstrates how to cross-compile the File Talker tool against ROS 2 Dashing, to run on an ARM64 Ubuntu system. You can generalize this workflow to any .repos file for your project.

NOTE: this tutorial assumes a Debian-based (including Ubuntu) Linux distribution as the host platform.

Creating a simple source workspace

  1. Create a directory for your workspace
    • mkdir cross_compile_ws
    • cd cross_compile_ws
  2. Create a .repos file for vcs
    • file_talker.repos
        type: git
        version: master
  3. Check out the sources to build
    • mkdir -p src
    • vcs import src < file_talker.repos

Running the cross-compilation

ros_cross_compile $(pwd) --rosdistro dashing --arch aarch64 --os ubuntu

Here is a detailed look at the arguments passed to the script:

  • --rosdistro dashing

You may specify both ROS and ROS 2 distributions by name, for example, kinetic (ROS) or dashing (ROS 2). ros_cross_compile -h prints the supported distributions for this option

  • --arch aarch64

Target the ARMv8 / ARM64 / aarch64 architecture (which are different names for the same thing). ros_cross_compile -h prints the supported architectures for this option.

  • --os ubuntu

The target OS is Ubuntu - the tool chooses the OS version automatically based on the ROS Distro's target OS. In this case for ROS 2 Dashing - 18.04 Bionic Beaver.

Outputs of the build

Run the following command

ls cross_compile_ws

If the build succeeded, the directory looks like this:

|-- file_talker/
|-- install_aarch64/
|-- build_aarch64/

The created directory install_aarch64 is the installation of your ROS workspace for your target architecture. You can verify this:

$ file cross_compile_ws/install_aarch64/lib/file_talker/file_talker                                                               0s
cross_compile_ws/install_aarch64/lib/file_talker/file_talker: ELF 64-bit LSB shared object, ARM aarch64, version 1 (GNU/Linux), dynamically linked, interpreter /lib/ld-, for GNU/Linux 3.7.0, BuildID[sha1]=02ede8a648dfa6b5b30c03d54c6d87fd9151389e, not stripped

Using the build on a target platform

Copy install_aarch64 onto the target system into a location of your choosing. It contains the binaries for your workspace.

If your workspace has any dependencies that are outside the source tree - that is, if rosdep had anything to install during the build - then you still need to install these dependencies on the target system.

# Run this on the target system, which must have rosdep already installed
# remember `rosdep init`, `rosdep update`, `apt-get update` if you need them
rosdep install --from-paths install_aarch64/share --ignore-src --rosdistro dashing -y

Now you may use the ROS installation as you would on any other system

source install_aarch64/setup.bash
ros2 run file_talker file_talker my_text_file.txt


This library is licensed under the Apache 2.0 License.

Build status

ROS 2 Release Branch Name Development Source Debian Package X86-64 Debian Package ARM64 Debian Package ARMHF Debian package
Latest master Test Pipeline Status N/A N/A N/A N/A
Dashing dashing-devel Build Status Build Status Build Status N/A N/A

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for ros-cross-compile, version 0.6.0
Filename, size File type Python version Upload date Hashes
Filename, size ros_cross_compile-0.6.0-py3-none-any.whl (35.1 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size ros_cross_compile-0.6.0.tar.gz (29.7 kB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring DigiCert DigiCert EV certificate Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page