This package supplies a docker image with everything needed to start using micro-ROS with the Zephyr RTOS. Use this together with micro-ROS docker images to build and run your first micro-ROS app!
This docker image contains a full toolchain to build and flash a micro-ROS application to a microcontroller (or to run it in host emulation). Combine this with the micro-ros-agent
image from micro-ROS docker to get a fully functional micro-ROS workbench.
Disclaimer: This package is still under construction and not yet ready for any serious use.
- Building the Image: How to build this docker image for your specific setup
- Running the Image: How to run this image correctly
- Example Applications: Examples of real-world application
- Serial Communication with Microcontroller Board: Using this image to build an example for a microcontroller board, using serial communication
- UDP-based Communication with Host Emulation: Using this image to run a micro-ROS application in a host emulation, using udp communcation
Before using this image, you need to install docker on your system. If you don't have this already, take a look at the docker installation instructions.
You can build the image using
$ git clone git@github.com:RobertWilbrandt/microros_zephyr_docker.git
$ docker build microros_zephyr_docker \
--tag wilbrandt/microros_zephyr:foxy \
--build-arg PLATFORM_BOARD=<Your Board>
The PLATFORM_BOARD
argument specifies which board you are using. The format it uses depends on:
- If your board is already supported by micro-ROS, use the name shown in the micro_ros_setup supported platforms section.
- If your board is not yet supported by micro-ROS (but supported by Zephyr) and you just want to try out this configuration, use the name used by Zephyr (as found in its board side in the documentation). This will give you warnings during your build.
You can use this image just like any other docker image. Recommended usage is
$ docker run -it --rm --privileged \
--volume <absolute_path_to_your_apps_folder>:/uros_apps \
wilbrandt/microros_zephyr:foxy
The effects of these flags are:
-it
: KeepsSTDIN
open and creates a pseudo-TTY for you to interact with. With these two options you can use this like a normal shell.--rm
: Removes the container after it exits. You might not want to use this if you want to keep an internal state in the image and attach to it later on.--privileged
: This is needed to give the container all permissions needed to flash your microcontroller.--volume
: Maps your custom apps into the image. This has to be an absoulte path, and it should contain the directories of your custom apps. The toolchain inside the image expects your apps to be mapped to/uros_apps
and will error out if it doesn't find any there.
Suppose you want to build one of the example applications distributed in zephyr_apps as part of micro-ROS, for example ping_pong. You use the STM32F429I-DISC1 discovery kit.
The first step to achieve this consists of getting the application and building this docker image:
$ git clone git@github.com:micro-ROS/zephyr_apps.git
$ git clone git@github.com:RobertWilbrandt/microros_zephyr_docker.git
$ docker build microros_zephyr_docker --tag wilbrandt/microros_zephyr_stm32f429i_disc1:foxy \
--build-arg PLATFORM_BOARD=stm32f429i_disc1
(Okay i am cheating a bit here, as the micro-ROS applications in zephyr_apps
will check if the board is supported, which is not the case for stm32f429i_disc1
. You can just comment out the check in CMakeLists.txt
though and i will try to get a patch in that removes this for basic applications)
This will take a while, as a full ROS foxy image with a complete micro-ROS and Zephyr installation needs to be installed. When this is done, you can run the image like this:
$ docker run -it --rm --privileged --volume $(pwd)/zephyr_apps/apps:/uros_apps \
wilbrandt/microros_zephyr_stm32f429i_disc1:foxy
Inside this image you can now configure, build and flash the application. The example board is connected using a serial transport (piped through the on-board debugger connected via USB) and the Zephyr board page tells us we want to use UART1
.
# ros2 run micro_ros_setup configure_firmware.sh ping_pong --transport serial --dev 1
# ros2 run micro_ros_setup build_firmware.sh -f
# ros2 run micro_ros_setup flash_firmware.sh
You have now flashed your micro-ROS image to your board! When developing, you might want to keep this docker image open (to prevent having to fully rebuild every time).
An easy way to connect to your MCU is to use the micro-ROS micro-ros-agent
docker image:
$ docker run -it --rm --net=host --privileged microros/micro-ros-agent:foxy \
serial --dev /dev/ttyACM0
Make sure to replace the /dev/ttyACM0
with your actual device. You might need to reset your microcontroller, for most development boards out there this should be possible by just pressing a button. The agent should show some lines notifying you about a new connection.
Now you should be ready to try out your demo: Open two shells and enter:
# Shell 1
$ source /opt/ros/foxy/setup.bash
$ ros2 topic echo /microROS/pong
# Shell 2
$ source /opt/ros/foxy/setup.bash
$ ros2 topic pub -1 /microROS/ping std_msgs/msg/Header "frame_id: 'Hello World!'"
You should now see your message in the first shell.
Zephyr also allows you to build an application right on your computer, emulating some of the driver interfaces seamlessly (even complex peripherals like displays, network stacks and BLE). If you don't have access to a microcontroller board right now but still want to take a look at micro-ROS, i encourage getting started using this. It can also be incredibly useful for prototyping and debugging embedded applications. More information on this can be found in the Zephyr documentation.
Using this host emulation is nearly identical to the above example. You start by cloning your application (in this case int32_publisher
) and building this docker image:
$ git clone git@github.com:micro-ROS/zephyr_apps.git
$ git clone git@github.com:RobertWilbrandt/microros_zephyr_docker.git
$ docker build microros_zephyr_docker --tag wilbrandt/microros_zephyr_host:foxy \
--build-arg PLATFORM_BOARD=host
After this, start the container. Inside it, you can now configure and build an image. Note that we need to replace the --privileged
flag with --net=host
, as we are now communicationg via network.
$ docker run -it --rm --net=host --volume $(pwd)/zephyr_apps/apps:/uros_apps \
wilbrandt/microros_zephyr_stm32f429i_disc1:foxy
# ros2 run micro_ros_setup configure_firmware.sh int32_publisher \
--transport udp --ip 127.0.0.1 --port 8888
# ros2 run micro_ros_setup build_firmware.sh -f
Note that we did not yet start the application (using flash_firmware.sh
), as this will just error out when it doesn't find an agent running. We can (again) start an agent in a seperate container, and as this doesn't require any device access we don't need any --privileged
flag for this to work:
$ docker run -it --rm --net=host microros/micro-ros-agent:foxy udp4 --port 8888
Now go back to your first container and start the application:
# ros2 run micro_ros_setup flash_firmware.sh
You should see some messages in the agent container. In a normal ROS 2 foxy machine, your can now check the functionality:
$ source /opt/ros/foxy/setup.bash
$ ros2 topic echo /zephyr_int32_publisher