Skip to main content

Developing Custom Applications


We now have a set of tutorials that walk you through creating custom applications that interact with cameras and drive your Amiga from the brain!

Check them out at Brain App Tutorials.

Quick Tips​

  • The Amiga user is able to install APT packages.
    • Edit the "" script in the sample git repository.
  • We leverage a venv per application.
    • The contents of requirements.txt will be installed by the script.
  • If you've updated the requirements remove the hidden ".lock" before relaunching the app from the Brain UI.
  • You may use VSCode remote development/containers to edit directly on the Amiga brain as well.


apt install wget

Prepare an app​

Follow the instructions below to select a Brain App Example prepared by farm-ng or create an app from the amiga-app-template repository.

Select an app​

To test the app deployment process of a fully made app developed by farm-ng, you can choose either the camera-streamer example or the virtual-joystick example from Brain App Example. To clone one of the examples click on one of the links in the previous sentence and click the green code button in GitHub and then copy the text highlighted in yellow. Create a new templated repository

Now in the terminal enter the following commands, but where it says git clone http... and replace the https.. part with your copied text.

cd <to-your-base-directory>
git clone # To clone the camera-streamer repo or other.

Now you can follow the instructions below as though you were developing a custom application and simply replace any <YOUR-USERNAME/YOUR-REPOSITORY> placeholder with that of the path to the prepared app's Git repository, e.g., farm-ng/virtual-joystick.

You can skip any sections specific to setting up a new project, including Create an app and Customizing an app.

Create an app​

Create a repository from the template​

To create your own custom application navigate to the amiga-app-template repository on GitHub.


You need to be signed into your GitHub account to see the option to Use this template.

Create a new templated repository

Written instructions (same as video above):

  1. Click on green Use this template button (top right) to create a new repository based on this repo
  2. Fill out:
    - Owner: (your_username)
    - Repository Name: whatever-you-want
    - Should be between between 4-17 characters, and
    - Use underscores (_) or dashes (-) for separating words (no spaces!)
    -Select Public box
    -Click [Create repository from template]

Now that you are done creating your repository, you have two options:

  1. Work in your local workspace.
  2. Work in a remote Remote-SSH session using in vs-code.

In both cases you will have to clone the repository you created above by following the commands below.

cd <to-your-base-directory>
git clone

Project structure​

In vs-code, you can see the project structure on the left side in the EXPLORER:

Below are listed the most important components.

repository-name/  # Root level of the project.
β”œβ”€β”€ libs/ # Contains private libraries.
| └── project_name/
| β”œβ”€β”€ # e.g a operators submodule
| └── project_name_subpackage/
| └── # e.g a utilities submodule
β”œβ”€β”€ src/ # Contains all code needed to run the main gui application.
| β”œβ”€β”€ assets/ # Contains files needed to run the application
| | └── app_logo.png # static images or images for buttons.
| β”œβ”€β”€ res/ # Contains the layout files and UI strings.
| | └── main.kv # main ui layout in Kivy languague
| └── # Is the main entry point for the gui application.
β”œβ”€β”€ test/ # Contains code for test of the private libs.
| └── # sets of unit test.
| # The script to setup the project, create a virtual env. and install dependencies.
└── setup.cfg
# The file containing the metadata of the package, including the name, versioning, etc. Learn more here:

How to setup the project​

Before any changes, lets see if we can run this app locally on your system.

./ # install and run

When running the above script, a virtual environment venv will be created and any dependencies will be installed.

If all goes well, you'll see an dummy kivy application on your screen.

Customizing an app​

In order to customize an app we leverage the setup.cfg that contains all the metadata and package configuration. More can be found here!

The most important first steps are to modify the metadata of the project and dependencies:

  1. Inside the setup.cfg file, adjust the fields under the tag [metadata]
    • For basic users, the package name (project_name in the project structure above) goes in the name field, and must match with the directory name right under the libs/ directory.
      • I.e., change both the name field and the directory name under libs/ to your new app name.
      • VScode should prompt you to change the import names in once you change the directory name.
        • If not, manually change:
          from amiga_package import __version__
          from amiga_package import ops
        • To:
          from <project_name> import __version__
          from <project_name> import ops
    • For advanced users, you can modify as much is compliant with Python setuptools.
  2. Adjust the package dependencies
    • Include whatever extra dependency you need in the install_requires field.
    • Our only requirements are:
      • wheel: for packaging the app.
      • kivy: to generate the graphical user interface (GUI).
      • farm-ng-amiga: the Farm-ng Amiga public SDK.

In case you face issues, connect with us and the community on Discourse

Development and Debug an app​

The workflow for development is pretty much the same as any standard gui application.

  1. Make changes in the code.
  2. Run the code with the play button in vs-code.
    • [Optionally] Add a breakpoint to any line and use the Debug Console to interact.
  3. Go to step 1)

[Optional] Test your application​

In order to validate your functionality, we suggest to add test cases for the internal libs/[package_name] located under test/test_[test_name].py.

To launch the tests: pytest test/ To run specific tests: pytest test/

How to SSH into the Amiga​

If this is your very first time ssh into Amiga follow the instructions under Initial SSH set up otherwise follow the steps at SSH Configuration.

Initial SSH set up​

Follow the steps listed below and then continue on to SSH Configuration to complete the process.

  • To set up ssh, make sure the Amiga and your computer are on the same network (check the Wifi on the Amiga and adjust accordingly).

  • Then check if a file ~/.ssh/config exists on your workstation with the steps below. ~/ is the home folder on your workstation.

  1. To check if this file exists put the commands below in your terminal. If this file does not exist, you'll have to create it.
cd ~ # Navigate to your home folder
ls -a # Check for a folder called `.ssh/`
  1. If ~/.ssh/ does not exist, run:
ssh-keygen # create ~/.ssh/ folder
  1. Create a file called config (no extension) in ~/.ssh
touch ~/.ssh/config

SSH Configuration​

Now it is time to configure your config file to ssh into your Amiga.

gedit ~/.ssh/config

Edit your .ssh/config file to include the below information.

Host amiga
HostName <ip address>
Port 22
User amiga
StrictHostKeyChecking no
  • <ip address> can be found on the bottom right of the Amiga Brain screen
  • Remove the < & > symbols.
  • Your development station and your Amiga should be on the same network.

This utilizes the IP address displayed on the bottom right of the brain display:

Let’s copy our keys to the robot as well (you may need to do this any time you are working with a new robot)

ssh-copy-id amiga

Now you are all set up to ssh into the Amiga

SSH Amiga​

ssh amiga
# When prompted for a user name and password enter the following :
# password: amiga
exit # If you want to exit the ssh

If, after a docker image update, you cannot ssh into your amiga or run the script, you may have an error similar to:

# $ ssh amiga
Someone could be eavesdropping on you right now (man-in-the-middle attack)!
It is also possible that a host key has just been changed.
The fingerprint for the ECDSA key sent by the remote host is
Please contact your system administrator.
Add correct host key in /home/<YOUR_NAME>/.ssh/known_hosts to get rid of this message.
Offending ECDSA key in /home/<YOUR_NAME>/.ssh/known_hosts:29
remove with:
ssh-keygen -f "/home/<YOUR_NAME>/.ssh/known_hosts" -R "192.168.XX.XX"
ECDSA host key for 192.168.XX.XX has changed and you have requested strict checking.
Host key verification failed.

Follow the instructions they provide and run:

# With correct values for <YOUR_NAME> and IP Address XX.XX
ssh-keygen -f "/home/<YOUR_NAME>/.ssh/known_hosts" -R "192.168.XX.XX"

You should now be able to ssh into your amiga and/or run the script

Develop and test in the robot​

In order to test your app in the robot, we require to push the changes to the remote repository created before and clone it to the robot.

We recommend to follow the next steps:


Make sure you are on the Linux tab when you go to the GitHub docs (links below). The page defaults to the system you are on and doesn't know you are SSH'd into a Linux machine, which the Amiga is.

  1. ssh to robot ssh amiga (if have not configured ssh do so here)
  2. Create an ssh key for github on the amiga ssh-keygen
  3. Copy and add the key to your github user
  4. Set your username in git config
git config --global "YOUR_EMAIL"
git config --global "YOUR-USERNAME"
  1. Clone your repository
cd apps/
# Replace 'YOUR-USERNAME/YOUR-REPOSITORY' below with the correct link to your new repo
# You can get this path from the SSH tab on the green '<> code' button of your new repo
git clone

If you cannot clone with SSH, you can try https:

# You can get this path from the HTTPS tab on the green '<> code' button of your new repo
git clone
  1. Run the app on your Amiga

The first time you launch will take a while, as the requirements have to install.

If your brain is on the launcher screen, your new app should appear! You can launch it by clicking your the app on the screen, and it should begin loading.

Alternatively, you can launch from the command line:

./ # install and run

If you launch your app with the command line, it is currently possible to have touch interactions with the launcher behind. This will cause other installed apps to unexpectedly launch over the app you are trying to use.

Please see instructions for a workaround under: FAQ - Brain App Development