Useful and Example Controllers developed for general starling usage

This folder contains a number of simple example general controllers for general usage of the starling system.

Building and Installation

Through Docker-Compose

Building

The examples can be built and run via docker-compose.

First, start up one of the docker-compose simulator stacks in ProjectStarling (i.e. docker-compose.yml), Murmuration or FenswoodScenario.

docker-compose -f <simulator/sitl/mavros compose file> up

Then, run the docker-compose.<firmware>.yaml file in this repository to use the examples

# For PX4:
docker-compose -f docker-compose.px4.yaml up
# For Ardupilot:
docker-compose -f docker-compose.ap.yaml up

This command will both build and run the local files (the compose file uses the 'build' entry to specify the location of building). If you make any local changes to the controller, the controller can be rebuilt by using the --build flag, e.g.

docker-compose -f docker-compose.px4.yaml up --build

Running and Networking

Instructions for networking differs on Windows and Linux, as well as which simulator/sitl/mavros scenario you have run. Each might require setting of some environment variables beforehand.

1. Windows - By default the compose files will attempt to connect to a custom Docker Network. The network name is created by the simulator compose file which must be run beforehand. Unless specifically changed, this will default to: <folder_which_contains_compose_file>_default, i.e:

• Projectstarling (default): projectstarling_default
• FenswoodScenario: fenswoodscenario_default
• Mumuration: px4_default or ap_default

If no STARLING_NETWORK environment variable is specified, it will default to projectstarling_default

Using WSL2 (all one command):

STARLING_NETWORK='<Name of compose network>'; docker-compose -f docker-compose.ap.yaml up

Using Powershell:

$env:STARLING_NETWORK = '<Name of compose network>';
docker-compose -f docker-compose.ap.yaml up

# Then when you are finished using these environment variables
Remove-Item Env:\STARLING_NETWORK

2. Linux - If running with the linux variant of the simulator compose files, you will not need to make any changes. By default the file already uses host networking. Whilst this fails on windows, therefore dropping to use the default network, on linux this succeeds.

docker-compose -f docker-compose.ap.yaml up

Through Makefile

This folder can be built locally by running make all from this folder. Each element can also be built separately by invoking make <controller-name>.

These are then available on the system under the tag uobflightlabstarling/example_controller_python_ap (or _px4) fo general running with docker run.

Kubernetes

To run the controller in kubernetes, k3s must be up and running and the images built via the Makefile. Then simply apply the k8 deployment script in the px4/ap folders, e.g.

kubectl apply -f example_controller_python_ap/kubernetes.yaml
kubectl apply -f example_controller_python_px4/kubernetes.yaml

This will run the local uobflightlabstarling/example_controller_python_ap/_px4 if it has successfuly built. The image will need to be pushed to a repository for use on a different machine in the cluster

Fenswood Scenario Example Controller

An example controller is provided which attempts to solve the Fenswood Scenario target landing problem in FenswoodScenario.

To run the fenswood ardupilot example controller, first start up the fenswood scenario simulation compose file for your particular operating system.

Then, the fenswood compose file will start the example fenswood controller.

docker-compose -f docker-compose.fenswood.ap.yaml up

The controller simply follows a trajectory, specified by a set of hardcoded points within the controller.

Example Python Controller

This project contains an example of a controller for the system written in Python with ROS2. It has been packaged up with a Dockerfile and an example kubernetes deployment file. The controller itself directly talks to mavros and tells the drone to lift off, trace a semi circle and land.

For greater details on how it works, refer to README_controller.md.

Examples of Controllers For Use in Starling

Example Python Controller

This is the controller within this repository

Starling Simple Offboard

This is an example of an onboard or offboard controller which can be used to control a single drone. It is based on the original Clover simple offboard, but also includes a trajectory follower module. This controller can be run to provide a higher level control of a real or simulated drone.

Note If running in the flight arena, this controller is already running on the Coex Clover drones

Starling Allocator

This is an example of a trajectory allocator node which attempt to smartly allocate trajectories to drones which can run them.