Swoop Aero Leverages AWS to Deploy a Fleet of Aircraft Across the Globe

Every day across the globe, Swoop Aero aircraft fly into hundreds of locations to deliver a range of services for the entire population, including medical transport, emergency management, disaster response, search and rescue, retail delivery, and ecommerce shipping. 

Founded in 2017, Swoop Aero offers an integrated drone logistics service and is approved to fly in 14 countries including Australia, the UK, Malawi, Mozambique, and the Democratic Republic of Congo. One year after its launch, the company became the first company in the world to transport a vaccine by commercial drone in Vanuatu with funding from UNICEF.

“Our goal is to make this service accessible to one billion people by 2030,” says Andrew Thomas, chief architect at Swoop Aero. “To achieve this, we need highly scalable technology to support our platform so we can have aircraft operating at the same time.”

Swoop Aero’s service must reach diverse geographical locations where communications and connectivity can be challenging. “Ultimately, lives depend on the operation of the platform, so we need high performance and reliability,” says Thomas. The business also requires analytics and machine learning to enhance the safety and reliability of its aircraft. “The drones are fully autonomous, which means we must always maintain situational awareness. We also need to use data analytics to monitor systems for constant visibility of each aircraft’s performance status,” Thomas adds.

Swoop Aero chose to launch its platform on Amazon Web Services (AWS) to meet scalability, reliability, and data analytics requirements. “We’ve been on AWS from startup to scale-up,” says Thomas.

The business uses AWS IoT Core to connect aircraft as IoT devices over cellular where available, and satellite communications to provide full global coverage. “AWS IoT Core allows us to connect our aircraft and control them remotely, beyond visual line of sight of the pilot, anywhere in the world,” Thomas explains. Swoop Aero uses the web socket capability in Amazon API Gateway to provide two-way communication between aircraft and remote pilots, who authenticate to Swoop Aero’s Remote Pilot Station using MFA credentials via Amazon Cognito. 

Swoop Aero also leverages computer vision technology at the edge for drones to ascertain the potential to land safely without pilot input. This technology is based on custom machine-learning (ML) models the company trains on Amazon SageMaker and the SageMaker Ground Truth data labelling service. With Amazon SageMaker Ground Truth, Swoop Aero efficiently labels thousands of landing images, continually improving the models’ accuracy and the safety of the system. The company also relies on Amazon SageMaker Neo to optimize its models to run onboard the aircraft in a poor communication environment. 

Additionally, Swoop Aero manages its drone fleets through digital twins based on the AWS IoT Device Shadow service. The service stores a shadow state for each drone and performs automated performance status checks against the twin to ensure the reliability of the aircraft. If a deviation is detected, it automatically logs a defect against the aircraft and the system blocks it from taking off.

By running its platform on AWS, Swoop Aero can scale to support a single remote pilot controlling 30 or more aircraft throughout the world from its Remote Operations Center (ROC) in Melbourne, Australia. Each pilot can fully fly aircraft across four continents beyond visual line of sight, maintaining continual situational awareness.

The business captures relevant data across each flight through on board sensors and control systems, storing this data in Amazon Simple Storage Service (Amazon S3). The data includes flight information, status of individual onboard sensors, and landing images. “With the number of flights we manage, having an individual engineer look at each flight’s data is not scalable,” says Thomas. “AWS helps us leverage the data in a scalable and automated way.” 

Swoop Aero takes advantage of ML and data analytics on AWS to increase the safety and reliability of its aircraft. For example, the company’s computer vision technology, based on AWS, allows pilots to remotely examine the ground before takeoff or landing and ensure there are no obstacles. “Across the aviation industry, redundancy and reliability are key,” says Thomas. “Our computer vision solution on AWS powers remote piloting beyond line of sight and provides insights to ensure a safe landing every time, even in areas with poor-to-zero network coverage.”

In addition, Swoop Aero developed a monitoring system using AWS Lambda to process each flight log, providing insights into the flight while looking for problem areas. “An example might be an aircraft that appears ostensibly fine but generates a bit of vibration,” Thomas explains. “AWS Lambda will identify the vibration and flag it as a defect against the digital twin. This blocks the system and that aircraft from flying again until the problem has been examined.” The company also generates reports from insights based on ML models in flight. Thomas continues, “We've combined ML approaches to track how the lifetime of components differs based on the environment, such as taking off in remote dirt or sandy environments versus launching from concrete."

The Swoop Aero development team has more time to create new features and enhancements for the company’s platform by using AWS. “We knew we didn’t want our team to be managing services, and AWS serverless technologies allow us to focus on developing and scaling our systems rather than spending time supporting infrastructure,” says Thomas. 

To date, Swoop Aero has safely delivered over one million items and counting worldwide. Thomas concludes, “Whatever we create, we know we’ll look at AWS first. AWS integrates seamlessly with our existing services and provides the scalability and reliability our platform requires.”

Learn more about how Swoop Aero reinvented commercial cargo aviation and saves lives.