Drones4Good

One project, two aims

The two-year project has two main aims. The first is to use a camera installed on a drone to image a larger area than ground-based systems can while flying at speeds of 80 to 140 kilometres per hour. The images will be processed in real time on board the drone. Humanitarian aid workers on the ground will thus receive processed information about damage to buildings, supply routes and people in need who are cut off from their surroundings via a radio link to the ground while the drone is in flight. The extracted geoinformation will then be distributed via the UN’s existing systems; this will ensure that the tools and technology developed by DLR can be evaluated together with the aid organisations involved. This scenario is due to be simulated in autumn 2021 at DLR’s National Experimental Test Centre for Unmanned Aircraft Systems in Cochstedt. One of the planned activities is to fly the equipped drone over mannequins and containers on the test site and transmit an image of the situation in real time to the emergency services on the ground.

The researchers' second primary aim for the project is to develop safe techniques for drones to drop supplies without endangering people and infrastructure below. For example, using specially developed AI technology, delivery drones could automatically detect people on the ground. With this information the remote pilot and drone would be able to assess the safety of the drop and only release the supply package – which can weigh up to 20 kilograms – if the drop site is clear.

Transferring technologies developed at DLR to operational activities for disaster relief will be particularly useful for the humanitarian aid organisations involved in Drones4Good. In the case of Wings for Aid, improving route planning and ensuring automatic access to affected areas are key to facilitating deliveries by transport drones. Being able to execute such tasks remotely is crucial to the organisation's success, but it is also proving a major challenge. For the World Food Programme, cutting down on the time between a disaster occurring and receiving information about the local situation is considered a huge benefit of Drones4Good. Following the Idai and Kenneth cyclones in Mozambique, it took 72 to 120 hours to process image data of the affected areas. The technologies developed at DLR should enable aid supplies to be delivered faster and more effectively to those in need.

Cooperation between DLR institutes

Drones4Good pools the efforts of four DLR institutes and facilities – the Institute of Optical Sensor Systems, the Remote Sensing Technology Institute, the Institute of Flight Systems and the National Test Centre for Unmanned Aircraft Systems. The Modular Aerial Camera System (MACS) developed by the Institute of Optical Sensor Systems has already made it possible to transmit aerial images from commercial drones to the ground in real time. Systems on the ground then merge these image tiles into a coherent image mosaic and overlay it on a digital map. This makes it possible to assess the current situation using digital mapping services such as Google Maps or OpenStreetMap.

The Remote Sensing Technology Institute is drawing upon its research into AI technologies to contribute towards the rapid analysis of aerial and satellite images. This research will help adapt drones for optical reconnaissance, transforming them into a 'pocket satellite'. The primary focus is on identifying people, building damage and roads. Within a short space of time, a digital image of the situation can be generated displaying the areas flown over by the drone; this is immediately available to emergency services in the crisis region. DLR will work with the partner organisations to assess the extent to which such technology and image data can assist international relief efforts. "We welcome direct feedback from our partners, as it helps us to continue to develop our methods and adapt them to specific applications," says Nina Merkle, who leads the Drones4Good project at the DLR Remote Sensing Technology Institute.

AI that can identify people at the drop site using information from aerial images is particularly important for dropping relief supplies using drones, as this dropping poses a danger to individuals on the ground. The Institute of Flight Systems has already conducted initial testing of such technology in the Dominican Republic. "Drones have the potential to perform tasks quicker and in a more targeted way than their alternatives. However, we have to be able to guarantee the highest standards of safety and build trust in the technology," explains Johann Dauer, Head of the Unmanned Aircraft Department at the Institute of Flight Systems. "As part of Drones4Good, a tailored version of DLR's MACS camera system will be integrated into DLR's superARTIS research helicopter. We will use specific hardware and software developments to determine the extent to which it is possible to identify people in aerial images in real time and determine whether relief supplies can be dropped safely."