Safely incorporating drones into airspace

There is an urgent need for clearly defined rules to govern the growing number of drones in German airspace. No-fly zones are key to ensuring the safe and equitable integration of these unmanned aircraft. Highly precise geodata can help to automatically define and indicate these prohibited areas. With this in mind, the fAIRport project was launched in May 2020—with Fraunhofer IGD providing its expertise in object recognition, geodata visualization, and artificial intelligence.

The future of drones is bright and multifaceted: They can replace hazardous helicopter monitoring and maintenance flights along high-voltage transmission lines and gas pipelines. They can accelerate the delivery of medications, organs, and blood. And their deployment for search-and-rescue missions can save lives. Moreover, drones can help protect the environment by significantly reducing CO₂ emissions. Yet turning these possibilities into practical solutions will require safely integrating these unmanned flying machines into Germany’s highly regulated airspace: Even when pilots have no line of sight to their drones, they must ensure their devices do not enter prohibited areas.
 

Using artificial intelligence to demarcate no-fly zones

DFS, Germany’s air navigation service organization, plans to provide drone pilots with high-quality geodata indicating no-fly zones via the UTM (Universal Transverse Mercator) system for managing unmanned aviation—operated by its subsidiary Droniq. These zones include, for example, wind turbines, rail and road networks, industrial facilities, and locations where large numbers of people can be expected, such as campsites. To this end, Fraunhofer IGD is developing a method based on artificial intelligence that interprets high-definition satellite images—specifically, attributes that have not been cartographically processed, but which represent areas drones are not permitted to enter. Machine vision and learning methods are capable of reliably finding and accurately categorizing the corresponding structures, and can add further relevant information. Artificial neural networks are employed to recognize 3D objects via 2D satellite images. “For instance, we are now able to determine exactly how high a wind turbine is by the shadow it casts,” explains project leader Mohamad Albadawi. Going forward, he and his team will teach the neural networks to perform further tasks, such as detecting additional important features, e.g., helicopter landing pads.
 

Comprehensive data make it possible to safely fly drones

The project goal is to make a geodata platform based on open standards available by 2023 and operated by wetransform, a Fraunhofer IGD spin-off. Via a dedicated interface, local government agencies will be able to add further information, for example, on mass gatherings such as markets or concerts, to create temporary no-fly zones. The platform will also continuously monitor and update existing flight-relevant data. This not only forms the basis for managing drone flights but also contributes to better airspace planning, helping to make general aviation safer. The Federal German Ministry of Transport and Digital Infrastructure (BMVI) has provided the fAIRport project with 1,205,000 euros in funding within the scope of the mFUND research initiative.