How can large areas be surveyed more quickly, more accurately, and more cost-effectively? By working together as a team rather than flying solo. In the “SmartFarmScape” project, a team led by Prof. Dr. Frank Künemund at Fachhochschule Dortmund is researching how autonomously operating drone formations can collect precise aerial data for agriculture and forestry.
Whether for the early detection of bark beetle infestations, the identification of deadwood, or the analysis of drought conditions on agricultural land, the need for high-resolution, large-scale environmental data is growing. To survey large areas particularly efficiently, multiple drones must be deployed simultaneously. This significantly increases the technical challenges. Only when flight altitude, distance, and flight paths are precisely coordinated can the captured image data be optimally merged into an accurate overall map.
This is where the “SmartFarmScape” research project at Fachhochschule Dortmund comes in. Under the direction of Prof. Dr. Frank Künemund, the Faculty of Computer Science, in collaboration with industry partners, has developed an innovative multi-agent system in which multiple drones operate as a coordinated team. The technical core of the system is the decentralized organization of the swarm. Unlike well-known drone light shows in the night sky, where each drone follows a fixed, pre-programmed route, the “SmartFarmScape” drones communicate directly with one another during flight. “There is a so-called ‘leader’ that the other drones follow. They regulate their exact distance and altitude completely autonomously, even when unexpected gusts of wind disrupt the formation,” explains Prof. Künemund.
A wide range of applications
This precise formation flight is not an end in itself, but rather a prerequisite for the subsequent data analysis. Only by flying in a precise formation can comparable, overlapping aerial images be captured, which can then be combined into a comprehensive photo mosaic and analyzed. In addition to agriculture and forestry, the system can also provide important insights into hazardous situations, such as droughts or floods. Its use is also conceivable in disaster response or during official search operations for missing persons. “With our system, we want to provide high-quality data that can be used for specific applications,” said Prof. Künemund.
Since the compact drones themselves do not have the necessary computing power for complex data analysis, the images are transmitted directly to a downstream ground station—such as an all-terrain robot or a mobile vehicle—located at the edge of a field or forest. There, they can be analyzed in near real time using specially trained artificial intelligence (AI).
The project is funded by the Federal Ministry for Economic Affairs and is currently in its final phase. But the researchers at Fachhochschule Dortmund are already thinking ahead. “One goal is to combine our system with freely available satellite data, such as that from the European Sentinel-2 program,” says Prof. Künemund. The idea behind this is an intelligent division of labor: AI first analyzes the large-scale satellite images, filtering out conspicuous changes—such as areas showing signs of extreme drought stress. The Fachhochschule Dortmund drone formation then takes off for a targeted, high-resolution inspection flight on site. The system is scalable to the specific application and allows for the rapid integration of additional drones into the formation.