Drones and AI for sustainable grassland farming

Grassland makes up around 40 percent of the Earth’s land mass; the livelihood of around two billion people worldwide is directly dependent it. Grassland serves as pasture for livestock, performs important ecosystem functions ssuch as carbon storage and guarantees food security for a growing world population.  

Digitalized and automated recognition of plant species and analysis of crop growth assists not only in grassland management but also in validating funding applications for environmental initiatives. UAS (Unmanned Aircraft Systems) technology shows great potential here: a variety of camera and sensor technologies collect high-resolution spatial and temporal information in the visible and infrared spectrum. With the aid of machine learning and artificial intelligence, the 2D image material and 3D models can be used to establish species composition in area of grassland as well as the quantity and quality of the vegetation to be found there. 

Such information is a prerequisite for precision farming and can therefore help to reduce the total quantity of fertilizers and pesticides consumed and to optimize pasture management.

Machine learning and AI algorithms are employed in recognizing and localizing plant species from the combined drone images and sensor data supplied. Individual plants or groups of the same species can be identified and localized down to the nearest centimeter, allowing maps to be drawn showing the distribution of plant species within the field that has been surveyed from above.

Plant recognition is of practical value for agriculturally relevant species, such as the distribution of nutritious forage plants, which are indicators of high-quality grassland. It is also useful in establishing the quantity and distribution of harmful or poisonous plants which can reduce the quality of the forage or even be injurious to the health of livestock grazing there.

Species recognition is of great environmental importance for monitoring the biodiversity of grassland, regardless of whether it is managed “intensively” or “extensively”. The extensive management of grassland (i.e. the sparing use of fertilizers, with mowing or grazing taking place only once or twice a year) promotes plant biodiversity, which also has a positive effect on the insect and bird population. The new organic regulations of the EU Common Agricultural Policy (CAP), which came into force at the beginning of 2023, allow farmers to receive support for maintaining or even increasing their acreage of extensively managed grassland. The Smart Farming department at Fraunhofer IGD is using AI algorithms to train individual species recognition models that not only recognize the so-called indicator species for species-rich grassland, but can also localize their distribution and, in some cases, individual plants with centimeter precision. In the near future, this application will also be certified for the substantiation of state funding applications.

Other applications served by the Smart Farming department in the area of drone-based visual computing are biomass estimation and forage quality monitoring in grassland, disease detection and assessment of damage to arable land and grassland caused by wildlife and adverse weather conditions, as well as monitoring carbon sequestration during rewetting of peatland.

Specialized drones are able to perform seed spreading and fertilizer spraying in areas that are not easily accessible due to overgrown vegetation or waterlogged soil. The needs-based use of fertilizers and spot spraying (i.e. the selective spraying of small areas or individual plants) is a valuable attribute, which allows large-scale savings on plant protection products with associated economic and environmental benefits. The vegetation is analyzed by means of visual computing on the basis of the image and sensor data collected by the survey drone. Spraying/spreading maps are then created and processed at the Integrated Graphics Device (IGD) for subsequent upload to the agricultural drone.