The fusion of data with high spatial and temporal resolution and their interpretation are essential innovation drivers for the realisation of more sustainable processes in plant cultivation. Here, economic potentials go hand in hand with ecological improvements such as resource savings, selective processes and the integration of flowering regions in plant stands and mixed crops. Different environmental conditions, such as the growth stages of the plants, soil properties, the emergence of weeds and weeds by-products, as well as weather conditions and machine influences, have a direct impact on the correctness and availability of sensor data and their interpretation with regard to a semantic environment perception.
AI-based algorithms offer the opportunity to develop robust sensor systems that generate valid and safe operations in a wide range of environmental conditions. The generation of reproducible test scenarios under different environmental conditions is essential for development of these systems. Therefore, an outdoor test environment is being build up in AI-Test-Field. The aim is to carry out autonomous long-term tests to generate sensor data under variable field, weather and plant conditions. In addition to the sensor signals, meta-data will be recorded for the development and evaluation of AI methods.
The project includes the use of different sensor systems (laser scanner, stereo cameras, ToF camera, ultrasonic and radar) as well as the exemplary transfer to real machines for different use cases (row cropping, grassland and bare ground). The contents of AI-Test-Field form an essential basis for the certification of such sensor systems for autonomous field robotics.
Hochschule Osnabrück, LEMKEN GmbH & Co. KG, Maschinenfabrik Bernard Krone GmbH & Co. KG, IFM