The AgroRobottiFleet project funded by the Innovation Foundation Denmark to foster autonomous collaborative farming activities with a fleet of agriculture robots. The fleet consists of AgroRobotti robots developed by the project partner Agro Intelligence ApS. The project has a full budget of 23.4MDKK and is supported with 13.4MDKK by the Danish Innovation Foundation. The project brings together expertise from Aarhus University, DTU, DTI, Business Region MidtVest, and Agro Intelligence ApS.

AgroRobottiFleet is divided into 5 interrelated work packages to enable safe, reliable, and autonomous collaborative farming. Each work package focusses on a specific aspect to achieve this challenging goal and enable farmers to perform farming tasks more efficiently

and with less manual labor. The work packages are as follows:

  • WP1: Project management
  • WP2: Ag-robotic cyber-physical scenario and safety simulator
  • WP3: Robotti fleets
  • WP4: Impact evaluation
  • WP5: Robotic fleet testing and validation, demonstration and dissemination


Department of Engineering at Aarhus University is specifically involved in safety evaluations and guarantees during runtime. For this, AU ENG leads WP2 and develops a cyber-physical safety simulator for fleet of agricultural robotics. This simulation environment enables testing of hardware/software in the loop for the overall system. The simulation capability supports the development of fleet control by virtually investigating scenarios before deployment. Numerous scenarios emerge when having multiple robots collaborating, such as route re-planning after a failed robot or new robots joining the fleet. The accompanying robots collaborate to finalize the operation, increasing operational reliability. The simulator further allows for easy configuration of different scenarios, definition of safety constraints, and how to handle adjustable autonomy for the selected robots. The developed models of the physical world for the simulator also utilizes a digital twin setting enabling a centralised control. Individual Robottis may automatically degrade their level of autonomy whenever the difference between the perceived data from the real world and the simulated data becomes too high. The digital twin follows the pre-defined adjustments of the autonomy or entirely switch to manual operation if the given situation has not been considered at all.