Autonomous weeding

Conventional agricultural systems are strongly dependent on herbicides to suppress the infestation of weeds. A more sustainable method would be to use an automated plant-specific weeding system. One of the bottlenecks in the development of a plant-specific weeding machine is the lack of a robust and versatile plant identification and localisation system.
In my research I am using state of the art deep-learning technology to develop a smart camera system that can identify, localize and map crops and weeds. The output of our camera system can be used by a robot to provide plant specific care. Furthermore the data produced by our camera system can be used on farm level by the farmer to adapt his farm management to the need of the crops.
With the help of an accuracy plant identification and localization system, the environmental impact of current agricultural practices can be reduced while maintaining a high yield.