Mobile-based autonomous system for real-time detection and removal of volunteer potato plants in sugar beet field

A major need for an automated weed control system together with consideration of environmental issues and concerns leeds to develop small and lightweight, but self-guided and fully autonomous, the mobile-based system for efficient control of volunteer potato in the field. The work mainly consists of three procedures including weed detection, weed control, and the integration into a mobile platform with Robot Operating System (ROS).

In 2015, we developed a novel discrimination (weeds and crop discrimination) procedure that performs robust under the natural outdoor conditions. The procedure is based on machine learning approach using counter-intuitive advanced features. The camera was fully exposed to outdoor sunlight conditions without having any covering material that protects the camera from direct and strong sunlight. Thanks to our novel discrimination procedure, varying sunlight and shadows are no longer the challenges for our weed control robot.