Design and development of automatic weed detection and smart herbicide sprayer robot

The conventional way of killing weeds in a crop plantation is to spray herbicides throughout the plantation. This results in contamination of the food crops and also the yield becomes less as some of the crop plants die along with the weeds. Thus there is a need for a smart weed control system. In this paper, an image processing algorithm is used to take images of the plantation rows at regular intervals and upon identifying the weeds in the image, the herbicide is sprayed directly and only on the weeds. The algorithm predominantly uses an Erosion and Dilation approach to detect weeds. The colour image is converted to binary by extracting the green parts of the image. The amount of white pixels present in the region of interest is determined and regions with higher white pixel count than the predefined threshold are considered as weeds. The herbicide is stored in a container fitted with water pump motors attached to spray nozzles. Once the weeds are identified, a signal is sent from Raspberry-Pi to the motor driver IC controlling the water pump motors to spray the chemicals over the weeds.