Intelligent Weeding Robot Using Deep-Learning

Abstract

Weeding is the most challenging process in agriculture. Traditional weeding requires a lot of labor, time and money. So, there is a need to develop an autonomous weeding method that reduces the time and cost of the weeding process without human intervention. Our system is designed to detect weeds using a deep learning model in order to eliminate the weeds using a laser beam, which is an effective weed control method that is echo-friendly and does not harm the environment. The laser weeding precisely targets the weeds and eliminates them without affecting the crop plants. The solution we suggest is a smart machine, based on a mobile application that processes the input images from the camera module to classify the detected objects in the image as weeds or not. After detecting the weed, the center point coordinates of the detected weed are sent to the Arduino microcontroller via the Bluetooth module, then the robot will move in 2-dimensionals to reach the position of the weed in the real environment, and activate the laser on the weed for an amount of time.