Desenvolvimento e análise experimental de um sistema robótico para plantio automatizado de vegetais utilizando agricultura de precisão e visão computacional.

Abstract

This work presents the development of a precision agriculture robot capable of managing vegetable, berry, and legume patches within a vegetable crop. It utilizes computer vision, Internet of Things (Iot) cloud-based processing, which, through an online webapp a digital twin coordinates planting activities such as sowing, weed control, and watering in a real vegetable crop raised bed where the physical robot is installed. An analysis of the computer vision methods for automatic weed detection under different conditions and artifacts is presented. According to experimental results, by the experiment scenarios, some computational theoretical improvements are proposed in order to reduce false detection of weeds such artifacts in the raised crop bed.