Construcción de un prototipo robótico vinculado con visión artificial para la asistencia en la eliminación de maleza en cultivos de fresa

Abstract

This project describes the construction of a robotic prototype to assist in the elimination of weeds in strawberry crops based on the use of artificial vision techniques in order to contribute to the agricultural sector in Chimborazo province. The analysis of the case study was performed, based on this, the robotic prototype was modeled with the use of SolidWorks software; the result was a mobile wheeled robot with a classic tricycle design that has the ability to mobilize it through the strawberry crop with solvency. The static analysis performed on the base structure with the material used, resulted that when applying a force of 500 N, there is a deformation of 0.000193 mm, this deformation is imperceptible and corroborates that the modeling is correct and supports the loads to which it was subjected. The control and image processing system was carried out by a computer that was responsible for communication with the prototype using SPI and UDP (User Datagram Protocol) protocols. The programming code of the machine vision system for path-path recognition and weed identification was done using the Python and Arduino IDE (Integrated Development Environment) programming language. The results obtained from the functionality tests indicate that the robotic prototype has a weed detection efficiency of 94.13% and a removal efficiency of 90.65%. This data is an indication that the prototype fulfils the functions for which it was built. The speed of image processing was on average 0.039 s, this time proves that there is a rapid processing that guaranteed the correct functioning of the machine vision system. It is recommended to use personal protective equipment while the prototype is performing cutting functions.