DefinePK

Abstract

This article presents the development of a virtual environment for testing industrial-type robotic applications, using artificial intelligence techniques through convolutional neural networks. The developed environment is made up of two RGB-D cameras for capturing information from the workspace in which the robot operates, a conveyor belt and a work table. A ResNet-50 model within a Faster R-CNN is used to detect a tool, in order to indicate to the robot the position of it, to generate a pick and place application, from the table to the conveyor belt, as a method for validating the environment, reaching an average precision of detection of the object of interest greater than 82% and an 84% success rate on the pick and place task.