Computational vision and the intelligent connection between perception and action in robotics

Abstract

The use of robotics in industrial logistics provides for increased efficiency in the production line, reduction of operating times and costs, increasing the productivity and accuracy of previously performed processes manually. FIESP defines Logistics as the process of efficiently planning, implementing and controlling the flow and storage of raw materials and stock during the manufacture of finished products, and the information related to these activities, from the point of origin to the place of consumption, in order to meet customer requirements. The role of mobile robots in industrial logistics refers to the automatic execution of tasks in the activities of factory floor where mobile robots collaborate in the execution of tasks that involve the flow and storage of raw materials, as well as in the process of manufacturing of products finished and stock. It is, therefore, a complex challenge since providing robots with such level of autonomy requires solutions that mix different techniques of Artificial Intelligence within a modular architecture. Computer processes have inputs and outputs, analogously an agent in action (autonomous mobile robot) has perception (inputs) and performs actions (outputs). Developing methods of intelligent connection between perception and action requires the existence of processes that define the characteristics of both areas. In this context, it is proposed in this project to consider results of works previously developed by students of the Control and Automation Engineering course of Unesp Sorocaba, in the form of Scientific Initiation projects supported by FAPESP that make up the SINPLI (Intelligent Industrial Production and Logistics System) in the areas of Image Processing, Path Planning, Task Planning and Task Manager, the latter module being responsible for the control of all tasks to be performed by the robot. In the context of SINPLI, this scientific initiation project aims to implement a software module that allows the connection between the perception of the robot, through the analysis of the environment through its different types of sensing, especially the computer vision and processing of images, and certain actions to be performed by the robot in the domain. The developed module will be integrated with SINPLI, making it possible to define scenarios and problem situations to validate the developed module. (AU)