Computer vision for perception robotic in simulated industrial environment for optimization of logistics

Abstract

This work aims to implement image processing and computer vision on a mobile robot, the Robotino FESTO ® plataform, for executing actions that complement the tasks described in the Robocup Logistics League competition, including identifying objects by color and shapes. The robot's structure includes a Logitech Webcam c250 camera, which will be used for image acquisition, for recognition the environment by robo, mainly to identify objects and assisting in the navigation task. The objects to be identified play the role of machines in a production line, raw material to be processed and the final product. These objects have specific shapes and combinations of colors. In particular, the machines have three colored lights and each combination represents a status of being recognized by the robot. Studies and research on filters techniques in images allow the student understands Open CV library functions, which will be the basis for implementation the application. This will be done in Robotino View development environment, which provides graphical language, C / C + + and Lua Script. Tests will be performed in environments and similar situations to the competition and the results will be analyzed according to desirable features such as: (I) correct visual identification of patterns and color combinations, (II) correct identification and visual formats (III) harmony between the visual processing and other modules responsible for additional tasks such as navigation and priorities. The subroutines developed will be adapted and integrated in a global solution for competition goal, under developing by Robo Samba team in Natel research group, at Unesp/Sorocaba. The complete system should run integrated into the ROS (Robot Operating System). Thus, the latest activities of this project include investigations into how to adapt and integrate the programs developed in said system.