Implementación de Rutinas en ROS2 para la Aplicación de Algoritmos Multiagentes en el Movimiento de Robots Tipo Soccer Small Size.

Abstract

The present work aims to address the implementation of routines inherent to multi-agent systems through the Robot Operating System in its second version (ROS2) for the STOx’s team; ROS2 facilitates the implementation and development of robotic systems. This is an innovative proposal in response to the various challenges faced by robotics, such as autonomous naviga- tion and efficient collaboration between robots. Therefore, with the support of ROS2, the aim is to improve the coordination and performance of the group of robots in test scenarios, thus contributing to the advancement of this research resource. Contributions of this work include the development of multi-agent behaviors for the STOx’s team robots and a computer vision mechanism that allows the localization of each robot-agent from a side camera perspective instead of the normally used top-down perspective.

The STOx’s team, from the Robotics Group of the Universidad Santo Tomás, developed a group of robots for the Soccer Small Size League (SSL), participating in the RoboCup competition from 2011 to 2017 with very good results. Due to the low usage of the group of robots over a long period of time (from 2017 to 2023), various challenges arise in developing the project. This is because several robotic platforms may be non-operational or may require maintenance. Therefore, it was decided to handle a small number of functional robots, which are ideal for tackling the development of the project due to their versatility and precision. This choice allowed for guaranteeing the project’s feasibility by taking advantage of the available resources and ensuring the validity in the implementation of the proposed algorithms. Initially, a literature review was conducted to obtain models of multi-agent algorithms, as well as tools that would allow real-time and stable detection of the group of robots. Subsequently, different techniques were implemented for the detection and localization of the group of robots,making it possible to test their movement in a delimited environment and to carry out tests satisfactorily. Finally, the functionalities corresponding to the movement of the robots, as well as their detection and localization, were transferred to ROS2. Likewise, the multi-agent models previously tested in simulation were adjusted to verify their functionality in the given environment through the robots, thus carrying out the validation process of the proposal on a physical scenario. As a result, the robots achieved coordinated movements based on the implemented multi-agent algorithms.