Research in cooperative multi-agent robotic systems (MARS) has received wide attention in recent years since this cooperation allows the system to perform complex tasks from relatively simple individual behaviors. One key problem to solve from the control engineer’s point of view in this setup is the achievement and maintenance of the formation of the multi-agent system, i.e., to drive multiple agents to achieve prescribed constraints on their states. In this regard, Robotic Park is an experimental platform that supports MARS experiences with different types of vehicles (aerial and ground robots). The architecture of Robotic Park allows working in a virtual environment, in a physical environment, or in a hybrid scheme. All the elements of the system are integrated into a ROS 2 network. Different aspects such as the performance of the local controllers, the number of communication messages, or the CPU usage are evaluated in the proposed benchmark.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

A Benchmark on Formation Control of Multi-agent Robotic System

  • Francisco José Mañas-Álvarez,
  • María Guinaldo,
  • Raquel Dormido

摘要

Research in cooperative multi-agent robotic systems (MARS) has received wide attention in recent years since this cooperation allows the system to perform complex tasks from relatively simple individual behaviors. One key problem to solve from the control engineer’s point of view in this setup is the achievement and maintenance of the formation of the multi-agent system, i.e., to drive multiple agents to achieve prescribed constraints on their states. In this regard, Robotic Park is an experimental platform that supports MARS experiences with different types of vehicles (aerial and ground robots). The architecture of Robotic Park allows working in a virtual environment, in a physical environment, or in a hybrid scheme. All the elements of the system are integrated into a ROS 2 network. Different aspects such as the performance of the local controllers, the number of communication messages, or the CPU usage are evaluated in the proposed benchmark.