| In recent years,as one of the important branches in the field of artificial intelligence,multiagent system has developed rapidly with the support of intelligent robot technology and network communication technology.Many theoretical research results of multi-agent system are often verified with the help of simulation environment,which is lack of actual multi-agent experimental platform.In this context,based on the self built multi-agent system platform,this thesis studies the algorithm and physical verification of the consensus of multi-agent system.The main research results are as follows:(1)The structural framework of the multi-agent experimental platform and the workflow of the whole platform are designed.Through hardware selection and software design,the software and hardware foundation of multi-agent platform is established,the two wheel differential robot is selected as the agent object for research,and the relevant model files are compiled in gazebo through model analysis.(2)A multi-agent experimental platform based on ROS is built.According to the requirements of multi-agent system for positioning,the positioning is completed through lidar mapping and IMU.According to the topic communication and publish subscribe mechanism of ROS,the distributed communication between agents is realized by creating handles to send and receive messages.(3)When the multi-agent system can verify the algorithm,the obstacle avoidance is designed for the possible obstacles on the forward path.Real time image acquisition by camera and obstacle avoidance by computer vision,including image processing and decision-making of agent behavior.(4)The model of single agent is analyzed,and the distributed containment control algorithm of multi-agent is studied,based on the pilot following method,a distributed control protocol for containment control is proposed,and the design method of core parameters is given.Finally,the algorithm is applied to the multi-agent experimental platform to verify the containment control algorithm. |
