ROS-based Predictive Target Robot Position Surrounding System Design And Implementation

With the rapid development of artificial intelligence and robotics,the multi-robot roundup system has more and more scientific research and engineering application significance.However,relevant research on multi-robot roundups has mainly focused on the effectiveness of roundup strategies in simulation environments,which has low practical applicability,and development approaches based on real environments are usually not open source and cannot be extended in terms of functionality.At the same time,as the roundup behavior often performs passive roundup due to the lack of a priori information,resulting in its low round-up efficiency.In addition,the driving path of the roundup robot is crucial to its roundup effect,and the path planning algorithm used needs to be reasonably configured with the actual situation.In this paper,a multi-robot system based on ROS platform for rounding up by predicting the location of target robot is implemented to address the above problems,in which the main work and contributions are as follows.To enhance the practical value and secondary development of the multi-robot roundup system,this paper uses the open-source ROS distributed software framework platform,combined with hardware devices such as pioneer robots,lidar sensors,external master control and servers,to build a multi-robot roundup system.The prediction service interface conducts information interaction and control with the robot in the form of subscription and publication,which reduces the power consumption of the robot while strengthening the computational capacity of the system.The robots complete escape and round-up tasks through sensors and external master control.In order to avoid passive roundup behavior and achieve interception-type roundup,this paper uses LSTM prediction model to train and learn the trajectory position of the target robot,which can effectively predict the location information of the target robot at the next moment.The predicted value is used to assign roundup points and drive the roundup robots in the predicted outcome directions.The experiments verify that multiple robots for predictive roundup can have a better roundup route,which reduce the roundup time,and improve the roundup efficiency of the robots.In order to realize the roundup behavior of the roundup robots to reach the target roundup points,this paper adopts the global path planning algorithm based on A~*algorithm and the local path planning algorithm based on DWA,and the important parameters are analyzed and reasonably configured.Through simulations and experiments,it is shown that path planning in a static environment and obstacle avoidance in a dynamic environment can be carried out effectively and the round-up task can be completed reliably.