A Stigmergy-Based Swarm Robots Target Search And Trapping Framework

As a classical but difficult problem,multi-target searching and entrapping in a swarm of robots have received more and more attention in recent years.The capabilities of individual robots are often considered limited,multi-robot systems(MRSs)can cooperate to complete complex tasks through local interaction.Compared with the single robot system,MRSs has the advantages of strong adaptability,strong expansibility,and high reliability.Like the swarm phenomenon in biological groups in nature,swarm robots can generate and transform the pattern of the whole cluster to realize various tasks.At present,swarm robot system has been widely used in a variety of realistic scenarios,and control algorithms of multiple kinds of swarm robots have emerged in an endless stream.However,most existing approaches for addressing this problem rely on unrealistic assumptions such as reliable communication links,available global coordinate system,known environmental information,and central coordination among robots.However,most of these approaches do not get good performance in real application environments.Therefore,this paper proposes a stigmergy mechanism-based framework for searching and entrapping targets in unknown and communication-restricted environments.Inspired by cell mechanism and biological collective behaviors,the framework improves the traditional inverse ant colony system and combines it with the Radial Basis Implicit Function(RBIF)based Hierarchical Gene Regulatory Network(H-GRN),which is mediated by pheromones,to realize the collaboration of swarm robots.Specifically,the framework is divided into three stages: target searching stage,target tracking stage,and target entrapping stage.In the first stage,the robots use their sensors to perceive the environment and search for targets,and then release pheromones during movement according to the improved inverse ant colony system.Each robot communicates indirectly through the pheromones remaining in the environment.In the second stage,some of the robots that find the target will track the target and release pheromones,while the rest of the robots will gather around the target along the pheromones.In the third stage,after the swarm of robots gather around the target,the pheromone information in the environment will be used as input to the RBIF model to automatically generate a flexible entrapping pattern.Finally,a directional vector will be generated based on the improved lower layer of H-GRN model to guide the robots into the entrapping patterns.Simulation experiments on different scenarios show that the swarm robots can quickly find the target in the unknown environment and collaborate to entrap the targets,which proves the better performance of the proposed framework compared with the traditional methods.