Research On Foraging Behavior Of Swarm Robots Based On Pheromone Communication

There are social organisms such as birds,fish,ants,bees and so on in nature.In social biota,the ability and wisdom of individual members are limited,but the whole group shows strong ability and extremely high intelligence,not only the simple linear superposition of individual ability.Individuals emerge complex group intelligence and group behavior through social interaction,resulting in self-organizing behavior beyond the individual.Swarm robots focuse on the emergence of complex self-organizing behaviors from simple behavior rules for swarm robots based on a biologically inspired approach.Aiming at the control theory of swarm robots system,this thesis focuse on foraging behavior,introduces pheromone communication mechanism into swarm robots foraging system by imitating the process of ants foraging,and upgrades the traditional pheromone model.A pheromone model based on recurrent neural network is proposed,which improves the overall efficiency of foraging system by improving the traffic environment.At the same time,the principle of division of labor and cooperation is introduced into the foraging system of swarm robots,and the foraging task is divided into two sub-tasks,search and transport,and the probability of performing sub-tasks can be dynamically changed by the interaction between robots,and the foraging task can be adjusted according to different environmental conditions by self-organization to adapt to the changes of the environment.The research results in this thesis theoretically deepen the understanding of the emergence mechanism of self-organizing behavior of swarm robots,which has important scientific significance and theoretical value for building a larger-scale robot cluster systemFirstly,starting from the development prospects and significance of swarm intelligence and swarm robots,this thesis expounds the purpose of this research,and introduces the research trends of swarm intelligence algorithm and swarm robots technology at domestic and abroad.Then,aiming at the shortcomings of traditional pheromone model,inspired by recurrent neural network,a pheromone model based on recurrent neural network is proposed.The simulation experiment is carried out in the same experimental environment on the simulation platform ARGoS,which verifies the validity of the model.The results are compared with those of traditional pheromone models,which proves that the pheromone model based on recurrent neural network can actually improve the foraging efficiency of swarm robots.Finally,aiming at the problem of reducing the foraging efficiency caused by traffic congestion in the experimental site,a task partition strategy is proposed by using the mechanism of division of labor and cooperation.The overall efficiency is improved by reducing the length of invalid paths.The simulation experiment is carried out on ARGoS simulation platform,which proves that the task partition strategy can improve the performance of the foraging system of swarm robots.