Research On Cooperation And Ad-hoc Network Algorithm Of Heterogeneous Robot Swarms In Uncertain Scenarios

Heterogeneous robot swarms are widely applied in my fields such as search,rescue,and surveillance,and usually communicate in the form of ad hoc networks.However,the changing topology,limited communication environment and uncertain task scenarios have brought great challenges to the design of robot cooperation and networking algorithm.In addition,with the growth of the communication demand of the task,the contradiction of this problem is further increased.Aiming at the above problems,this paper focus on the workflow cooperation in uncertain environment and the task cooperative communication of multi heterogeneous robots.Firstly,aiming at the influence of uncertain task scenarios,a workflow consistency detection and coordination algorithm based on distributed auction algorithm for multi-robot(WCDC)is proposed.In the task coordination phase,workflow constraints are transformed into time constraints.With the help of broad learning system,the execution time of task sequence is predicted and estimated.Based on this prediction,the task bidding strategy and consistency checking mechanism are designed.On this basis,the subnet coordination and neighborhood coordination methods are proposed to realize the allocation and maintenance of task workflow under local communication.Through the comparison of simulation experiments,it is verified that WCDC can effectively improve the completion rate of tasks,and has stable performance in communication energy consumption,average waiting time and average scheduling distance.Secondly,facing the task collaboration problem of heterogeneous robot swarms,this paper has proposed a task cooperative communication algorithm of robot swarms based on improved contract network protocol(CA-CNP).The improved contract network protocol realizes the conflict free task allocation scheme.The communication consumption in the task announcement stage is reduced,and the parallel coordination of multiple tasks is allowed.The lightweight training of deep neural network with few samples is realized through meta learning architecture.With the help of the trained network model,the success rate of the contract is predicted,assist the announcement robot to reduce the generation of invalid contracts and improve the efficiency of task allocation.Through the simulation tool NS3,the performance of the CA-CNP is compared with various combined auction algorithms.The experimental results show that the CA-CNP has excellent performance in task completion rate and communication energy consumption.