Research And Design Of Auction-Based Multi-Agent Task Allocation In Storage

There are many requirements in the storage areas,such as expanding warehouse scale,increasing cargo types and reducing hours.As the increasing scale of warehouse,the complexity of distribution and cargo is increasing gradually.It is a feasible solution that taking multi-agent system to improve warehouse management efficiency,which has practical significance to accelerate the automation production process.Multi-agent intelligent warehouse based on robots can improve its operational efficiency.One of the key issues is rational task allocation,and the dynamic task assignment is the focus of current research.A basic study is given on dynamic task assignment of multi isomorphic agent system in this paper.When the multi-agent system is introduced into the warehouse management,a question arises that which robot should be sent to carry out new tasks.This problem requires the task assignment algorithm for judging.First of all,the current research progress and difficulties of MAS theoretically,as well as MATA,are summarized.Then summarizing the common task allocation methods,market based auction method is used as a research method combined with the storage application environment.Considering the characteristics of the task allocation of MAS,the multivalent auction algorithm is selected as the basis for further research.Then,based on the general theory of auction,the basic model,the necessary process and the performance evaluation criteria of the algorithm are build up.Through the formal language,it is carried out that describing the storage system including task description,the implementation of the cost and benefits and so on.The performance index of the task allocation algorithm is represented by the formula.A model of MATA is presented.Further,a task allocation method of MAS is proposed,which is called task-grouped task allocation auction based.Static and dynamic task allocation strategies are designed by auction algorithm.Taking into account the characteristics of warehouse tasks,a new task-grouped approach is proposed,which is based on the concept of “the global tasks are grouped first and redistributed later ”,improving the traditional static allocation algorithm is applied to the dynamic task in warehouse,where dynamic tasks can be processed in the form of relatively static.The method that gets near optimal assignments enhances the flexibility of the algorithm.The experiment platform is set up by mixed programming of MATLAB and VS2010,realizing strategy calculation and entity control in storage.The task-grouped auction algorithm is implemented by MATLAB.Controlling of multi-Amigobot is realized by VS2010.Finally,the simulation of MATA in the specific storage mode is provided.Experimental results indicate that the method reduces the total time of the global task allocation,improves the performance of the global task path equilibrium,and the complexity become lower.The algorithm is simulated and the verification experiment is carried out on Amigobot system,and simulation results represent that the algorithm is reasonable and effective.