Research On Distributed Multi-robot Task Allocation Based On Hoplites Framework

In the field of industry,service,medical and search and rescue,robots have broad application prospects.Robots can replace humans to accomplish some boring,dangerous tasks and even tasks that humans can't finish.Although the capabilities of robots are already very powerful,there are still many complex tasks and scenarios that a single robot can't do or can only finish extremely inefficient.At this time,multiple robots are needed to perform tasks together.When multiple robots work together,it is necessary to cooperate each other to complete all tasks reasonably.Therefore,an efficient and robust multi-robot task allocation method is crucial.The multi-robot task allocation method is mainly divided into the centralized method,the distributed method,and the hybrid method.The centralized method has the disadvantage of low fault tolerance.And the most widely used now is the distributed multi-robot task allocation method.The distributed method is highly fault-tolerant and the dependence between the robots is small.The Hoplites framework is a distributed multi-robot task allocation framework.This paper first studies the fault tolerance of the Hoplites framework.After many experiments,it is found that fault tolerance is crucial for multi-robot teams.If the taskallocation algorithm is not fault-tolerant,it may lead to the entire team fail to execute tasks.Therefore,this paper designs a multi-robot dynamic priority queue rotation mechanism and a timeout re-planning mechanism,which greatly improves the fault tolerance of the multi-robot team.Secondly,this paper designs a threshold self-calibration mechanism and a threshold sharing mechanism based on the Hoplites framework to improve the self-adaptability of the Hoplites framework in solving multi-robot task allocation problems.According to the experimental experience,the switch between the passive coordination mechanism and the active coordination mechanism of the Hoplites framework is mainly realized by a threshold.In the past,the threshold is determined according to experience or through a large number of experiments.Therefore,this paper designs the threshold self-calibration and sharing mechanism,which gives the robots the ability to adjust themselves,so that the robots can autonomously correct the threshold during the task execution,and achieve a better understanding of the environment.Finally,this paper studies the task allocation problem of multiple unmanned underwater vehicles.Because of the special underwater working environment,the limitation of communication range among unmanned underwater vehicles has to be considered.Many of the previous work on multi-robot task allocation was done based on perfect communication,but in practice it is almost impossible.Therefore,based on the actual situation,this paper designs a communication-maintaining profit and communication relay task node based on the Hoplites framework,which ensures that in the underwater environment,multiple unmanned underwater vehicles can still maintain close contact with certain communication range restrictions and collaborate to complete all the tasks.In general,based on the Hoplites framework,this paper studies the fault tolerance,self-adaptability and communication range limitations of multi-robot distributed task allocation problem considering from the practical application,gives corresponding solutions,and verifies the effectiveness and advantages of the solutions through a lot of experiments.