一种基于滤波的分布式任务分配算法

尽管同步模式协同算法允许团队在许多通用网络拓扑中达成一致状态感知并执行任务分配，但收敛到一致状态感知需要花费大量时间，传输大量数据。在低带宽环境中这会导致严重的潜在问题，增加求解任务方案时间。分析了现存协同拍卖算法例如、CBBA，HIPC等的优缺点，并对HIPC进行改进。采用将滤波框架引入分布式协同拍卖的方法，在建立任务束阶段采用滤波算法对局部状态感知进行估计。打破了HIPC对局部场景感知的完美假设，使得新算法可以在局部感知不可靠条件下良好运行，该算法可以异步进行并行任务分配和冲突分解。通过试验得出相比于HIPC，新算法减少了任务冲突次数，缩短任务执行时间。

A filter embedded distributed task allocation algorith

Although consensus algorithms allow a fleet of vehicles to converge on the SA and perform an assignment over many generic network topologies,convergence to a consistent SA may take a significant amount of time and can often require transmitting large amounts of data to do so.This can cause severe latency in low-bandwidth environments and can substantially increase the time it takes to find an assignment for the fleet.This dissertation discuss advantages and disadvantages of extant distributed cooperative algorithms such as CBBA,HIPC.then extends HIPC by introducing filter into HIPC framework.This dissertation replaces TAA （task assignment algorithm） with FETA （filtered-embeded task assignment） during task bundle building phase of HIPC in order to estimate neighbours state information,which breaks the assumption of perfect local situation awareness and ensure the new algorithm running well under uncertainty.the new algorithm asynchronously assign task and resolve conflicts in parallel.Simulation result demonstrates when compared with HIPC,new algorithm can decrease task conflict times, shorten task execute time obviously.