旋翼无人机协同任务指派问题研究与算法改进

Study and Algorithm Improvement on Collaborative Task Assignment of Rotor UAVs

摘要:

在应用旋翼无人机群时,每架无人机自主快速选择目标空域点,对于提升无人机群整体性能有很大作用。对于到达目标空域点即可降落的情况,可视作二分图最佳匹配问题,利用匈牙利算法进行求解。对于到达空域点后,需要等待其他无人机全部就位后一同工作的情况,需要同时优化平均移动距离和最大移动距离两个耦合参数。经多次实验和测试,将匈牙利算法中代价矩阵各元素值替换为各自值的平方,可以在平均移动距离与最优解十分接近的情况下,使最大移动距离减小,从而减小无人机悬停等待的消耗,缩短过程完成的时间。针对此问题设计了只变异不交叉、父辈子辈一同评价的遗传算法进行求解与改进算法对比,验证了改进后匈牙利算法在此问题中具有更好的效果。由于算法计算速度快,适用于无人机上嵌入式系统实时计算。

Abstract:

In the application of the rotor unmanned aerial vehicle (UAV) cluster, the quick selection of a predetermined point of each UAV plays a great role for improving the overall performance. In the case of reaching the target airspace point and then landing, it can be regarded as the best matching problem of the bipartite graph, which can be solved by the Hungarian algorithm. In the case of reaching the airspace point then waiting for the other UAVs to work together after they are fully in place, it is necessary to optimize both of the coupling parameters: the mean moving distance and the maximum moving distance at the same time. The value of elemental in the cost matrix of the Hungarian algorithm is replaced after experiments and tests, so that the maximum moving distance can be reduced when the average moving distance is very close to the optimal solution, and the time to complete the process can be shortened. Genetic algorithm for this problem was designed to compared with the improved algorithm. It is verified that the improved Hungarian algorithm performs better in this problem. Due to the fast calculation of this algorithm, it is suitable for real-time calculation of embedded system in UAV.