Distributed Optimal Transport for the Deployment of Swarms
Distributed Optimal Transport for the Deployment of Swarms
复制标题
集群部署的分布式最优传输
DOI：
10.1109/cdc.2018.8619816
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
S. Martínez
中科院分区：
文献类型：
--
作者：
Vishaal Krishnan;S. Martínez
可扩展分布式空间部署算法的分析与设计是多机器人系统研究领域的一个重要问题。对于非常大的群体，这可以通过对群体行为的宏观目标来规定，并通过代理之间的本地感知和通信来完成。在本文中，我们解决的问题，分布式的最优运输，以最小化部署大群的成本为目标。工作与宏观PDE模型的连续性方程给出的群集，我们首先制定了一个一般的部署目标，并制定部署算法收敛梯度流。然后，我们设计和分析了一种新的基于拉普拉斯的分布式算法和相应的加权梯度流的最优运输。最后，我们用模拟来说明我们的结果。
The analysis and design of scalable distributed algorithms for spatial deployment is an important problem in the area of multi-robot systems. For very large swarms, this can be prescribed via macroscopic objectives on the behavior of the swarm, and accomplished by local sensing and communication between agents. In this paper, we address the problem of distributed optimal transport, with the aim of minimizing the cost of deployment of large swarms. Working with a macroscopic PDE model of swarms given by the continuity equation, we first formulate a general deployment objective, and formulate deployment algorithms as convergent gradient flows. Then, we design and analyze a novel Laplacian-based distributed algorithm and a corresponding weighted gradient flow for optimal transport. We conclude the manuscript with simulations that illustrate our results.