CAREER: Mothership/Daughtership Architectures for In Situ Science by Robotic Sensor Networks

这个项目的目标是发展的控制策略，可以利用互补的计算，传感和通信能力的母船/子船机器人传感器网络的成员进行原位体积感测的基本理解。从网络化的无人系统，合作控制，在ad hoc网络，运动规划，和最佳分布式传感控制的移动性的概念相结合，创建一个新的异构机器人团队组成的两类机器人具有显着不同的，但互补的属性。研究活动分为三个主要方面：（一）在强电流存在下使用有序逆风方法进行运动规划的子系统控制和在间歇通信存在下用于协作采样的分散覆盖控制; ii.）基于强化学习的母舰控制算法，以提供延迟容忍网络和随机动态编程，用于子系统的最佳部署和重新分配;以及iii.）在室内机器人传感器网络上进行的实验演示，在室外异构无人驾驶飞机系统上进行的实验演示，以及与大气科学家进行的实地活动的一部分。拟议的研究项目将提供对复杂大气现象进行原位体积采样的新能力，这将直接提高对复杂环境的理解能力。该项目将通过将受益于机器人传感器网络进行现场科学的应用产生最广泛的影响。合作的重点是研究在严重风暴，海冰变化和气团边界的时空演变的龙卷风发生。此外，研究和教育被整合到旨在改善K-12学生对工程领域的看法的活动中，保留那些最初被工程吸引的学生，将动手主动学习带到几个空中机器人课程，并促进科学家和工程研究人员之间的互动。

The objective of this project is to develop fundamental understanding of control strategies that can exploit the complimentary computation, sensing, and communication capabilities of the members of a mothership/daughtership robotic sensor network performing in situ volumetric sensing. Concepts from networked unmanned systems, cooperative control, controlled mobility in ad hoc networks, motion planning, and optimal distributed sensing are combined to create a new heterogeneous robot team comprised of two classes of robots with significantly different, though complementary, attributes. Research activities are organized into three main thrusts: i.) daughtership control using ordered upwind methods for motion planning in the presence of strong currents and decentralized coverage control for cooperative sampling in the presence of intermittent communication; ii.) mothership control algorithms based on reinforcement learning to provide delay-tolerant networking and stochastic dynamic programming for optimal deployment and reassignment of the daughterships; and iii.) experimental demonstration performed on an indoor robotic sensor network, outdoors on a heterogeneous unmanned aircraft system, and part of field campaigns with atmospheric scientists.The proposed research project will provide new capabilities for in situ volumetric sampling of complex atmospheric phenomena which will lead to a direct improvement in the ability to understand complex environments. This project will have its broadest impact through the applications that will benefit from robotic sensor networks performing in situ science. Collaborations focus on the study of tornadogenesis in severe storms, sea ice change, and spatio-temporal evolution of airmass boundaries. Additionally, research and education are integrated into activities aimed at improving the perception of the engineering field by K-12 students, retaining those students who are initially attracted to engineering, bringing hands-on active learning to several aerial robotic courses, and fostering interaction between scientists and engineering researchers.