Scalable Randomized Scheduling of Mobile Sensors with Observability Guarantees

自主的空中、地面和水下机器人已经成为无数传感应用的有前途的平台。空中无人机可以监测龙卷风和森林火灾等自然灾害，水下机器人可以检测有害的藻华并跟踪入侵鱼类。要收集的信息通常由一些非线性动态过程控制。一个自然的问题是，给定有限数量的移动的传感器，它们应该如何动态放置以最好地观察感兴趣的量，特别是在有限的能量下，同时要求机器人保持连接？可能的传感器放置空间是巨大的，能源和连接的限制增加了挑战。本计画将发展有效的演算法，以在这些限制下调度移动的感测器，并借由实地实验评估其追踪移动目标的效能。这项研究的跨学科性质将与推广和教育活动相结合，以扩大K-12和本科生的参与，特别是来自代表性不足的群体。该项目结合了研究人员在控制，网络理论和快速随机计算方面的互补专业知识，因为该项目将产生一个新的视角和一个可推广的，具有可观测性保证的移动的传感器的可扩展调度的原则框架。该项目的目标将通过四个综合的研究重点，跨越理论研究，算法开发和实验验证来实现。推力1的重点是集成一个基于Gramian的非线性可观测性度量与随机采样的有效计算接近最佳的传感器放置下感知和通信的不确定性。推力2扩展了框架，以适应能源和连通性的限制，其中特别强调分布式计算方法。受鱼类跟踪应用的启发，Thrusts 1和2中开发的理论和算法将通过一组跟踪移动声学标签的自主水面车辆进行实验验证。该项目的第三个目标涉及实验测试平台的开发，包括机器人及其相关模型和控制器，而第四个目标则通过模拟和希金斯湖的现场实验来评估所开发的移动的传感器调度算法，该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查进行评估，被认为值得支持的搜索.

Autonomous aerial, ground, and underwater robots have emerged as promising platforms for a myriad of sensing applications. Aerial drones can monitor natural calamities like tornadoes and forest fires, and underwater robots can detect harmful algal blooms and track invasive fish species. The information to be gathered is typically governed by some nonlinear dynamic processes. A natural question to ask is, given a limited number of mobile sensors, how should they be dynamically placed to best observe the quantity of interest, especially with limited energy while requiring the robots to remain connected? The space of possible sensor placements is vast and constraints on energy and connectivity add to the challenges. This project will develop efficient algorithms to schedule the mobile sensors under these constraints and evaluate their performance in tracking a moving target through field experiments. The interdisciplinary nature of this research will be integrated with outreach and educational activities to broaden participation of K-12 and undergraduate students, especially from underrepresented groups.The project combines the investigators’ complementary expertise in control, network theory and fast randomized computation as the project will result in a fresh perspective and a generalizable, principled framework for scalable scheduling of mobile sensors with observability guarantees. The project goals will be realized through four integrated research thrusts that span theoretical investigation, algorithmic development, and experimental validation. Thrust 1 focuses on integrating a Gramian-based nonlinear observability metric with randomized sampling for efficient computation of near-optimal sensor placements under sensing and communication uncertainty. Thrust 2 extends the framework to accommodate energy and connectivity constraints, where special emphasis will be on distributed approaches for computation. Motivated by the fish-tracking application, the theory and algorithms developed in Thrusts 1 and 2 will be validated experimentally with a fleet of autonomous surface vehicles tracking a moving acoustic tag. Thrust 3 of the project involves the development of the experimental testbed, including the robots and their associated models and controllers, while Thrust 4 evaluates the developed mobile sensor scheduling algorithms via both simulation and field experiments in Higgins Lake, Michigan.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.