Autonomous Formation and Operation of Fractionated Spacecraft

这项研究是对新太空时代对可扩展、分布式架构的迫切需求的回应，该架构可以整合多个小型航天器，称为分块航天器，每个航天器具有有限但互补的能力，以便共同执行复杂任务，如地球观测、在轨服务和深空探测。这种空间基础设施的一个关键促成因素是航天器组队和操作的自主权。最根本的问题是：“如何在组织和功能上组建一支小型化的异构航天器团队，使它们能够在现实世界的条件下协同自主地执行集体任务，这些条件包括信息不完整、感官噪声、环境扰动以及硬件和通信故障。”该研究承诺，通过使团队能够a)根据可用资源动态调整任务计划，使由分块航天器组成的空间任务实现完全自主；B)根据更新后的计划重新配置飞行编队；c)在非结构化和不可预测的环境中，在集体（个人功能）、合作（任务重新分配和编队重新配置）和协作（资源和知识共享）三个层面上提高其在任务期间的性能。它将采用由PI为移动机器人团队开发的基于自由控制概念的主机-代理-化身组织架构，并定义用于分项操作的各种软件代理，包括任务规划、任务分配和编队飞行，作为独立于安装在航天器平台不同模块（化身）上的机载处理器（主机）的分布式马尔可夫决策过程。这样的框架不仅允许分布式处理和网络以及可扩展的特别团队组成，而且还为使用健壮的并发马尔可夫决策过程的故障恢复和统一的组学习机制的开发铺平了适当的基础。功能架构将基于并行工程方法设计，通过对最先进的小型化航天器模块进行系统的功能评估，并通过蒙特卡罗分析考虑其可靠性。作为功能架构的关键组成部分，该研究还将开发基于分散虚拟结构的分布式控制方案，用于具有有限处理和通信以及欠驱动能力的小型航天器的临时小组的自主编队重新配置（包括轨道和姿态）。为了验证研究成果，将设计一个分块航天器基础设施，用于在不同轨道上部署各种有效载荷进行地球观测。这项研究将培养2名硕士。2名博士生和1名博士后，将从不同的候选人中招聘。

This research is a response to the urgent need of the new space era for scalable, distributed architectures that can consolidate multiple small space vehicles, dubbed as fractionated spacecraft, each with limited but complementary capabilities, in order to perform complex missions collectively, such as Earth observation, on-orbit servicing and deep space exploration. A critical enabler for such space infrastructures is autonomy in the formation and operation of teams of spacecraft. The fundamental question is: “how to form, organizationally and functionally, a team of miniaturized heterogeneous spacecraft enabling them to synergistically carry out a collective task autonomously, under real-world conditions such as incomplete information, sensory noise, environment perturbations, and hardware and communication malfunctions.” The research promises to make full autonomy a reality for space missions composed of fractionated spacecraft, through enabling the team to a) dynamically adapt the mission plan to the available resources; b) reconfigure its flight formation given the updated plan; and c) enhance its performance during the mission at the three levels of collective (individual functionality), cooperative (task reallocation and formation reconfiguration), and collaborative (resource and knowledge sharing), in unstructured and unpredictable environments. It will adopt the Host-Agent-Avatar organizational architecture, based on the notion of Control ad Libitum, developed by the PI for mobile robot teams, and define various software agents for the fractionated operation, including mission planning, task allocation, and formation flying, as distributed Markov decision processes independently from the onboard processors (hosts) mounted on different modules (avatars) of spacecraft platforms. Such a framework would not only allow for distributed processing and networking as well as scalable ad hoc team formation, but also pave a proper ground for the development of failure recovery and unified group learning mechanisms using robust concurrent Markov decision processes. The functional architecture will be designed based on a concurrent engineering methodology, through a systematic functional assessment of state-of-the-art modules for miniaturized spacecraft, considering their reliability by means of Monte Carlo analyses. As a key constituent of the functional architecture, the research will also develop a distributed control scheme, based on a decentralized virtual structure, for autonomous formation reconfiguration (both orbit and attitude) of an ad hoc team of small spacecraft with limited processing and communication and underactuated capabilities. To validate the research outcomes, a fractionated spacecraft infrastructure will be designed for deploying various payloads in different orbits for Earth observations. The research will train 2 M.A.Sc. and 2 Ph.D. students as well as 1 postdoctoral fellow, to be recruited from a diverse pool of candidates.