Resilient Control of Human Controlled Robotic Networks

这项建议的研究目标是调查人类控制的机器人网络对网络攻击的弹性。这些系统在机器人系统和通信网络的帮助下，扩展了人类远程操纵物体的能力。一些应用领域，如危险环境中的作业、海底勘探、机器人手术、钻井等，都受到了这项技术的重大影响。然而，使用通信网络可能会导致中间人或恶意软件攻击，从而严重危及这些系统的安全和性能。在这些观察的推动下，这项提议旨在调查导致安全和弹性人类控制的机器人网络的算法、工具和实现。从根本上说，拟议中的工作将广泛影响控制系统和机器人领域的网络安全。从应用的角度来看，拟议的研究将影响远程外科手术、空间机器人和其他半自主远程操纵系统的领域。一名研究生将直接受益于拟议的研究，预计数百名本科生和研究生将受益于控制理论、机器人和网络安全方面丰富的跨学科教育。这项拟议的为期三年的研究工作预计将对安全和有弹性的半自主系统的设计和控制方式产生重大影响。这将通过研究用于人类控制的机器人系统中的边缘和节点攻击检测的分布式和计算轻量级算法来实现。此外，利用和改进对这些系统中的串通攻击者的基本界限的计算将有助于理解对上述攻击的可实现恢复能力。拟议工作的具体目标包括调查用于边缘和节点攻击检测和识别的分析工具和基于物理的加密技术，通过利用混合物理-虚拟机器人团队来发展人类控制的机器人网络的复原力，以及在由不同类型的机器人系统组成的试验台上进行实验研究，以实现对实现漏洞的理解。

The research objective of this proposal is to investigate resilience of human controlled robotic networks to cyber attacks. These systems extend the human capability to manipulate objects remotely with the help of robotic systems and communication networks. Several application areas such as operations in hazardous environments, undersea exploration, robotic surgery, drilling, etc. are significantly impacted by this technology. However, the use of communication networks can lead to man-in-the-middle or malware attacks that can severely compromise the safety and performance of these systems. Motivated by these observations, this proposal is aimed at investigating algorithms, tools, and implementations that lead to secure and resilient human controlled robotic networks. From a fundamental standpoint, the proposed work will broadly impact the field of cyber security in control systems and robotics. From an applications perspective, the proposed research will impact the field of telesurgery, space robotics, and other semiautonomous telemanipulation systems. One graduate student will directly benefit from the proposed research, and it is expected that several hundred undergraduate and graduate students will benefit from the enriched interdisciplinary education in control theory, robotics, and cyber security. The proposed three-year research effort is expected to significantly impact the manner in which secure and resilient semiautonomous systems are designed and controlled. This will be realized by investigating distributed and computationally light algorithms for edge and node attack detection in human controlled robotic systems. Additionally, utilizing and refining the computation of fundamental bounds to colluding attackers in these systems will aid in understanding achievable resilience to the aforementioned attacks. Specific objectives of the proposed work include investigating analysis tools and physics based encryption techniques for edge and node attack detection and identification, developing resilience in human controlled robotic networks by utilizing mixed physical-virtual robot teams, and experimental studies on a test bed consisting of heterogeneous robotic systems to achieve an understanding of implementation vulnerabilities.