Secure Telerobotics

Abstract:

Telerobotic systems, such as those used in rescue operations, remotely-operated vehicles or the next-generation robotic surgery, human operators interact with robots through a communi- cation network. It is anticipated that the next generation of teleoperated systems will combine existing communication networks with temporary ad-hoc wireless and satellite networks, to send audio, video and other sensory information. This project is developing tools to prevent security threats in telerobotics, by monitoring and detecting malicious activities and correcting for them. To develop tools to prevent and mitgate security threats against telerobotic systems, this project adapts cybersecurity methods and extends them to cyber-physical systems. Knowledge about physical constraints and interactions between the cyber- and physical components of the system are leveraged for security. A monitoring system is being developed which collects operator commands and robot feedback information to perform real-time verification of the operator. Timely and reliable detection of any discrepancy between real and spoofed operator movements enables quick detection of adversarial activities. Methods to enhance safety are also being implemented. The results are evaluated using the UW-developed (NSF-supported) RAVEN surgical robot, under a variety of different conditions and threat scenarios. This project brings together research in robotics, computer and network security, control theory and machine learning, in order to gain better understanding of complex teleoperated robotic systems and to engineer telerobotic systems that provide strict safety, security and privacy guarantees. The results are relevant and applicable to a wide range of applications, including telerobotic surgery, search and rescue missions, military operations, underwater infrastructure and repair, cleanup and repair in hazardous environments, mining, as well as manipulation/inspections of objects in low earth orbit.The project algorithms, software and hardware will be made available to the non-profit cyber-physical research community. In the medium term, this project is expected to dramati- cally impact the community of robotics researchers. In particular, researchers involving surgical robotics and search and rescue missions will be able to incorporate our results into their work. In the longer term, this project will also benefit a larger cyber-physical community by broadening our knowledge about the interconnection between the cyber and the physical components of the system.As a part of this project, graduate and undergraduate students are being trained in cyber- physical systems security topics, and K-12, community college students and under-represented minority students are being engaged.