David Musliner

Staff
dmusliner at sift dot net

Ph.D., Computer Science and Engineering, University of Michigan, Ann Arbor, MI, 1993

B.S.E., Electrical Engineering & Computer Science, Princeton University, Princeton, NJ, 1988

During more than twenty years of leadership in intelligent systems research, Dr. Musliner has developed extensive experience with planning and control for autonomous systems, including space, air, and ground vehicles.  In addition, Dr. Musliner has led major research thrusts in formal verification including statistical model checking, hybrid systems verification and controller synthesis, as well as autonomous cyber security.

For his Ph.D. dissertation, Dr. Musliner designed and implemented the Cooperative Intelligent Real- Time Control Architecture (CIRCA), one of the first AI planning and control architectures capable of reasoning about and interacting with dynamic, hard real-time domains. CIRCA is designed to make rigorous safety guarantees about its plans, using formal verification methods to ensure that its plans avoid predicted failures while also achieving system goals.

In collaboration with other current SIFT researchers, Dr. Musliner incorporated formal model checking methods into CIRCA. Taking advantage of the unique aspects of CIRCA's real-time plans, this research team developed a novel model checking algorithm that improved system scalability by two orders of magnitude and led to a U.S. patent on incremental verification.  Dr. Musliner also led the extension of CIRCA to reasoning about probabilistic systems, developing the first statistical model checking systems for use in verifying CIRCA's automatically-designed controllers.  Most recently, Dr. Musliner has led efforts to apply CIRCA to control autonomous satellites and federated assets for space situation awareness.

Dr. Musliner has pursued more than a decade of work on autonomous cyber security.  Recently he led the DARPA-funded team building SIFT's Fuzzbuster system, a unified plug-in architecture for automatically identifying, characterizing, and repairing or shielding software vulnerabilities.   This work is currently being transitioned to the Navy Cyber Warfare Development Group. Teamed with Prof. Stephen McCamant at the University of Minnesota, Dr. Musliner currently leads SIFT's FuzzBOMB team entry in the DARPA Cyber Grand Challenge.

In earlier work, Dr. Musliner led the DARPA-funded CIRCADIA project, which applied CIRCA concepts to computer security, developing real-time reactive control systems to defend against high-speed cyber-attacks.  Dr. Musliner then led the DARPA-funded CORTEX project to demonstrate an early version of the Fuzzbuster concept, including active experimentation to characterize vulnerabilities and automatically shield them.  This led to two computer security patents.  The CORTEX project was demonstrated successfully numerous times and underwent live red-team testing twice, meeting all of its performance metrics and demonstrating the feasibility of active self-adaptive defense.

In collaboration with other current SIFT researchers (Michael Pelican and Eric Engstrom), Dr. Musliner led the development of the planning and control software for the DARPA-funded Heterogeneous Unmanned Reconnaissance Team (HURT) project. For HURT, up to six small UAVs were flown fully autonomously and sent specialized RSTA imagery directly to ground soldiers, at their request. The HURT team spent many weeks in field testing with Marine infantry, conducting live-flight exercises in urban operations using the UAVs and soldier-carried RSTA display systems. In the latest demonstration, HURT coordinated an even wider range of assets including a manned C-12 aircraft flying at 6000 feet, unmanned Hunter, Shadow, and ScanEagle vehicles flying at 2000 feet, and smaller Pointer, Raven, and Wasp vehicles flying as low as 100 feet over the combat zone.

Dr. Musliner also designed and developed a distributed airspace management system for NASA’s AATT advanced air traffic technology program, demonstrating the value of Contract-Net based approaches to air route planning. In earlier work, Dr. Musliner led the development of Playbook-based planning and control software for commanding teams of small indoor robots, providing ISR imagery and security in support of urban small combat unit operations during the DARPA Tactical Mobile Robots program. Dr. Musliner also led a DARPA-funded Coordinators project developing distributed task coordination software to help optimize human teams in the performance of dynamic, inter-dependent tasks such as combat operations or searchand- rescue.

Dr. Musliner has published extensively in journals and conferences including AI Journal, IEEE Computer, IEEE Intelligent Systems, IEEE Systems, Man, and Cybernetics, AAAI, and numerous symposia and workshops. Dr. Musliner participates on program committees and review boards for NASA, DARPA ISAT, IEEE, AAAI, and NSF.