The Robotics Institute

RI | Seminar | December 9

Robotics Institute Special Seminar, December 9
Time and Place | Seminar Abstract | Speaker Biography | Speaker Appointments

Achieving Intelligent Real-Time Control

 

David D. Musliner

Senior Principal Researcher

Honeywell Laboratories

 

 

Time and Place

NSH 1507
Talk 2:00 pm

Abstract

 

Most research into applying AI techniques to real-time control problems has limited the power of AI methods or embedded ``reactivity'' in an AI system.  Dr. Musliner will present an alternative approach that uses separate AI and real-time subsystems to address the problems for which each is designed; a structured interface allows the subsystems to communicate without compromising their respective performance goals.  By reasoning about its own performance and resource restrictions, the Cooperative Intelligent Real-time Control Architecture (CIRCA) can guarantee that it will meet hard deadlines while still using unpredictable AI methods.  Dr. Musliner will describe in detail the model of agent/environment interactions that CIRCA uses to build its real-time control plans, the methods used to prove that these control plans meet domain requirements, and the architectural mechanisms required to support guaranteed behavior.  With its abilities to guarantee or trade off the timeliness, precision, confidence, and completeness of its output, CIRCA provides more flexible performance than previous systems.

 

Speaker Biography

 

David J. Musliner received his B.S.E. degree with high honors in Electrical Engineering and Computer Science from Princeton University in 1988.  While at Princeton, he received the Charles Ira Young Award for excellence in electrical engineering research; his independent research project developed a highly-successful neural network system for handwriting recognition.

 

Dr. Musliner received his Ph.D. in Computer Science from the University of Michigan in 1993.  His graduate work was supported by fellowships from the National Science Foundation and the Machine Intelligence Foundation at the University of Michigan.  For his dissertation, Dr. Musliner designed and implemented the Cooperative Intelligent Real-Time Control Architecture (CIRCA), one of the first AI control architectures capable of reasoning about and interacting with dynamic, hard real-time domains.  CIRCA was demonstrated in both simulated and real-world robotic applications.  Dr. Musliner's thesis resulted in three journal articles as well as a funded NSF grant that paid for two subsequent graduate students.  CIRCA remains an area of active research at Michigan, with two faculty members and three graduate students currently funded on subcontracts to Dr. Musliner at Honeywell.

 

After receiving his doctorate, Dr. Musliner spent a year and a half at the University of Maryland as a postdoctoral fellow, lecturer, and Project Director of the Autonomous Mobile Robotics Laboratory.  While at Maryland, Dr. Musliner developed novel low-cost sonar multiplexing techniques, applied CIRCA to autonomous mobile robot control, and directed the development of an autonomous hovercraft.

 

In 1995, Dr. Musliner joined Honeywell Laboratories, where he is currently a Staff Research Scientist in the Automated Reasoning group.  Dr. Musliner leads a research thrust based on CIRCA, emphasizing real-time planning and guaranteed intelligent control.  The CIRCA thrust has had several projects funded by DARPA to investigate extensions and applications of the architecture.  TheDARPA/ITO MASA-CIRCA project extended CIRCA to multi-agent applications, with explicit real-time negotiation and coordination between CIRCA agents controlling simulated unmanned aerial vehicles. The DARPA/ISO CIRCADIA project applied CIRCA concepts to computer security, developing real-time reactive control systems to defend against scripted cyber-attacks. The DARPA/IPTO CORTEX project is combining CIRCADIA with the Scyllarus cyber situation assessment system to provide closed-loop cyber security control.

 

Dr. Musliner also leads a large DARPA/IPTO Coordinators project to develop multi-agent coordination assistants for large teams of human performers.  In addition, Dr. Musliner is co-leading a DARPA/IXO HURT project on controlling multiple unmanned vehicles in urban environments, bringing advanced NASA autonomous systems technologies to DARPA applications.

 

 

In earlier work, Dr. Musliner has made major technical contributions to a broad range of HTC projects including reactive planning and plan execution for large-scale petroleum refineries, the multi-agent SARA system for assisting military search and rescue operations, design of networked tool support for integrated constraint-based system design, and distributed scheduling for satellite data processing.

 

 

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 IEEE, AAAI, Agents, NSF, and NASA.  He has organized workshops, tutorials, and symposia for ICRA, AAAI, and AIPS.  Dr. Musliner is a member of Sigma Xi, the scientific research honor society, as well as AAAI and Tau Beta Pi.

 

Speaker Appointments

For appointments, please contact Kristen Schrauder

Related Material
  • Presentation (ppt)

 

The Robotics Institute is part of the School of Computer Science, Carnegie Mellon University.