PROJECT TITLE: Agent-based Hierarchical Task Network Planning and Resource Managemnt
TECHNICAL ABSTRACT (LIMIT 200 WORDS)
This document proposes the development of Agent-Based Planning, a new approach to planning for autonomous control systems that addresses fundamental limitations of classical AI techniques. The current state of the art in autonomous planning technology cannot be used in applications where the control system must react to change in the world within minutes or seconds. We propose a component based architecture that supports planning and re-planning concurrently with plan execution. These systems quickly find solutions to planning problems, and then improve these solutions as time allows. This planning architecture provides a foundation for incorporating application-specific planning modules and competencies, which in turn supports integration of capabilities into more and more complex systems. Specifically, we propose to develop a prototype planning and control system for routine satellite operations, which has the potential to save tens of millions of dollars annually in maintenance costs. To the extent feasible, our approach will be generic, isolating fundamental issues in resource management that must be addressed in order to make this technology viable. If successful, this approach to planning will significantly enhance our ability to engineer and build robust and reliable automatic planning systems in a wide variety of application domains.
POTENTIAL COMMERCIAL APPLICATIONS
The immediate commercial potential of this research will be to greatly reduce the cost of routine satellite operations and maintenance. This technology also has the potential to increase system utilization, especially in response to unanticipated events. This technology also has commercial potential in a broad spectrum of indirect applications that Intelligent Automation, Inc. is involved in, including factory automation, distribution route planning, and transportation logistics.
NAME AND ADDRESS OF PRINCIPAL INVESTIGATOR
Robert Kohout
Intelligent Automation, Inc.
2 Research Place, Suite 202
Rockville , MD 20850
NAME AND ADDRESS OF OFFEROR
Intelligent Automation, Inc.
2 Research Place, Suite 202
Rockville , MD 20850