NASA SBIR 00-II Solicitation

FORM 9B - PROPOSAL SUMMARY


PROPOSAL NUMBER: 07.01-9541 (For NASA Use Only - Chron: 000460 )
PROPOSAL TITLE: Agent-based Optimization of Distributed Air Traffic Management

TECHNICAL ABSTRACT (LIMIT 200 WORDS)
NASA is developing new concepts for Air Traffic Management (ATM) that redistribute flight planning authority across decision-makers (pilots, controllers, and dispatchers) in the National Air Space (NAS). To explore these concepts, new modeling and simulation (M&S) initiatives focus on developing models of the NAS components (vehicles, radars, communications, etc), but it is clear that equal emphasis must be given to modeling the distributed decision-making, via high-fidelity Human Behavior Representations (HBRs). In Phase I, we explored approaches to this, concluding that HBRs need to: 1) represent human capabilities, limitations, and rules of behavior for decision-makers within the NAS; 2) be agent-based and extensible within larger scale simulations; and 3) provide ?hooks? that support exploration of alternative decision-making protocols that drive ATM safety and performance. We developed a concept demonstration that showed how agent-based HBRs, embedded in a medium-fidelity simulator, could be used to optimize ATM performance, via a Genetic Algorithm (GA) procedure that ?evolves? well-formed conflict resolution procedures within the agents. In Phase II, we will extend this prototype, expanding the scope of decision-making and negotiation behaviors, enhancing the ?fitness functions? the GA used for assessing and optimizing performance, and validating the approach in high-fidelity ATM simulations of future NAS operations.

POTENTIAL COMMERCIAL APPLICATIONS
The proposed technology will directly support simulation of the developed agents in the advanced ATM system. These simulations will support the design and specification of rules for negotiation for pilots, air traffic controller and airline dispatchers in the advanced ATM environment of the future. The underlying models of information processing, situation assessment, and distributed decision-making will also support requirements in other domains (e.g. intelligent vehicle highway systems, strategic warfare gaming industry, etc.). We also plan to extract the decision-making and communications components to embed in a generalized Intelligent Agent Toolkit (IAT) configurable to any domain requiring intelligent agent interaction.

NAME AND ADDRESS OF PRINCIPAL INVESTIGATOR (Name, Organization Name, Mail Address, City/State/Zip)
Karen Harper
Charles River Analytics Inc.
725 Concord Ave
Cambridge , MA   02138 - 1040

NAME AND ADDRESS OF OFFEROR (Firm Name, Mail Address, City/State/Zip)
Charles River Analytics Inc.
725 Concord Ave
Cambridge , MA   02138 - 1040


Form Printed on 11-26-01 17:18