NASA SBIR 2004 Solicitation


PROPOSAL NUMBER: 04 X7.02-9175
SUBTOPIC TITLE: Intelligent Onboard Systems
PROPOSAL TITLE: Seamless Mode Switching for Shared Control of Semiautonomous Vehicles

SMALL BUSINESS CONCERN (Name, E-mail, Mail Address, City/State/Zip, Phone)
Intelligent Automation, Inc.
15400 Calhoun Drive, Suite 400
Rockville, MD 20855-2785

PRINCIPAL INVESTIGATOR/PROJECT MANAGER (Name, E-mail, Mail Address, City/State/Zip, Phone)
Donald Myers
15400 Calhoun Drive, Suite 400
Rockville, MD 20855-2785

Whether it be a crew station, the Shuttle Remote Manipulator System (SRMS), an unmanned ground rover (UGV) or air vehicle (UAV), or teams thereof, the controllers for such systems will be complex, multilevel, usually distributed, systems. When a human user desires to switch between automatic and manual control, the transition must occur at all levels of the controller. There exist no well-developed strategies for managing such transitions and no proven methods for guaranteeing overall stability in the classical control-theoretic sense, or even safety and reliability in the general sense. These type of issues will span virtually every shared-control application in future NASA exploration systems. Intelligent Automation, Inc. proposes to use a two-level, distributed robot controller with multimodal user interface (UI) and demonstrate a technique to seamlessly transition between teleoperation and autonomous operation. The technique is based on using Hidden Markov Models to identify the current active state at each level of the controller. The demonstration platform was developed for a previous NASA project for JSC to develop automatic programming methods for astronaut assistants.

As NASA extends its reach, it will be necessary to use unmanned vehicles to facilitate autonomous exploration and to support telepresence for human personnel. Since at such great distance, human interaction will be limited, these vehicles must learn both on their own and from the limited interaction when human assistance is possible. Learning from human demonstration can be used to automatically "program" robots for space missions during the earth-based training periods when humans are also learning how to use the robots. Beyond robots, as the functions of all mission subsystems become increasingly complex, forms of adaptive learning will be necessary.

These technologies will be applicable to commercial avionics systems, unmaneed ground systems, and control and decision-making systems. The technology developed will provide greater integration at the system level, more affordable configurations, more efficient and supportable control architectures, and the ability to operate air vehicles safely and effectively in an inter-netted environment. All commercial aircraft manufacturers, suppliers, and airline could also benefit from this technology.