NASA SBIR 2004 Solicitation


PROPOSAL NUMBER: 04 X7.03-9393
SUBTOPIC TITLE: Mission Training Systems
PROPOSAL TITLE: Adaptive Distributed Environment for Procedure Training

SMALL BUSINESS CONCERN (Name, E-mail, Mail Address, City/State/Zip, Phone)
Stottler Henke Associates, Inc.
951 Mariner's Island Blvd. Suite 360
San Mateo, CA 94404-1560

PRINCIPAL INVESTIGATOR/PROJECT MANAGER (Name, E-mail, Mail Address, City/State/Zip, Phone)
Dr. Eric A. Domeshek
280 Broadway, 1st Floor
Arlington, MA 02474-5311

With its constantly evolving portfolio of highly technical systems requiring human construction maintenance and operation, NASA has an extreme form of a common yet challenging training problem: how to ensure that personnel are qualified on the (often changing) procedures required to work on or with these systems. Simulation-based training that enables learning while doing is a proven approach, but dependence on hardware-based simulators and the requirement for human instructors to develop and supervise training scenarios raise costs and limit flexibility in delivering training and retraining. We propose to build a distributable intelligent tutoring system (ITS) exploiting a unified representation of human and robotic mission activities that can be used to (1) trace student activity to assess, prompt, and correct their actions, (2) simulate robotic activity, (3) control training scenario generation/selection, (4) cover both general and specific cases, (5) allow for varying degrees of detail in human and robotic activity, (6) support extended scenarios involving multiple procedures, and (7) track detailed re-training requirements resulting from changes in procedures. The innovative merger of general procedure descriptions with specific scenario scripts will facilitate more efficient authoring of consistent broad-coverage automated simulation-based training while retaining the ability to author specific scenarios when needed.

Projected future missions will require controllers and astronauts to interact with more capable (semi) automated and robotic systems. Current training practices, tied as they are to scripted scenarios and human instructors, are too costly, inflexible, and inefficient in the face of escalating training needs: more systems, greater complexity, and evolving procedures. The next rounds of lunar exploration will likely involve robotic vehicles with greater potential for autonomy, and controllers will need to learn to supervise these systems in varied ways depending on factors such as immediate mission, environmental setting and expected hazards, latest operational procedures and system health limitations, etc.

The modern world is full of organizations that must construct, maintain, and operate highly complex technical systems, and which in consequence must be able to train their workforce to follow prescribed procedures. Training and retraining are a constant burden as workers move between jobs, and as procedures change to reflect changes in the system, mission, context, operating regimen, or regulations. Industries with heavy reliance on production, distribution, transportation, and information processing systems are all examples of potential consumers of the technology to be developed here.