Lockheed Martin Robotics Seminar
Cognitive Robotics and Human Robot Interaction
Greg Trafton, Ph.D
Head, Intelligent Systems Section
Naval Research Laboratory
Today's robotic systems need to be able to interact with, deal with, and understand people. How do we build computational and robotic systems that don't just ignore the person, but actually take the person into account? I will discuss a series of systems that extract information from people, build models about how people think and reason, and then test the efficacy of the models. My primary example will center around building Human-Robot-Interaction systems for firefighting robots.
Dr. Greg Trafton received his BS in computer science with a second major in psychology from Trinity University, San Antonio, TX in 1989. He received an MA (1991) and Ph.D (1994) in cognitive science from Princeton University.
Dr. Trafton has three primary areas of interest: human robot interaction/cognitive robotics, error prediction, and predictive supervisory control systems.
His work in HRI involves creating computational cognitive models that perceive and think the way that people do, putting those models on embodied platforms (Mobile/Dexterous/Social robots), and using those models to increase interaction capabilities between robots and people.
The error prediction work has created theoretical models of why people make procedural errors, creating both process and statistical models of error behavior, and then using those models to predict and prevent errors that people are likely to make.
Dr. Trafton's work on predictive supervisory control systems has allowed him to build a theoretical dynamic system that predicts when an operator becomes overloaded and then provide different types of facilitation to the user to allow the operator to accomplish their tasks.