NASA Logo

NTRS

NTRS - NASA Technical Reports Server

Back to Results
Sensory Motor Coordination in RobonautAs a participant of the year 2000 NASA Summer Faculty Fellowship Program, I worked with the engineers of the Dexterous Robotics Laboratory at NASA Johnson Space Center on the Robonaut project. The Robonaut is an articulated torso with two dexterous arms, left and right five-fingered hands, and a head with cameras mounted on an articulated neck. This advanced space robot, now driven only teleoperatively using VR gloves, sensors and helmets, is to be upgraded to a thinking system that can find, interact with and assist humans autonomously, allowing the Crew to work with Robonaut as a (junior) member of their team. Thus, the work performed this summer was toward the goal of enabling Robonaut to operate autonomously as an intelligent assistant to astronauts. Our underlying hypothesis is that a robot can develop intelligence if it learns a set of basic behaviors (i.e., reflexes - actions tightly coupled to sensing) and through experience learns how to sequence these to solve problems or to accomplish higher-level tasks. We describe our approach to the automatic acquisition of basic behaviors as learning sensory-motor coordination (SMC). Although research in the ontogenesis of animals development from the time of conception) supports the approach of learning SMC as the foundation for intelligent, autonomous behavior, we do not know whether it will prove viable for the development of autonomy in robots. The first step in testing the hypothesis is to determine if SMC can be learned by the robot. To do this, we have taken advantage of Robonaut's teleoperated control system. When a person teleoperates Robonaut, the person's own SMC causes the robot to act purposefully. If the sensory signals that the robot detects during teleoperation are recorded over several repetitions of the same task, it should be possible through signal analysis to identify the sensory-motor couplings that accompany purposeful motion. In this report, reasons for suspecting SMC as the basis for intelligent behavior will be reviewed. A robot control system for autonomous behavior that uses learned SMC will be proposed. Techniques for the extraction of salient parameters from sensory and motor data will be discussed. Experiments with Robonaut will be discussed and preliminary data presented.
Document ID
20030064070
Document Type
Other
Authors
Peters, Richard Alan, II (Vanderbilt Univ. Nashville, TN, United States)
Date Acquired
September 7, 2013
Publication Date
March 1, 2003
Publication Information
Publication: National Aeronautics and Space Administration (NASA)/American Society of Engineering Education (ASEE) Summer Faculty Fellowship Program - 2000
Subject Category
Cybernetics, Artificial Intelligence and Robotics
Funding Number(s)
CONTRACT_GRANT: NAG9-867
Distribution Limits
Public
Copyright
Work of the US Gov. Public Use Permitted.

Available Downloads

NameType 20030064070.pdf STI

Related Records

IDRelationTitle20030064055Analytic PrimaryNational Aeronautics and Space Administration (NASA)/American Society of Engineering Education (ASEE) Summer Faculty Fellowship Program - 2000