Acquisition and Use of Internal Models for Human Motor Learning

Abstract

We study decision making in dynamic environments in general, and human motor learning in particular. Our approach focuses on the acquisition and use of libraries of representational primitives. This approach is motivated by computational considerations -- learning new motor plans by linearly combining primitives from a library ameliorates the "curse of dimensionality". It is also motivated by evidence from the field of cognitive neuroscience indicating that biological organisms (including humans) linearly combine motor primitives (known as motor synergies) when planning and executing motor actions. We have made excellent progress showing that linear combinations of "global" primitives can achieve near-optimal performance on tasks requiring the control of a simulated two-joint robot arm. We have also shown that new linear combinations for novel tasks can be learned rapidly. In more recent research, we have explored the strengths of libraries of "local" primitives where primitives are linearly combined using a "local" additive regression procedure.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 17, 2009
Accession Number
ADA574167

Entities

People

  • Robert A. Jacobs

Organizations

  • University of Rochester

Tags

Communities of Interest

  • Autonomy
  • Energy and Power Technologies
  • Human Systems

DTIC Thesaurus Topics

  • Acquisition
  • Algorithms
  • Bayesian Networks
  • Cognitive Neuroscience
  • Cognitive Science
  • Computational Science
  • Computer Simulations
  • Dimensionality Reduction
  • Dynamic Programming
  • Environment
  • Information Processing
  • Learning
  • Machine Learning
  • Mathematical Models
  • Neurosciences
  • Psychology
  • Reinforcement Learning

Readers

  • Neural Network Machine Learning.
  • Robotics and Automation.
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • Autonomy