Model-Based Motion Filtering for Improving Arm Gesture Recognition Performance

Abstract

We describe a model-based motion filtering process that when applied to human arm motion data leads to improved arm gesture recognition. By arm gestures, we mean movements of the arm (and positional placement of the hand) that may or may not have any meaningful intent. Arm movements or gestures can be viewed as responses to muscle actuations that are guided by responses of the nervous system. Our method makes strides towards capturing this underlying knowledge of human performance by integrating a model for the arm based on dynamics and containing a control system. We hypothesize that by embedding the human performance knowledge into the processing of arm movements, it will lead to better recognition performance. We present details for the design of our filter, our analysis of the filter from both expert-user and multiple-user pilot studies. Our results show that the filter has a positive impact on the recognition performance for arm gestures.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 2003
Accession Number
ADA499029

Entities

People

  • Donald H. House
  • Greg S. Schmidt

Organizations

  • United States Naval Research Laboratory

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Accuracy
  • Angular Acceleration
  • Computational Science
  • Computers
  • Control Systems
  • Filtration
  • Hidden Markov Models
  • Human Body
  • Human-Machine Interaction
  • Joints (Anatomy)
  • Kalman Filters
  • Motion Capture
  • Neural Networks
  • Probability Distributions
  • Random Variables
  • Recognition
  • Template Patterns

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Computer Vision.
  • Robotics and Automation.