A Computer Vision and HCI System for Robotic Arm Control
We developed a computer vision based object classification method and a human computer interaction (HCI) system which uses a Cyber Glove and a Flock of Birds® motion tracker to translate hand gestures into commands for a robotic arm. Our system consists of a conveyor, a simulated robotic manipulator, an HCI setup, and a vision system. First, the speed and the position of the objects are dynamically computed on a moving conveyor using standard motion estimation techniques. Then, objects are classified by an artificial neural network (ANN) using various image processing techniques and a moment invariant method. After the classification step, object specific commands are issued using hand gestures to further control the robotic arm which are translated into a command sequence via the HCI system. The system is successfully tested for different objects moving on a conveyor. The hardware implementation and the overview of the algorithms used along with the results obtained are presented.
C. Oz et al., "A Computer Vision and HCI System for Robotic Arm Control," 35th International Conference on Computers and Industrial Engineering, Unknown, Jan 2005.
Mechanical and Aerospace Engineering
National Science Foundation (U.S.)
University of Missouri--Rolla. Intelligent Systems Center
Keywords and Phrases
Artificial Neural Network; Moment Invariants; Object Classification
Library of Congress Subject Headings
Article - Conference proceedings