Image-based robot motion simulation

Qingjie Zhao*, Zengqi Sun

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

16 Citations (Scopus)
Plum Print visual indicator of research metrics
  • Citations
    • Citation Indexes: 16
  • Captures
    • Readers: 4
see details

Abstract

This paper describes an image-based robot animation technique. The technique utilizes eigen space method to generate compact representations of images from a set of reference images. A wavelet neural network is used to learn the relationship between robot joints and image representations. The learning algorithm is Quasi-Newton algorithm with Levenberg-Marquardt modifications. The trajectory in the joint space is first planned to generate a joint sequence. A corresponding sequence of representations can be obtained through the trained wavelet neural network and the image sequence can be synthesized using the result of eigen space method. No a priori models of robot and camera or any calibrations are needed. The virtual motion of the robot UP6 is demonstrated.

Original languageEnglish
Pages (from-to)257-263
Number of pages7
JournalOptics Communications
Volume205
Issue number4-6
DOIs
Publication statusPublished - 1 May 2002
Externally publishedYes

Keywords

  • Eigen space method
  • Quasi-Newton algorithm
  • Robot motion simulation
  • Wavelet neural network

Fingerprint

Dive into the research topics of 'Image-based robot motion simulation'. Together they form a unique fingerprint.

Cite this

Zhao, Q., & Sun, Z. (2002). Image-based robot motion simulation. Optics Communications, 205(4-6), 257-263. https://doi.org/10.1016/S0030-4018(02)01357-3