Applications of Back Propagation Neural Networks to Robot Position Control



A method using neural network techniques for solving picking-up problems for robotic manipulators is presented in this article. An error back propagation network is used twice to process an image of a randomly placed object in order to find the gripping point and to obtain the required joint angles of the robot arms. Techniques of probabilistic and periodic coding are employed to improve generalisation ability of the network.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    A.P. REEVES, R.J. PROKOP, S.E. ANDREWS and F.P. KUHL, (1988) Three Dimensional Shape Analysis using Moments and Fourier Descriptors, p937–943, IEEE Trans. Pattern Anal. Machine Intell.Google Scholar
  2. 2.
    J. L. McCLELLAND and D.E. RUMELHART, (1988) Parallel Distributed Processing: Explorations in the Microstructure of Cognition, The MIT Press, Cambridge, Massachusetts.Google Scholar
  3. 3.
    C. LEE,(1990) A Vision System Using Pattern Associators for Finding Part Location and Orientation, Proc. 7th Int. Conf. Sys. Eng., p761–768, Las Vegas, USA.Google Scholar
  4. 4.
    W.A. WRIGHT,(1989) Probabilistic Learning on a Neural Network, Proc. IEE 1st Neural Network Conference, London.Google Scholar
  5. 5.
    R.P. LIPPMANN,(1989) Pattern Classification Using Neural Networks, p47–64, IEEE Communication Magazine, November.Google Scholar
  6. 6.
    J J CRAIG,(1986) Introduction to Robotics: Mechanics & Control, Addison-Wesley.Google Scholar

Copyright information

© Department of Mechanical Engineering University of Manchester Institute of Science and Technology 1992

Authors and Affiliations

  • C. Lee
    • 1
  1. 1.University of LiverpoolUK

Personalised recommendations