Advertisement

Abstract

The “intelligence” of robots is not yet sufficient to deal with complex situations in many potential applications, for example in the service area or for industrial assembly. For this reason, engineers are looking for solutions which are based on biology rather than on classical technical approaches, as animals have shown that they can cope successfully with complex problems. Using neural networks is one step in this direction and has become quite fashionable. However, not each possible application of a neural net is technically useful. Three different areas will be presented where the use of neural networks has advantages over classical methods. This classification will be substantiated by practical experience from ETH projects on the gripping of unknown objects, on a vision system for a robot playing ping pong, and on localizing addresses on postal parcels.

Keywords

Neural Network Mean Square Error Hybrid Approach Ping Pong Address Block 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. Ade, F., F. Akeret, A. J. Baerveldt, R. Gutmann, & J. Wen (1993, June). A two-step procedure to find the destination address block on parcels. Proceedings of the JET POSTE ‘93, 1st European Meeting on Postal Technologies. Nantes, France.Google Scholar
  2. Bozic, S. M. (1990). Digital and Kaiman filtering. London: Edward Arnold.Google Scholar
  3. Burdet, E. (1996). Algorithms of human motor control and their implementation in robotics. Doctoral thesis No 11692, Eidgenössische Technische Hochschule, Zürich, Switzerland. Fässler, H.Google Scholar
  4. R. H. A. Beyer, & J. Wen (1990). A robot ping pong player: Optimized mechanics, high performance 3D vision and intelligent sensor control. Robotersysteme 6:161–170.Google Scholar
  5. Fischler, M. A. , & O. Firschein (1987). Intelligence, the eyes, the brain, and the computer. Reading, PA: Addison-Wesley.Google Scholar
  6. Hornik, K. , M. Stinchcombeand, & H. White (1989). Multilayer feedforward networks are universal approximators. Neural Networks 2:359–366.CrossRefGoogle Scholar
  7. Kosko, B. (1992). Neural networks and fuzzy systems. Englewood Cliffs, NJ: Prentice Hall.zbMATHGoogle Scholar
  8. Maren, A. J. , D. Jones, & S. Franklin (1990). Configuring and optimizing the back-propagation network Handbook of Neural Computing Applications. San Diego, CA: Academic Press.Google Scholar
  9. Marsh, P. (ed. ), (1985). Robots (p. 1). Salamander Books Ltd.Google Scholar
  10. Scherrer, H. K. (1993). Intelligenter Greifer für Roboter. Doctoral thesis 10249. Eidgenössische Technische Hochschule, Zurich, Switzerland.Google Scholar
  11. Schweitzer, G. (1991). The robot as an intelligent interactive machine. Mechatronics 1 (4), 525–533.CrossRefGoogle Scholar
  12. Shadmehr, R. , & FA. Mussa-Ivaldi (1993). Geometric structure of the adaptive controller of the human arm. A. I. Memo 1437, Cambridge, MA: MIT Press.Google Scholar
  13. Vischer, D. (1992). Cooperating robot with visual and tactile skills. Proceedings of the IEEE International Conference on Robotics and Automation. Nice, France.Google Scholar
  14. Wen, J. , A. J. Baerveldt, & F. Ade (1991). Location of address labels on postal parcels using neural networks based on multifeatures. Proceedings of the International Conference on Industrial Electronics, Control and Instrumentation IECON’91 I. Kobe, Japan.Google Scholar
  15. Wen, J., & G. Schweitzer (1991, November). Hybrid calibration of CCD cameras using artificial neu-ral nets. Proceedings of the International Joint Conference on Neural Networks (pp. 1337–1342). Singapore.Google Scholar
  16. Wen, J. (1995). Hybrid approach of neural networks with knowledge-based explicit methods. PhD Thesis No 11009, Eidgenössische Technische Hochschule, Zurich, Switzerland.Google Scholar
  17. Xu, G. , H. K. Scherrer, & G. Schweitzer (1990). Application of neural net-works on robot grippers. Proceedings of the International Joint Conference on Neural Networks IJCNN ‘90. San Diego, CA.Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 2000

Authors and Affiliations

  • Gerhard Schweitzer
    • 1
  • Jianyong Wen
    • 1
  1. 1.Eidgenössische Technische Hochschule ZürichSwitzerland

Personalised recommendations