OF THE 2 ND INTERNATIONAL IEEE Conference on Tools for Artificial

A new programming paradigm for the contro/ of a robot manipu/ator by leaming the mapping between the Cartesian space and thejoint space (inverse Kinematic) is discussed. Jt is based on a neural network model of optimal mappings between two high-dimensional spaces by Kohonen. This paper describes the approach and presents the optimal mapping, based on the principle of maximal information gain. Funhermore, the principal control error made by the leamed mapping is evaluated for the example of the commonly used PUMA robot. By introducing an optimization principle for the distribution of irifonnation in the neural network the optimal system parameters, including the number of neurons and the optimal position encoding reso/utions, are derived.