Abstract
In this paper we present an experiment of the visuomotor coordination of a simple mobile robot. Our approach to sensorimotor modelling belongs to the category of learning by doing. First, < perception, action > pairs are collected by observing the robot behavior during operation. Then, a learning by examples method is used to estimate the parameters of the model. In our experiment the learning machine is an Extended Kohonen Map. We have observed that this neural network model has the property of being naturally invertible. Given an input pattern, the network output value is retrieved by competition on the neuron fan-in weight vectors. This is the standard use of the input-output mapping that we call forward mode. Viceversa, given an output value, a corresponding input pattern can be obtained by competition on the neuron fan-out weight vectors. We call this use of the network backward mode. The invertibility property makes the Extended Kohonen Map worth considering for sensorimotor modeling. Our experiment shows that by training the network on the robot direct kinematics (the forward mode), one obtains at the same time a solution to the inverse kinematics problem (the backward mode). The experiment has been performed both in a computer simulation and by using a real robot.
Supported by the No. 21-36365.92 project of the Fonds National de la Recherche Scientifique, Berne, Suisse.
This is a preview of subscription content, log in via an institution to check access.
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
Béssiere, P., Dedieu, E., Mazer, E. Representing Robot/Environment Interactions Using Probabilities: the “Beam in the Bin” Experiment Proc. “From Perception to Action”, Lausanne, Switzerland.
Bullock, D., Grossberg, S., Guenther, F. A Self-Organizing Neural Network Model for Redundant Sensory-Motor Control, Motor Equivalence and Tool Use. Journal of Cognitive Neuroscience, 5, 408–435.
Kohonen, T. Self-Organization and Associative Memory. Springer Series in Information Sciences, 8, Heidelberg.
Massone, L. L. E. Sensorimotor Learning. To appear in “The Handbook of Brain Theory and Neural Networks”, M. A. Arbib Editor, MIT Press.
Mondada, F., Franzi, E., Ienne, P. Mobile Robot Miniaturization: a Tool for Investigation for Control Algorithms. Proc. of the Third International Symposium on Experimental Robotics, Kyoto, Japan.
Ritter, H., Martinetz, T., Schulten, K. Neural Computation and Self-Organizing Maps. An Introduction. Addison-Wesley.
Ritter, H., Schulten, K. Extending Kohonen's Self-Organizing Mapping Algorithm to Learn Ballistic Movements. Neural Computers, R. Eckmiller and E. von der Marlsburg (eds.), Springer, Heidelberg.
Zalama, E., Gaudiano, P., López Coronado, J. L. A Real-Time, Unsupervised Neural Network Model for the Control of a Mobile Robot in a Nonstationary Environment. Boston University Technical Report CAS/CNS-94-002.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1995 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Versino, C., Gambardella, L.M. (1995). Learning the visuomotor coordination of a mobile robot by using the invertible Kohonen Map. In: Mira, J., Sandoval, F. (eds) From Natural to Artificial Neural Computation. IWANN 1995. Lecture Notes in Computer Science, vol 930. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-59497-3_288
Download citation
DOI: https://doi.org/10.1007/3-540-59497-3_288
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-59497-0
Online ISBN: 978-3-540-49288-7
eBook Packages: Springer Book Archive