Perception-action map learning in controlled multiscroll systems applied to robot navigation
Authors:
- 1 December 2008
- journal article
- research article
- Published by AIP Publishing in Chaos: An Interdisciplinary Journal of Nonlinear Science
- Vol. 18 (4) , 043119
Abstract
In this paper a new technique for action-oriented perception in robots is presented. The paper starts from exploiting the successful implementation of the basic idea that perceptual states can be embedded into chaotic attractors whose dynamical evolution can be associated with sensorial stimuli. In this way, it can be possible to encode, into the chaotic dynamics, environment-dependent patterns. These have to be suitably linked to an action, executed by the robot, to fulfill an assigned mission. This task is addressed here: the action-oriented perception loop is closed by introducing a simple unsupervised learning stage, implemented via a bio-inspired structure based on the motor map paradigm. In this way, perceptual meanings, useful for solving a given task, can be autonomously learned, based on the environment-dependent patterns embedded into the controlled chaotic dynamics. The presented framework has been tested on a simulated robot and the performance have been successfully compared with other traditional navigation control paradigms. Moreover an implementation of the proposed architecture on a Field Programmable Gate Array is briefly outlined and preliminary experimental results on a roving robot are also reported.Keywords
This publication has 14 references indexed in Scilit:
- TURING PATTERNS IN RD-CNNs FOR THE EMERGENCE OF PERCEPTUAL STATES IN ROVING ROBOTSInternational Journal of Bifurcation and Chaos, 2007
- Chaotic Neurodynamics for Autonomous AgentsIEEE Transactions on Neural Networks, 2005
- Design and Analysis of Multiscroll Chaotic Attractors From Saturated Function SeriesIEEE Transactions on Circuits and Systems I: Regular Papers, 2004
- HOW AND WHY BRAINS CREATE MEANING FROM SENSORY INFORMATIONInternational Journal of Bifurcation and Chaos, 2004
- SENSORY FEEDBACK IN CNN-BASED CENTRAL PATTERN GENERATORSInternational Journal of Neural Systems, 2003
- Habituation in the kiii olfactory model with chemical sensor arraysIEEE Transactions on Neural Networks, 2003
- FAMILIES OF SCROLL GRID ATTRACTORSInternational Journal of Bifurcation and Chaos, 2002
- Local-global interactions and the role of mesoscopic (intermediate-range) elements in brain dynamicsBehavioral and Brain Sciences, 2000
- How brains make chaos in order to make sense of the worldBehavioral and Brain Sciences, 1987
- Planning of Minimum- Time Trajectories for Robot ArmsThe International Journal of Robotics Research, 1986