A novel topological map of place cells for autonomous robots

  • Authors:
  • Vilson L. DalleMole;Aluizio F. R. Araújo

  • Affiliations:
  • Federal Technology University of Paraná, Toledo;Federal University of Pernambuco, Recife

  • Venue:
  • ICANN'10 Proceedings of the 20th international conference on Artificial neural networks: Part II
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents a novel Topological Map of Place Cells model for autonomous robots. In such a model the robot acquires and stores perceptions using a basic memory provided by our proposed growing self-organizing map. Context sensitive cells aim to obtain Place Cells whose activation is dependent on a remembrance process that fires the recollection of stored memories from current robot perceptions. The map is a graph of interconnected and topologically organized Place Cells. The robots notion of localization is primary guided by the recollection process, while vestibular stimuli estimates and a historic of lastly visited places disambiguate conflicting simultaneously activated Place Cells. The results are promising.