Skip to Main content Skip to Navigation

Modèles de mémoires pour la navigation autonome

Abstract : We propose a method for autonomous agent navigation problem grounded on a connectionist architecture inspired by neurobiological data concerning cortex and hippocampus. Our approach is grounded on distinction between memories : cerebral cortex is believed to be implied within declarative memory, allowing explicit memorization of facts and events. Our works concerns modeling of these memories in the framework of autonomous behavior. The study of these different memories underlines the importance of this procedural/declarative dichotomy in cognitive tasks while autonomous navigation study underlines the topological cognitive map notion as well as the necessity of having at disposal ad hoc memory systems. Our modeling work is separated in two parts.Available data concerning human cortex allow us to model a procedural memory for the construction of links between places of a simulated environment allowing a goal-guided search as well as the possibility of reaching anytime any known place in the environment. Furthermore, available data concerning hippocampus allow us to model a declarative memory system. The architecture of the model as well as the original mechanisms implied make this system able to handle both simulated and real-world data. THis model finally allows to characterize and memorize the different places of a given environment.
Document type :
Complete list of metadata
Contributor : Thèses Ul Connect in order to contact the contributor
Submitted on : Thursday, March 29, 2018 - 10:38:36 AM
Last modification on : Saturday, October 16, 2021 - 11:26:08 AM

Links full text


  • HAL Id : tel-01746477, version 1


Nicolas Rougier. Modèles de mémoires pour la navigation autonome. Autre [cs.OH]. Université Henri Poincaré - Nancy 1, 2000. Français. ⟨NNT : 2000NAN10122⟩. ⟨tel-01746477v1⟩



Record views