Deep State Representation: an unified internal representation for the robotics cognitive architecture CORTEX

  1. Calderita Estévez, Luis Vicente
Dirixida por:
  1. Juan Pedro Bandera Rubio Director
  2. Pablo Bustos García de Castro Director

Universidade de defensa: Universidad de Extremadura

Fecha de defensa: 22 de xaneiro de 2016

Tribunal:
  1. José Moreno del Pozo Presidente/a
  2. Pablo Lanillos Pradas Secretario/a
  3. Luis Jesús Manso Fernández-Argüelles Vogal
  4. Ismael García Varea Vogal
  5. Antonio Bandera Vogal

Tipo: Tese

Teseo: 401051 DIALNET

Resumo

CORTEX proposes a agent---based architecture, where the agents are in charge of classical perceptual, actuation and planning abilities. The agents should be autonomous and obedient, and they are in charge of a well defined functionally --- whether it be reactive, deliberative of hybrid---. They communicate through a shared data structure called Deep State Representation. This Thesis focuses on the developing of Deep State Representation. The goal that is pursued seeks the design of a multi---labelled dynamic graph object that can hold several levels of abstraction, from basic geometry and sensorial state to high---level symbols and predicates describing the state of the robot and the environment during a short time lapse. The final view of the new robotics cognitive architecture CORTEX can be seen as cylindrical surface splitted in several vertical slices, which are the agents. In the centre of the cylinder, the new data structure holds the current internal belief of the robot, affecting and being affected by all the agents, and always struggling for an equilibrium that the reality fades away.