The cognitive architecture of a robotic salesman
Romero-Garces, Adrian,Calderita, Luis Vicente,Martínez-Gómez, Jesús,Bandera-Rubio, Juan Pedro,Marfil-Robles, Rebeca,Manso, Luis J.,Bustos, Pablo,Bandera-Rubio, Antonio Jesús
- Published
- 2015-12-04
- Language
- en
Abstract
This paper describes a robotics cognitive architecture for social robots named CORTEX. This architecture integrates di fferent levels of abstraction (from basic geometry to high-level predicates) into a unique Deep Space Representation (DSR) that diff erent agents interface. These agents update the contents of the DSR with new data from the outer world, and execute, plan and design behaviours. The design of CORTEX as an unified deep representation allows to fit both the subsymbolic processing and exibility requirements of robot control. In this paper a first implementation of CORTEX has been integrated into Gualzru, a robotic salesman, and tested in real scenarios. Results show that this cognitive architecture allows this robot to adequately execute its use case, and that it has a promising adaptability to achieve new tasks and be used in new scenarios.
Full text
The cognitive architecture of a robotic salesman Adrián Romero-Garcés, Luis Vicente Calderita, Jesús Martínez-Gómez, Juan Pedro Bandera, Rebeca Marfil, Luis J. Manso, Pablo Bustos, Antonio Bandera
Index •Introduction •CORTEX •Agents •Gualzru, the robot •Experiments •Conclusions The cognitive architecture of a robotic salesman 2
Introduction •Robotic architectures –Modules that allow the robot to work – Perception, motion, planning, learning… •Cognitive architectures –General-purpose reasoning –Problem solving –High-level cognitive processes •Robotic cognitive architectures –Integrate high-level simbolic cognitive processes in a robot operating in the real world. Introduction 3
CORTEX 1. Percepts represent objects and their properties 2. Goals are explicitly represented 3. Behaviours are represented by discrete actions and their parameters 4. Processing occur in cognitive cycles Introduction 4 CORTEX: Agent-based robotic cognitive architecture -Use a unique dynamic graph to model the reality at different abstraction levels (symbolic and geometric). -Use internal emulation to imagine.
Real tests for a real architecture: Objective •CORTEX is tested in a real scenario: the ITC20111030 ADAPTA Project •Use case: Gualzru, the robotic salesman Introduction 5 Bla bla bla… Here it is Thank you!
Index •Introduction •CORTEX •Agents •Gualzru, the robot •Experiments •Conclusions The cognitive architecture of a robotic salesman 6
•RoboCog: –First implementation of a cognitive architecture –World model: Two shared objects (InnerModel, AGM) –Changes have to be validated •CORTEX: –Only one object to model the world (DSR) –Two types of agent actions (structural and non-structural changes) RoboCog vs CORTEX CORTEX 7
CORTEX CORTEX 8 •DSR includes symbolic and geometric data •Agents are synchronized with the DSR and update it •Planning agents can execute cognitive cycles
Index •Introduction •CORTEX •Agents •Gualzru, the robot •Experiments •Conclusions The cognitive architecture of a robotic salesman 9
The ADAPTA use case Experiments 16
Final tests: Setup Experiments 17 10m x 8m Escuela Politécnica Superior Universidad de Málaga People walk freely through the test area x
Final tests: Results Experiments 18
Final tests: Results Experiments 19
Final tests: Results Experiments 20
Final tests: Results Experiments 21
Final tests: Results Experiments 22 Improved results after adding the touch panel
Index •Introduction •CORTEX •Agents •Gualzru, the robot •Experiments •Conclusions The cognitive architecture of a robotic salesman 23
Conclusions •CORTEX is able to work with different levels of abstraction in the same representation •DSR appears as a powerful unified representation of the world •Tested in real scenarios •Future research: –Inject raw data in the representation (proxies) –Emulations Conclusions 24
The cognitive architecture of a robotic salesman Adrián Romero-Garcés, Luis Vicente Calderita, Jesús Martínez-Gómez, Juan Pedro Bandera, Rebeca Marfil, Luis J. Manso, Pablo Bustos, Antonio Bandera Thank you very much! Questions?
