Abstract
In the framework of open-ended learning cognitive architectures for robots, this paper deals with the design of a Long-Term Memory (LTM) structure that can accommodate the progressive acquisition of experience-based decision capabilities, or what different authors call "automation" of what is learnt, as a complementary system to more common prospective functions. The LTM proposed here provides for a relational storage of knowledge nuggets given the form of artificial neural networks (ANNs) that is representative of the contexts in which they are relevant in a configural associative structure. It also addresses the problem of continuous perceptual spaces and the task- and context-related generalization or categorization of perceptions in an autonomous manner within the embodied sensorimotor apparatus of the robot. These issues are analyzed and a solution is proposed through the introduction of two new types of knowledge nuggets: P-nodes representing perceptual classes and C-nodes representing contexts. The approach is studied and its performance evaluated through its implementation and application to a real robotic experiment.
Highlights
Cognition encompasses a series of behavioral processes in animals, involving gathering sensory information, converting it into perceptions, making decisions and producing actions, that allow them to deal with dynamic or changing environments.[1,2] A cognitive architecture is a system that tries to implement cognition
This mechanism is inspired by the concepts of Conceptual Spaces[24] introduced by Gardenfors and of Network Memories[25] as proposed by Fuster. It revolves around the Long-Term Memory (LTM) and we propose a series of components and functionalities that allow for the integrated and task related formation of perceptual classes as well as context representations
This paper proposes a general structure for a LTM within a cognitive architecture in open-ended learning settings based on knowledge nuggets represented as artificial neural networks (ANNs)
Summary
Cognition encompasses a series of behavioral processes in animals, involving gathering sensory information, converting it into perceptions, making decisions and producing actions, that allow them to deal with dynamic or changing environments.[1,2] A cognitive architecture is a system that tries to implement cognition. Traditional general purpose cognitive architectures in the literature such as ACT-R,10 CLARION,[3] EPIC,[11] GLAIR,12 4CAPS13 or SOAR,[14] implement these decision processes within a set of internal structures and representations that encode the different knowledge nuggets as production rules or other types of externally imposed symbolic representations These representations lend themselves to very explicit and self-explanatory or selfdescriptive mechanisms to work with this knowledge, as well as to relatively straightforward computational indexing mechanisms when storing the knowledge nuggets in memory. It revolves around the Long-Term Memory (LTM) and we propose a series of components and functionalities that allow for the integrated and task related formation of perceptual classes as well as context representations This will permit progressively obtaining associative relationships among these structures and the rest of the knowledge nuggets present in LTM (such as models, policies, goals) in order to facilitate the selection of the appropriate policy or behavior for each situation without interference when the robot is faced with a priori unknown sequences of environments. These tests are designed to show how the characteristics that were required from the LTM are met
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
