Enhancement of Cognitive Abilities of an Agent-Robot on the Basis of Image Recognition and Sound Perception

Abstract

The objective of this work is to enhance the cognitive abilities of an agent robot (ARb), a model of a human. This article considers a new approach to ARb simulation, consisting of the utilization of a direct analogy between the functions of specific organs of the human brain/head and the "brain/head" of a model, between the life of a human and the "life" of a model. The model of a Homo Sapiens is constructed as an intellectual agent integrated with a humanoid robot. The task set is to construct an ARb that must be able to read and write, to understand its situation in the world, the meaning of its actions, and the semantics of the text it processes. An ARb achieves the high cognitive abilities mentioned above because, like a human, from the moment of its "birth" it learns a natural language, first and foremost such as the names of objects and phenomena it sees and recognises. It is proposed that an ARb will be taught, including language teaching, in a group of both fellow robots and humans, under the tutelage of a teacher. The growth of the cognitive abilities of an ARb is also achieved thanks to the evolution of a "population of reproducing agents" (our article [1]). For "reproduction", it is proposed that we should separate the "private life" of an ARb from its operating functions ("service"), i.e. to separate the "private" and the "service" spheres in the ARb software.