A Cognitive Architecture for Self Learning in Humanoid Robots

摘要

Cognitive is the mental process of knowing, including characteristics such as perception, awareness, judgment, and reasoning. Today humanoid robots need to become self-learner like humans, in this way they can be able to experience different things and learn from their experience, relating to, being, or involving conscious intellectual capable of being reduced to empirical factual knowledge. Considering the advantages of humanoid robots, in this study we propose a novel framework called Cognitive Architecture for Self Learning in Humanoid Robots (CASLHR). It combines the active memory, action schematical engine and sensor listener layers which try to produce human-like intelligence by analyzing the internal processes and the architecture of the human brain. The proposed CASLHR architecture may result in robust, safe, reliable, and flexible machines that can substitute humans in multiple tasks. This architecture is illustrated through case studies about fire-fighting task in the building and communication with the real-world. It can feel and perceive similar to a human being and will be able to learn from its experience and simultaneously updates its actions based on the success rate of its attempts to achieve a goal.