The desire to enhance cyber-physical systems (CPS) with cognitive capabilities represents a significant step forward in the evolution of robotics and intelligent automation. This paper focuses on the application of cognitive architectures to create cognitive CPS with the ability to perceive, reason and learn autonomously and also capable of interacting with the environment and human users in a meaningful and adaptive way. The analysis compares various cognitive architectures, highlighting their strengths and limitations for integrating cognitive functions into CPS. It examines how each framework supports cognitive processes such as sensory integration, attention management, action selection, memory recall, learning mechanisms and reasoning abilities.