modified on 12/10/14


An intelligent robot would seem to be an obvious place to implement a cognitive architecture, yet most robotics applications generally have complex software systems that bear little resemblance to cognitive systems. Furthermore, architectures are usually heavily engineered for specific applications. In this paper, we discuss why general architectures that include deliberation and learning are rarely used in robotics and we pose some challenges for the designers of cognitive architectures to make them more suitable for robotic applications.