Robotic Action Control: On the Crossroads of Cognitive Psychology and Cognitive Robotics

Abstract

T hefield of robotics is shifting from building industrial robots that can per-form repetitive tasks accurately and predictably in constrained settings, to more autonomous robots that should be able to perform a wider range of tasks, including everyday household activities. To build systems that can handle the uncertainty of the real world, it is important for roboticists to look at how humans are able to perform in such a wide range of situations and contexts-a domain that is traditionally the purview of cognitive psychology. Cognitive scientists have been rather successful in bringing computational systems closer to human performance. Examples include image and speech recognition and general knowledge representation using parallel distributed processing (e.g., modern deep learning models).