ONA for Autonomous ROS-Based Robots

Patrick Hammer, Peter Isaev, Tony Lofthouse, Robert Johansson

Abstract: In this paper we will show that OpenNARS for Applications (ONA) can be used for enhanced autonomous robots operating in the real world. It is done by utilizing state-of-the-art object detection along with Simultaneous Localization and Mapping techniques while inheriting the strengths of means-end reasoning and adding robust learning at runtime into the picture. This is showcased in an experiment where an ONA-controlled mobile robot with manipulator arm is learning about its environment to collect bottles to be returned to a human operator.