Towards an Integrated Architecture for Open-World Human-Robot Interactions

Abstract

“Open-world” human-robot interactions differ significantly from “closed-world” human-robot interactions because not all task-relevant information is available at the beginning of the interaction in the former (while it is in the latter). As a result, robots need to be able to acquire new knowledge online during task execution in open-world scenarios, which requires additional functionality in the cognitive robotic architecture to deal with unknown entities (such as words, objects, locations, actions, events, goals, etc.). In this presentation, I will provide an overview of our recent efforts towards developing an integrated architecture for open-world HRI and demonstrate some of the novel algorithms using videos from human-robot interaction scenarios.