A flexible and scalable architecture for human-robot interaction

Diego Reforgiato Recupero;Danilo Dessì;Emanuele Concas

2019-01-01

---

Abstract

Recent developments and advancements in several areas of Computer Science such as Semantic Web, Natural Language Understanding, Knowledge Representation, and more in general Artificial Intelligence have enabled to develop automatic and smart systems able to address various challenges and tasks. In this paper, we present a scalable and flexible humanoid robot architecture which employs artificial intelligent technologies and developed on top of the programmable humanoid robot called Zora. The framework is composed by three different modules which enable the interaction between Zora and a human for tasks such as Sentiment Understanding, Question-Answering, and automatic Object Recognition. The framework is flexible and extensible, and can be augmented by other modules. Moreover, the embedded modules we present are general, in the sense that they can be easily enriched by adding training resources for the presented sub-components. The design of each module consists of two components (i) a front-end system which is responsible for the interaction with humans, and (ii) a back-end component which resides on server side and performs the heavy computation.