Utilizing Semantic and Contextual Information during Human-Robot Interaction

摘要

Humans are aware of the context and the environment while communicating with an interaction partner. Although equipping a robot with such cognitive skills is still a far cry, integrating contextual information into a robot is a step towards making Human-Robot Interaction (HRI) more meaningful, realistic or “human-like”. This paper focuses on establishing HRI based on semantic and contextual information in the conversation. The humanoid robot called ROBIN recognizes the user utterances using a speech recognizer called Julius. The approach proposed extracts named entities, contextual information and the sentiment of each user utterance, with robot’s responses triggered by a chatbot while gestures and facial expressions invoked by the sentiment labels of the user speech. The human-centered evaluation results based on the experimental scenarios show that the approach ensures more fluid and natural interaction.