User Settings
Open AccessArticle

Situated Language Understanding for an Anthropomorphic Robot based on Linguistic Constructions and Embodied Simulation

Matthias Priesters,Malte Schilling,Stefan Kopp-2015-01-01-PUB – Publications at Bielefeld University (Bielefeld University)
0PDF

TL;DRAbstract

How do actors gain an understanding of their environment? On the one hand, they familiarize themselves with physical objects through bodily interaction and exploration; on the other hand, they gain understanding through linguistic interaction with other knowledgeable agents. We report work in progress from a project investigating such familiarization processes through modeling them in an anthropomorphic robot. The robot consists of a torso equipped with two arms with touch-sensitive hands enabling familiarization with novel objects and their affordances through bimanual action. This process is to be guided through situated dialogue with a human instructor, possibly including for example deictic expressions, anaphors or underspecified references to actions, such as in the following: “Open this box.” – “Okay.” – “No, you have to grasp it at the top.” – “Like this?” – “Yes, good, keep going”. Our approach aims to model the needed linguistic and dialogue abilities. We follow an embodied co

Chat with Paper

AI Agents for this Paper

How do actors gain an understanding of their environment? On the one hand, they familiarize themselves with physical objects through bodily interaction and exploration; on the other hand, they gain understanding through linguistic interaction with other knowledgeable agents. We report work in progress from a project investigating such familiarization processes through modeling them in an anthropomorphic robot. The robot consists of a torso equipped with two arms with touch-sensitive hands enabling familiarization with novel objects and their affordances through bimanual action. This process is to be guided through situated dialogue with a human instructor, possibly including for example deictic expressions, anaphors or underspecified references to actions, such as in the following: “Open this box.” – “Okay.” – “No, you have to grasp it at the top.” – “Like this?” – “Yes, good, keep going”. Our approach aims to model the needed linguistic and dialogue abilities. We follow an embodied co

Keywords

Embodied cognitionDeixisSituatedComputer scienceHuman–computer interactionAffordanceGestureAction (physics)

Chat

Click to start Chat