Linguistic, Kinematic and Gaze Information in Task Descriptions: The LKG-Corpus

16173

Knowledge Triples