Explaining Intelligent Agent’s Future Motion on Basis of Vocabulary Learning With Human Goal Inference