Embodied Quantifiers

摘要

This paper studies how quantificational expressions such as few, three and all can be grounded in real-world perception. Based on findings from psycholinguistics, we propose a computational model designed for use in robot-robot interaction scenarios which involve discrimination tasks for objects in the real world. We test the performance of our model and contrast it with a type theory based model. We show that our design choices make our model more suitable for real-world applications.