The award will be divided evenly between Artzi and fellow principal investigator, Mohit Bansal of the University of North Carolina.
According to Artzi: “Spatial language reasons about the composition of physical spaces and actions in them, and is one of the most commonly used forms of language in daily life. While studied extensively, it has often been in severely constrained environments. This abstracts away issues of scale, and the simplifying assumptions result in language drastically different from the real problem. We propose to study spatial language comprehension and generation in interactive settings using resources that provide real-life visual input and environment configurations. Realistic inputs will enable the study of the complexity of natural language that occurs in real-life environments.”