UW-MRS: Leveraging a Deep Grammar for Robotic Spatial Commands

Proceedings of the 8th International Workshop on Semantic Evaluation (SemEval 2014), pages 812–816, Dublin, Ireland, August 23-24, 2014. UW-MRS: Leveraging a Deep Grammar for Robotic Spatial Commands Woodley Packard University of Washington sweaglesw@sweaglesw.org Abstract This paper describes a deep-parsing ap- proach to SemEval-2014 Task 6, a novel context-informed supervised parsing and semantic analysis problem in a controlled domain. The system comprises a hand- built rule-based solution based on a pre- existing broad coverage deep grammar of English, backed up by a off-the-shelf data- driven PCFG parser, and achieves the best score reported among the task participants. 1 Introduction SemEval-2014 Task 6 involves automatic transla- tion of natural language commands for a robotic arm into structured “robot control language” (RCL) instructions (Dukes, 2013a). Statements of RCL are trees, with a fixed vocabulary of content words like prism at the leaves, and markup like action: or destination: at the nonterminals. The yield of the tree largely aligns with the words in the command, but there are frequently substitu- tions, insertions, and deletions. A unique and interesting property of this task is the availability of highly relevant machine- readable descriptions of the spatial context of each command. Given a candidate RCL fragment de- scribing an object to be manipulated, a spatial planner provided by the task organizers can auto- matically enumerate the set of task-world objects that match the description. This information can be used to resolve some of the ambiguity inherent in natural language. The commands come from the Robot Com- mands Treebank (Dukes, 2013a), a crowdsourced corpus built using a game with a purpose (von Ahn, 2006). Style varies considerably, with miss- ing determiners, missing or unexpected punc- This work is licensed under a Creative Commons At- tribution 4.0 International Licence. Page numbers and pro- ceedings footer are added by the organizers. Licence details: http://creativecommons.org/licenses/by/4.0/ tuation, and missing capitalization all common (Dukes, 2013b). Examples (1) and (2) show typi- cal commands from the dataset. (1) drop the blue cube (2) Pick yellow cube and drop it on top of blue cube Although the natural language commands vary in their degree of conformance to what might be called standard English, the hand-built gold stan- dard RCL annotations provided with them (e.g. Figure 1) are commendable in their uniformity and accuracy, in part because they have been automat- ically verified against the formal before and after scene descriptions using the spatial planner. (event: (action: drop) (entity: (color: blue) (type: cube)) Figure 1: RCL corresponding to Example (1). 2 Related Work Automatic interpretation of natural language is a difficult and long-standing research problem. Some approaches have taken a relatively shal- low view; for instance, ELIZA (Weizenbaum, 1966) used pattern matching to somewhat con- vincingly participate in an English conversation. Approaches taking a deeper view tend to parse utterances into structured representations. These are usually abstract and general-purpose in na- ture, e.g. the syntax trees produced by main- stream PCFG parsers and the DRS produced by the Boxer system (Bos, 2008). As a notable ex- …

Content truncated to 500 words