Robotic NLP vs. Human Subset Language Acquisition

This extended abstract investigates the appropriateness of natural language processing (NLP) in the context of robot-assisted navigation for the visually impaired. Several assumptions of corpus-based robotics are examined. It is argued that, in the short term, NLP may be inadequate and, in the long term, it may not be necessary to enable robotic guides to solicit route directions from bystanders. Human sublanguage acquisition is presented as one feasible alternative.