Architecture and applications of Language-Centered Intelligence for unmanned underwater vehicles

摘要

Language-Centered Intelligence (LCI) is an approach to artificial intelligence currently under development at the University of Idaho. In this approach, the existing intervehicle communication language and associated logics are harnessed for parallel use to enable more advanced cognitive abilities, such as anticipation and hypothetical reasoning, and expand the behavioral repertoire of unmanned underwater vehicles (UUVs). We begin with a brief background of relevant UUV research and continue by defining LCI. Next, we propose an architecture for LCI and describe a few applications of LCI to collaborative UUV behaviors. Here, we detail the three primary sub-modules of LCI, namely, the Look-Ahead Inspections Module, the Imagination Replacement Approach, and the newly-proposed Message Anticipation Module (MAM). We conclude that the LCI approach represents a sophisticated extension of cognition in artificial agents and systems that use language to communicate.