Natural language-based correction method for autonomous underwater vehicle communications

University of Idaho (UI) researchers have designed and constructed a fleet of autonomous underwater vehicles (AUVs) that are used to investigate a range of collaborative behaviors. The fleet was initially developed to investigate the use of AUVs to detect and inspect sea mines and mine-like objects in mine countermeasure (MCM) missions. Due to the collaborative nature of the mission, communication is critical to success. The nature of underwater acoustic communication is such that bandwidth is severely limited and errors in messages frequently occur (i.e only 80-90% transmission successes). In this paper we describe a method for correcting messages with errors by anticipating the possible messages being sent. This approach to message anticipation views fleet communication as conversational, allowing application of concepts related to natural human language. Because all the AUVs share the same language and the logic associated with that language, they can all anticipate, to some extent, the message being sent by another AUV. We refer to the use of language and its logic to anticipate and evaluate future hypothetical scenarios as language-centered intelligence (LCI). By applying the principals of fuzzy logic and probability, we expand upon previous message anticipation work at UI and demonstrate that such an approach can be used to repair the majority of a specific class of errors.