Behavior-based planning and prosecution architecture for Autonomous Underwater Vehicles in Ocean Observatories

This paper discusses the autonomy framework proposed for the mobile instruments such as Autonomous Underwater Vehicles (AUVs) and gliders. Paper focuses on the challenges faced by these clusters of mobile platform in executive tasks such as adaptive sampling in the hostile underwater environment. Collaborations between these mobile instruments are essential to capture the environmental changes and track them for time-series analysis. This paper looks into the challenges imposed by the underwater communication infrastructure and presents the nested autonomy architecture as a solution to overcome these challenges. The autonomy architecture is separated from the low-level control architecture of these instruments, which is called the `backseat driver'. The back-seat driver paradigm is implemented on the Mission Oriented Object Suite (MOOS) developed at MIT. The autonomy is achieved by generating multiple behaviors (multiple objective functions) linked to the internal state of the platform as well as the environment. Optimization engine called the MOOS-IvP is used to pick the best action for the given instance based on the mission at hand. At sea operational scenarios and results are presented to demonstrate the proposed autonomy architecture for Ocean Observatory Initiative (OOI). Keywords: Autonomous Underwater Vehicles (AUVs), Underwater Gliders, MOOS, MOOS DB, MOOS-IvP, OOI-CI, behavior-based autonom