Intelligent subsea lander network to support resident AUVs for long duration under-ice ocean observations

Abstract

Resident autonomous underwater vehicles (RAUV) have been identified as having the potential to significantly impact coastal, mid-ocean and polar/under-ice observations. An enabling technology for persistent missions is underwater docking at a subsea lander (node), such as the Kongsberg K-Lander MK2, to recharge and store data collected. The proposed network of nodes is scalable, reconfigurable and economical compared to cabled observatories. The contribution of this work is the development of a fault detection identification and recovery (FDIR) framework for a seabed node. A simulator of the K-Lander MK2 is developed focusing on the powering system using the Robot Operating System (ROS) middleware. This simulator can be used for training, mission planning, sensor data review, and to evaluate the efficacy of FDIR measures. The K-Lander was deployed in April 2019 to validate the simulator and evaluate payload sensors. FDIR based on a Partially Observable Markov Decision Process (POMDP) model was implemented for two cases and yielded realistic responses compared to archived missions. The FDIR framework shows the potential of a subsea lander node supporting one or more RAUV(s) deployed for an extended period.