A Two Level MPC Architecture for Controlling Cooperative Underwater Gliders

Exploration by means of unmanned multi-robot formations is gaining interests in different fields. Autonomous underwater vehicles (AUVs) can be used as an efficient data acquisition system for surveying, monitoring and exploring in environments too risky for manned vehicles or too long or too hard operations for other kinds of unmanned vehicles. In particular, the marine exploration by means of autonomous glider fleets allows to save money and to safely perform dangerous missions. A hierarchical control architecture based on two Model Predictive Control (MPC) loops is here considered. The outer loop accomplishes the task of kinematic control for tracking a desired trajectory by setting the speeds for the inner loop. This loop implements a dynamic control on the actuators in order to track the desired speeds. Details on models and optimization problems for the controllers are provided and validated through simulations.