Flatness-based MPC for underactuated surface vessels in confined areas

Abstract: A two-phase model predictive controller (MPC) is proposed for underactuated surface vessel operation in confined environments. For general driving maneuvers (phase one) the ship's geometry is not considered explicitly while in more restricted areas (stage two) which occur, e.g., in mooring maneuvers, the ship's geometry is approximated to ensure collision avoidance. To remove the dynamical constraint in the problem setup, the differential flatness of the fully actuated system is exploited and the flat outputs are parameterized using B-spline functions. Underactuated behavior is retained by means of inequality constraints that are imposed on the non-controllable input. In an effort to solve the MPC, a static nonlinear optimization problem is formulated and feasibility w.r.t. obstacles and actuator constraints is ensured at collocation points. Static obstacles are considered as constructive solid geometry functions in the MPC which also takes into account disturbances induced by wind.