Two-layer MPC for Unmanned Surface Vehicle under Signal Temporal Logic Tasks

This paper studies the motion control problem of the unmanned surface vehicle (USV) under signal temporal logic (STL) tasks. Considering the kinematic and dynamical models of USV, a two-layer model predictive control (MPC) control framework is proposed. In the upper layer, the specified STL tasks are formulated into mixed-integer linear program (MILP) constraints and are incorporated into the MPC optimization problem to achieve the path planning; then in the lower layer, robust tube-based MPC (RMPC) is employed for path following; both upper and lower layers are computed online and interact in real-time to guarantee the task satisfaction for USV against uncertainties. Simulation results demonstrate the effectiveness of this two-layer MPC control framework.