Sliding mode observer based non-causal model predictive control for collision-free marine launch and recovery system

• Nonlinear time-varying implementation of the novel tube-based MPC based on Sliding Mode Observer. • SMO reduces the bound of the disturbances that the tube-based MPC has to handle, resulting in lower steady-state error. • Monte-carlo simulation of various initial conditions is conducted to verify the robustness of the proposed control scheme. • SMO-based MPC ensures safe control across SNR levels and prediction errors under uncertain conditions. This paper investigates a non-causal control scheme for the Marine Launch and Recovery (L&R) systems involving a mothership and a small rigid-hulled inflatable boat (RHIB). The controller is considered non-causal as it requires predicted future wave information for decision-making. The main challenges are: 1) the system is underactuated with the tension force in the connecting cable as the only control input; 2) both mothership and RHIB are subject to uncertain wave-induced disturbances that pose challenges to collision avoidance; 3) the overall system exhibits nonlinear dynamics with coupling between the cable dynamics and the swing dynamics. Traditional Tube-based Model Predictive Control (MPC) addresses these challenges but leads to overly conservative control actions. In this paper, a Sliding Mode Observer (SMO) based MPC is proposed to reduce the conservatism, leading to reduced steady-state errors. Furthermore, the online computational load for the proposed scheme is similar to the traditional MPC, as the SMO is computationally cheap. Numerical simulations have been conducted to verify the effectiveness of the proposed scheme.