Hand position-based L1-model predictive control for robust trajectory tracking of underactuated surface vehicles

This paper proposes a hand position-based L 1 -MPC framework for robust trajectory tracking of underactuated unmanned surface vehicles under environmental disturbances and model uncertainties. In maritime missions such as hydrographic surveys, environmental monitoring, docking, and dynamic positioning, accurate regulation is often required at a mission-relevant point, such as a sensor location or the bow tip, rather than at the center of gravity (CoG). Moreover, because underactuated surface vehicles have no direct sway control input, CoG-based dynamics contain unmatched sway uncertainty, limiting direct disturbance compensation. To address both aspects, this work adopts the hand position, an established output-redefinition concept defining a virtual point ahead of the CoG along the vehicle centerline, as the control reference. Expressing the dynamics at the hand position converts the sway disturbance unmatched at the CoG into matched uncertainty at the control point, enabling effective L 1 adaptive compensation despite the vehicle underactuation. This enables the proposed framework to combine constraint-aware MPC with L 1 adaptive augmentation. MPC regulates the hand position while handling vehicle dynamics and input constraints, whereas the L 1 controller compensates for disturbances and model uncertainties in real time. Simulation and field experiments demonstrate improved trajectory-tracking performance over the evaluated baseline methods.