Path-Following Control of Autonomous Tugs based on Data-Enabled Predictive Control

Tugs are often used to tow, escort, rescue, and provide other forms of assistance to large cargo ships in ports. They play a crucial role in enhancing the safety and efficiency of large ships when entering and leaving ports. This paper proposes a data-driven non-parametric strategy based on Data- Enabled Predictive Control (DeePC) to address the path-following control problem for autonomous tugs. In contrast to traditional approaches that rely on priori parametric models, the DeePC method implicitly characterizes the system based on input-output data, avoiding the complexity of mechanism models and system identification processes. Firstly, a line-of-sight (LOS) guidance strategy is used to transform the reference path points into reference heading angles. A heading controller is constructed based on DeePC, which directly utilizes the rudder angles and heading angles collected in real environment to construct a system trajectory matrix. This matrix serves as an implicit non¬parametric model that reflects the correlation between rudder angle and heading angle. Due to the nonlinear characteristics of ship dynamics, regularization techniques are used. Then a rolling horizon optimal control problem is formulated and solved in each control interval to determine optimal control inputs. Simulation results show that the proposed method performs well in the path-following control task without the necessity of previously acquired information regarding the hydrodynamic coefficients and ship parameters.