A Study on Generation of Optimal Route and Route Following Control for Autonomous Vessel

Abstract

In this study, basic research was conducted to adapt to the era of autonomous vessels. When a vessel operates autonomously, it sholuld determine the optimal route by itself and accurately follow the determined route. Even when environmental disturbances are applied to the vessel, it must have the ability to accurately follow the route. First, the optimal route should be generated to minimize fuel consumption. In addition, it should be able to satisfy the under keel clearance according to the specifications of the vessel, safety and regulations for offshore vessel. To achieve this purpose, this study used the Q-learning algorithm based on reinforcement learning. When the optimal route has been generated, the vessel should be able to accurately follow the generated route without deviating. Therefore, the vessel requires an autopilot controller. In the past, the PD (Proportional Derivative) type autopilot controller, which has a simple structure and convenient design, was applied and used, but it has problems such as rough alter course, low stability in nonlinearity, and overshoot owing to large heading angle changes. Therefore, in this study a velocity type fuzzy PID (Proportional Integral Derivative) autopilot controller was designed to compensated for such problems. However, even if the velocity type fuzzy PID autopilot controller with excellent performance is applied to the vessel, if environmental disturbances are applied to the vessel, it cannot accurately follow the designated route. Therefore, a method for estimating environmental disturbances with a fuzzy disturbance estimator using the innovation process characteristics of the Kalman filter is proposed in this thesis. If environmental disturbances are estimated using a fuzzy disturbance estimator, and converted them into thrust and rudder angle added to the vessel, the vessel can attain the ability to follow the designated route without deviating.