A model predictive autopilot for autonomous inland vessels

The European Commission has confirmed the importance of shipping via inland vessels based on the EU green deal. It means there will be a growth in their utilization, which would result in a shortage of crew members to operate these vessels. Hence, it is essential to add autonomous features to the navigation systems to moderate the burden of the navigation from the crew’s shoulders so that operating a vessel requires fewer crew members. On the other hand, using an optimal computer-aided navigation system also ensures an optimal path for the journey. Journeys through an optimal way, plus having fewer crew members on board, would lead to reduced fuel consumption and greenhouse gas emissions. Current vessels benefit from various navigation systems to keep the voyage’s navigation as autonomous as possible. Among all these systems, the Track Keeping Pilot (TKP), a means of automation, regulates ship steering to keep it on a pre-defined track. Existing TKPs use only user-defined waypoints as input and have no information regarding the vessel’s surroundings. To achieve a higher level of automation on board, considering the vessel’s surroundings situation, the vessel maneuvering dynamics, and external disturbances seem crucial. Therefore, this dissertation utilizes onboard instruments and navigation sensors as inputs to calculate a list of waypoints based on a Model Predictive Control (MPC) approach. In other words, at each iteration, the algorithm predicts the suitable reference waypoints of the ship over a finite horizon based on a constraint optimization problem. Therefore, since the MPC can handle constraints, these waypoints will form a reachable desired trajectory for the TKP. Then, the conventional onboard TKP or a renovated TKP will steer the ship to track the defined course and get to the desired waypoint. In the first phase, the necessary constraints and the MPC algorithm will be worked out and assessed in simulation environments. Afterward, the approach will be validated on an actual inland vessel model to introduce a higher level of automation on board.