Formation control of autonomous surface vessels (ASVs) is researched extensively in the last years, since it has promising applications. However very few designed strategies have been validated during real experiments. In this research, two
control methods for distributed leader-follower formation control are proposed: A Nonlinear Model Predictive Control (NMPC) method and an MPC method using Feedback Linearization (FL). These two designed methods are compared with a conventional Proportional-Integral (PI) control method. The performance of the proposed strategies is evaluated through simulations and real experiments using small-scale Roboat units. Both designed control methods outperform the conventional PI control method in simulation and real experiments.

In the automotive industry, automation is on the rise and it increases safety while decreasing costs. Improved sensor performance and greater computing power steers the future of the shipping industry in the same direction. Avoiding obstacles in close proximity is one of the current challenges. This research provides a generic and open- source design guideline for mapping, path planning and control for autonomous sailing in an environment with obstacles in close proximity. For each of the three modules, several options are discussed and the best one is chosen. Mapping is applied in the form of mapping with positioning sensors in an occupancy grid map. The original map is devided into cells containing information about the probability that they represent an obstacle. The mapping algorithm is applied in several maps and with noise and compared with a simultanously localization and mapping (SLAM) method. The resulting maps are created within the path planning requirements. Path planning is performed by inflating the occupancy grid map from the mapping algorithm. The inflated map is sampled into an 8-connected grid matching the grid cell size. A Dijkstra algorithm is applied combined with an quadratic approximation in the costmap. A steepest descent method will return a continous and shortest path which is suitable for control and does not collide with the obstacles in the map. Proportional and integral is applied to steer the vessel along the path. The proportional controller is tuned based upon the look ahead distance and its maximum error with the original path. The controller is tuned using a block shaped path as a reference and evaluated in a path from the path planner. Both static errors like wind or current and Gaussian noise on the position estimation are included for validation. Finally the controller is able to maneuver the vessel safely from its starting point to its final destination The functionalities of the modules are indivividually demonstrated, and eventually interconnected into one system. Autonomous maneuvering of the complete system is then demonstrated in various simulated environ- ments and settings, including unknown areas and global planning challenges. The vessel is able to maneuver autonomously from the start to its goal given only the sensor data, its location and its goal position.