Motion Planning Algorithms for Marine Vehicles

Abstract

In this Master's thesis some of the most popular techniques in the motion-planning field are implemented in two-dimensional areas with obstacles. The Voronoi diagram method, the probabilistic roadmap method and the rapidly-exploring random trees method (RRT). Because the RRT has proved to be very efficient in finding paths connecting two points on a map while avoiding obstacles in earlier work the method is in this thesis combined with differential constraints such as curvature continuity which can be achieved using Dubin's vehicle kinematics with Fermat's spiral and clothoid transitions. To determine whether the method is suitable for collision-avoidance applications for marine vehicles the main part of the thesis is dedicated to test the generated paths as input for an extended version of a trajectory tracking guidance system. Even when introducing ocean current to the vehicle while tracking the method show promising results, and future work could revolve around further testing or expanding the method with another dimension for 3-D applications.