| dc.description.abstract | This thesis is the result of work focusing on collision avoidance in connection with formation control for unmanned surface vehicles (USVs) using coordinated target tracking. Especially utilization of such vessels in marine survey operations is a subject of importance. Such operations include seabed mapping, geological surveys and seismic services, to name a few. The potential use of USVs alongside a manned leader vessel in a formation within a leaderfollower framework is in this thesis explored, together with development of algorithms rendering this possible. The objective is to enable the usage of such formations in the context of a marine survey operation. The range of which USVs are used in survey operations today is at best limited. Hence, possible areas of application within which unmanned craft can make a difference are definitely present in this segment. The main focus of this thesis is thus to develop purposeful algorithms enabling usage of USVs in formation with a manned leader vessel in a marine survey operation. Analyzing the performance and assessing whether or not the algorithms behave as desired are other important subjects considered. The overall focus of this thesis is on collision-free formation control of USVs within the scope of marine survey operations. The operation at hand is in this thesis divided into three phases; start-up, underway and end. Formation control algorithms are thus developed for each phase, jointly fulfilling the requirement of controlling the USVs filling the roles of followers of a manned leader vessel in the formation proposed used. This formation consists of two USVs, in addition to the manned leader vessel. The start-up and end phases are both controlled using the developed Start-up algorithm, as the objectives of these phases are basically the same. The challenges of the underway phase are handled by the Underway algorithm. Together, these two formation control algorithms make up the main contribution of this thesis, namely a formation control system allowing USVs used safely in marine survey operations, in collaboration with a manned leader vessel. Two important concepts form the basis of the formation control algorithms; collision avoidance and target tracking. Target tracking is employed as means to achieve formation control, while inter-vessel collisions are prevented by employing a collision avoidance module developed in this thesis. Besides these common characteristics, the two algorithms differ in both objectives faced and how these objectives are completed. Both algorithms are reactive, and are thus only dependent on instantaneous measurements. They are also co-operative, meaning that the USVs jointly achieve the goals of avoiding inter-vessel collisions while tracking their respective formation positions. Beginning with the Start-up algorithm, this imposes velocity commands to the USVs that prevents inter-vessel collisions while making them converge to their initial positions in the formation. These positions are defined as targets for the USVs relative to the leader, in a pattern of choice. The target tracking module of the algorithm ensures tracking of these targets, while the collision avoidance module prevents hazardous situations occurring. Collision avoidance is based on the USVs giving the vessel approaching from their starboard side the right of way. The start-up phase is concluded when the USVs have reached their positions relative to the leader vessel in the formation (i.e. converged to their targets). The underway phase concerns the execution of the marine operation, and the Underway algorithm ensures that the formation pattern is sustained during this phase, while also allowing the leader vessel to alter the USVs’ positions in the formation. Collision avoidance is also achieved using this algorithm, this time by obligating the USV making its positional change to give the other vessels in the formation the right of way. Collision avoidance is certainly also ensured during the periods when the formation pattern is maintained. Numerical simulations were performed in order to assess the performance of both algorithms. Velocity reference models and a closed-loop velocitycontrolled model of the Viknes 830 USV were used to mimic the behaviour of the USVs in a realistic way during the simulations. The simulation results indicate that the algorithms performed well in the simulation scenarios, and were exhibiting the desired behaviour by preventing collisions and also sustaining formation control through target tracking. The empirical proofs of stability of the algorithms provided by the simulations are further emphasized by derived proofs of analytic stability by means of Lyapunov theory. | nb_NO |
