Ship Berthing Path Planning And Model Predictive Track Keeping Control

Autonomous berthing control is the key point and difficulty in the research of autonomous surface ships.With the implementation of the Maritime Autonomous Surface Ships(MASS)strategy,the berthing of ship will inevitably develop from qualitative to quantitative,and from experience to science.Compared to ocean navigation,the berthing process has complex planning paths and needs frequent manipulations.In this study,the berthing problem is divided into three levels: global path planning,local motion planning and motion control.Path planner and controller are designed based on this thinking.For global path planning,this thesis uses quadtree method to establish a multi-scale chart,and uses an improved A* algorithm to find the path.Reeds Shepp(RS)curve is used to design the final path to meet the non-integrity constraints existing in the berthing stage.The global berthing path planning is realized with constraints on berthing position and heading.For local motion planning,this thesis realizes the automatic path design through coordinate transformation and polynomial curve,and realizes the automatic switch of “collision avoidance-path tracking” strategies.For motion control,this thesis designs a controller based on nonlinear model predictive control(NMPC)to control the ship’s motion along the path,and uses a moving horizon estimation(MHE)method to reduce sensor errors.The entire planning and control method is coherent,and the planning and control effect is satisfactory.This thesis proposes an automatic path planning method and motion control method for sternway berthing,which helps to achieve automatic berthing of ships that require sternway berthing due to the nature of their cargos.The main contribution of the thesis is given in four aspects:(1)In order to meet different accuracy requirements of the berthing process,namely low offshore accuracy and high nearshore accuracy,a multi-scale method is proposed to achieve fast global path planning.The planning process first performs edge detection on the coast to remove obstacle information within large obstacle areas,so that the number of obstacles to be detected are reduced.Then,a quadtree is used to recursively partition the chart into blocks until there are no obstacles within the blocks or specified depth is reached.The advantage of multi-scale method is that they reduce the number of planning nodes.The nodes in open water are sparse,while in nearshore water are dense,which solve the problem of traditional uniform grid based planning method of “either low accuracy caused by large grid or low speed caused by small grid”.Based on multi-scale chart,one can achieve functions such as high-speed location search,adjacent nodes search and fast collision detection.The proposed multi-scale method can serve as a research foundation for various path planning strategies.(2)In order to solve special berthing problems such as Mediterranean mooring(sternway berthing),an end path design method based on RS curve is proposed.The RS curve refers to the shortest path with up to 2 cusps(caused by the change of advance and retreat motions)and constraints on the maximum curvature,as well as direction constraints on the start and finish points.The optimal RS path which satisfies non-integrity constraints is obtained by weighted selection.By the combining use of multi-scale A* algorithm for path following process and RS curves for berthing stabilization process,a path planning method for sternway berthing is proposed.(3)To solve the problem of strategy switch of path following and collision avoidance,a local motion planning method based on Frenet coordinate transformation and polynomial curve is proposed.Local motion planning can quickly respond to the dynamic changes of the environment,making up for the inadequacy of global planning in responding to navigation dynamics.A design method based on online double S trajectories is proposed for the design of motion speeds and related states.The advantage of this method is that the changes in speed and acceleration are gentle,which prevent the tremble of controller from sudden changes of references during navigation,and the states on path are time related,which provides state references for the control process.(4)A trajectory tracking method based on model predictive control is proposed for the tracking problem of discrete-time path states.The thesis discussed predictive control based on linear and nonlinear models,and utilized moving horizon estimation to reduce the impact of environmental interference and observation errors,good control results are achieved.In summary,this thesis realized berthing path planning and trajectory keeping control with four aspects of work,which contributes to the improvement of maritime safety guarantee capabilities and the development of MASS strategy.