Research On Trajectory Planning Algorithm For Unmanned Surface Vehicle In Complex Environments With Obstacles

Unmanned surface vehicle(USV),as an unmanned surface motion platform with autonomous navigation capabilities,has been widely used in civilian and military fields.Whether a high-quality path can be planned has become one of the evaluation indexes of the USV's autonomous navigation capability.Therefore,the research on the USV's trajectory planning algorithm in complex environments with obstacles has important theoretical significance and practical value for the USV's autonomous and safe navigation.The research of the USV trajectory planning algorithm needs to solve the following problems:(1)When a USV sails in the ocean,it needs to consider multiple navigation objectives and be subject to multiple constraints.Therefore,it is necessary to plan the USV global trajectory from the perspective of multi-objective and multi-constraint optimization problems.(2)When planning the local traj ectory of the USV,it is necessary to consider the simultaneous situation of encountering multiple moving obstacles with time-varying velocity.(3)Since it is difficult for the USV to obtain the complete motion information of the obstacle,the USV needs to consider the uncertainty of the obstacle's motion information when planning the local trajectory.Based on the above analysis,this paper will study the USV trajectory planning problem in complex environments with obstacles from three aspects:the USV global trajectory planning,the USV local trajectory planning when the obstacle motion information is known,and the USV local trajectory planning when the obstacle motion information is uncertain.The main research of this paper includes the following aspects:The USV global trajectory planning problem under the influence of the marine environment is modeled.Firstly,a USV kinematic model under the influence of the marine environment is established.The USV global traj ectory planning space,the expression of global trajectory and obstacles,and the judgments of trajectory feasibility are introduced.Finally,a multi-objective and multi-constraint optimization mathematical model representing the USV global trajectory planning problem is established.Among them,the optimization objectives include traj ectory length,energy consumption,trajectory safety,and trajectory smoothness.The constraints include the maximum voyage,the maximum turning angle,and the sailing velocity limit.A multi-objective quantum particle swarm optimization(MOQPSO)algorithm is designed to solve the multi-objective and multi-constraint optimization problem that represents the global traj ectory planning of USV.Firstly,an improved quantum particle swarm optimization(IQPSO)algorithm is designed to overcome the shortcomings of particle swarm optimization(PSO)and quantum-behaved particle swarm optimization(QBPSO).On the basis of the IQPSO algorithm,the MOQPSO algorithm is designed by adding the selection strategy based on feasibility,the external archive set update strategy,the optimal solution selection strategy,and the constrained Pareto domination strategy.Finally,through the simulation experiments,it is verified that the MOQPSO algorithm has well convergence and solution diversity.The global trajectory set under the influence of the marine environment planned by the MOQPSO algorithm has better overall performance on the multiple optimization objectives.For the collision avoidance problem of obstacles moving in a known time-varying velocity,a USV local trajectory planning algorithm based on nonlinear finite-time velocity obstacle(NLFVO)is designed.Firstly,a nonlinear finite-time velocity obstacle(NLFVO)method is proposed for the situation where the USV encounters multiple velocity-varying obstacles.Secondly,the velocity variation and the course variation of the USV are used as the optimization objectives of the USV local trajectory planning,so that the USV local trajectory planning is transformed into a multi-objective optimization problem.Subjecting to the USV navigation rules and the kinematic constraints,the velocity and course of the USV that avoid collisions with moving obstacles and minimize the cost function value can be calculated.Finally,the simulation results verify the effectiveness of the USV local trajectory planning algorithm when the obstacle's motion information is known and certain.For the problem that it is difficult for USV to obtain the motion information of moving obstacles,a local trajectory planning algorithm is proposed when the motion information of obstacles is uncertain.Firstly,in order to obtain the future motion information of moving obstacles,an obstacle's velocity prediction model based on LSTM-GMR is designed according to Long Short-term Memory(LSTM),Gaussian Process Regression(GPR)model,and Gaussian Mixture Model(GMM).Secondly,the velocity interval prediction results obtained by the LSTM-GMR model are calculated to obtain the possible collision time between the USV and the moving obstacle.When the obstacle motion information is uncertain,a nonlinear finitetime velocity obstacle with uncertainty(UNLFVO)is designed to obtain the velocity and course that can avoid a collision.Finally,through the simulation experiments,it is shown that the velocity prediction model based on LSTM-GMR proposed in this paper can predict and obtain reliable future velocity of the moving obstacles.The USV local trajectory planning algorithm based on UNLFVO can make USV effectively avoid collisions with the obstacles with uncertain motion information.In summary,this paper studies the USV trajectory planning algorithm problem in complex environments with obstacles and validates the algorithm by simulations.The results show that the algorithms designed in this paper can effectively solve the USV global trajectory planning problem,the USV local trajectory planning problem when the obstacle motion information is known,and the USV local trajectory planning problem when the obstacle motion information is uncertain.The USV trajectory planning algorithm in complex environments with obstacles can provide theoretical basis and technical support for US V's autonomous and safe navigation.