Research On Local Environment Construction And Motion Planning Of UUV

As the human devote more energy into marine exploration and development,the tasks undertaken by UUV become increasingly complex and diverse,putting forward higher requirements for the autonomous navigation ability of UUV.Local environment construction and motion planning are key technologies to achieve autonomous voyage for UUV.Considering the practical application requirements,this paper concentrates on the research of static and dynamic environment construction as well as motion planning of UUV for safe voyage,and makes every effort to provide a systematic solution involving perception and decision-making.The results of static and dynamic environment construction act as the input information of motion planning;the motion planning scheme integrates both deliberative collission avoidance method and reactive collission avoidance method for guarantee of safety.The main research contents and results of this paper include the following aspects:(1)The oriented cuboid static environment model and its construction method are proposed.The computing method of confidence values of the grids based on bayesian filter is explained,as well as the moving and updating principle of the three-dimensional rolling map.Furthermore,from the perspectives of reducing the storage space of environmental information and facilitating the decision-making of motion planning,an oriented cuboid environment model is proposed on the basis of three-dimensional grid map.A PSO based construction method of the oriented cuboid is proposed.Simulation results show that the environment modeling method based on three-dimensional rolling occupied grid map provides reasonable local environment information against the false alarm and missing detection effects of the sonar.An oriented cuboid set that completely covers the obstacle grids could be achieved by utilizing PSO method,which would save storage space,improve the computational efficiency,and ensure the safety of subsequent motion planning decisions.(2)A Clustering Statistic Hough Transform(CSHT)based estimation method for motion elements of multiple underwater targets is proposed.The detection information of the moving targets by the sonar appears as binary images,so Hough transform is emploied to extract motion elements characteristics.Considering the limits of unknown target number and the detection feature of the sonar,a clustering based statistic method is proposed,which helps overcoming the dependency of threshold as well as the false alarm and missing detection effects.Simulation results demonstrate that CSHT method is insensitive to environment noise,false alarm and missing detection effects of the sonar and offers favorable estimation accuracy and tracking performance,indicating engineering reliability.CSHT method would contribute to enhancement of situation assessment capability of the UUV towards the surrounding environment and would provide basic reference information for further decision-making behaviors like collision avoidance and path planning.(3)An adaptive differential evolution particle swarm optimization(ADEPSO)algorithm is proposed and is utilized for UUV trajectory planning in underwater three-dimensional dynamic space.Considering the UUV trajectory planning application background and the progressive environment information provided by the sonar,an adaptive differential evolution particle swarm optimization(ADEPSO)algorithm is proposed to improve the global optimization ability and adapt to dynamic constraints.The mutation,crossover and selection mechanisms of differential evolution algorithm are introduced into the basic PSO method for enhancement of diversity of the population,so as to improve the global optimization ability.In addition,the differential evolution ratio of the particle swarm is adjusted according to the variance of fitness value of sensitive particles,promoting adaptability to the dynamic constraints.Optimization results of typical functions prove that ADEPSO algorithm improves the ability of global optimization and the adaptability towards dynamic constraints compared with the basic PSO algorithm and differential evolution PSO algorithm.The objective function for optimization of UUV trajectory in three-dimensional dynamic environment is designed from safety and rapidity aspects with consideration of autonomous voyage mission requirements,and ADEPSO algorithm is utilized in the optimization process.The mutation,crossover,and selection mechanisms in the UUV trajectory planning problem and the method of setting sensitive particles are explained.Simulation cases of typical static obstacle environment and autonomous voyage of UUV in three-dimensional dynamic environment demonstrate the effectiveness and practicability of the ADEPSO based trajectory planning method.(4)An Self Adaptive Vector Field Histogram(SAVFH)method is proposed to realize the reactive collision avoidance of UUV in underwater three-dimensional dynamic space.On the basis of local trajectory planning,for guarantee of voyage safety,a reactive collision avoidance decision-making mechanism based on intelligent VFH is put forward.SAVFH method expands the basic VFH method to three-dimensional space,where the candidate direction set involves heading and pitch.The fitness of each candidate direction is calculated according to the distribution of the local obstacles with consideration of safety and rapidity,and the velocity could be adjusted according to comprehensive environmental threat degree and uncertainty.Simulation results indicate the collision avoidance capability of SAVFH method in static and dynamic environment,and it overcomes the sensitive threshold issue of basic VFH method.SAVFH method realizes safe collision avoidance decision-making in typical obstacle environments,and behaves more efficiently by utilizing oriented cuboid set static environment modelling method.The heading and velocity adjusting functions of the proposed SAVFH algorithm have been tested by lake and sea experiments,demonstrating the validity and practicability.(5)Comprehensive simulation experiments in the three-dimensional dynamic marine environment are performed.The motion planning simulation case in three-dimensional dynamic marine environment is designed,and two experiments are performed.One experiment is performed by merely utilizing SAVFH reactive collision avoidance method,and the other one combines the advantages of both the ADEPSO based trajectory planning method and the SAVFH method.Simulation results show that the integrated application of deliberative trajectory planning method and reactive collision avoidance decision-making mechanism realizes iterative trajectory optimization in three dimensional dynamic space by utilizing the progressively accumulated environment information,which guides UUV moving towards safe direction.Additionally,the SAVFH reactive collision avoidance decision-making mechanism can further ensure the safety in case of emergent obstacles.The integrated solution achieves more effective and reliable motion planning compared with merely utilizing deliberative method or merely utilizing reactive method.The research work of this paper provides theoretical reference and technical support for UUV autonomous voyage in unknown three dimensional dynamic environment.