【作者】 赵鹏；

【导师】 张钦宇；

【作者基本信息】 哈尔滨工业大学 ， 信息与通信工程， 2019， 硕士

【摘要】 近年来,无人机及其相关技术研究受到各个领域的广泛关注,不仅在军事领域具有战略意义,如侦查、作战等,在民用领域也愈发受到企业与爱好者的追捧。相对于单架无人机,无人机编队有着许多优势,如可以降低无人机负载等。未来的无人机编队需要兼备自主性、灵活性、稳定性,且能够高成功率高效率地完成探测、物品运送、甚至更复杂的任务,满足不同场景的实用需求。从降低成本的角度来讲,无人机编队相较单架无人机更有优势,仅装载少量计算设备与传感器的无人机组成的大规模蜂群将有着广阔的应用领域。无人机编队任务包括了编队、队形变换、分裂重组等编队算法的设计,考虑到现实环境中存在障碍物以及动态威胁的问题,无人机编队方案设计中还需要结合避障方法的研究。为了设计一种具有避障能力和扩展规模能力的自主编队,本文中提出了一种基于群集行为法的无人机分布式编队方案。该方案中通过规定了无人机个体的基本行为,通过群体中按照运动规则互相影响最终产生群体共识,结合虚拟领导的设置并利用人工势场法进行改进。结合基于行为法规则设计的思想与人工势场法中势场函数改变无人机运动状态具有更平滑的优点,该方案中无人机运动参照队内邻居无人机运动状态使自身运动与群体达成一致,同时结合势场函数对环境中感应范围内障碍物自主避障。传统的领航跟随式编队鲁棒性差,领机掉线基本上会导致任务失败,所以本文中采用了虚拟领导的方式引导无人机编队的运动。针对现实生活中的具有动态障碍物的情况,通过为无人机设计依据动态障碍物计算虚拟速度障碍加强了避障成功率,并通过为编队设计虚拟壳的方法,减少无人机脱队的情况,增强编队稳定性。本文针对未来大规模可扩展的自主编队需求,创新性地结合了编队算法与避障算法,形成无人机-障碍-虚拟领航点的框架,且针对有动态障碍物的情景优化了无人机行为策略。通过MATLAB仿真50架无人机在自由空间和静、动态障碍物环境下运动验证了算法的有效性和稳定性,初始状态随机的无人机能够成功避开障碍驶出障碍区,并形成具有期望尺寸的编队。更多还原

【Abstract】 In recent years,research on Unmanned Aerial Vehicle(UAV)and related technologies has received extensive attention in various fields.Not only has military strategic significance,such as investigation and combat,but also in the civilian sector,it has become more and more sought after by enterprises and enthusiasts.Compared to a single drone,drone formation has many advantages,such as reducing the load on single drone.The future UAV formation needs autonomy,flexibility,stability,and high-performance and high-efficiency to complete detection,item transportation,and even more complex tasks to meet the practical needs of different scenarios.In nature,group behaviors such as fish,birds,or swarms formed by groups of simple rules are inspiring for UAV group design.From the perspective of cost reduction,the UAV formation has more advantages than the single UAV.The large-scale bee colony consisting of only a few UAVs with computing equipment and sensors will have a wide application field.The UAV formation task includes formation algorithms such as formation design,formation transformation,split reorganization,etc.Considering obstacles and dynamic threats in the real environment,research on design of the UAV formation scheme also needs to be combined with the obstacle avoidance method.In order to design an autonomous formation algorithm with the ability to avoid obstacles and expand its scale,we proposed a distributed formation strategy of the drone combined with the flock behavior method and the artificial potential field method.In the flock behavior method,only the basic behaviors of the individual drones are specified,and the group consensus can be generated by mutual image according to the movement rules in the group.The artificial potential field method sets the potential field for objects in space,and the drones determine their motion according to the influence of the potential field.Combining the rule design of flock behavior method with the potential field function in artificial potential field method which smoothly change the motion state,the UAV refers to the motion state of the neighboring drones in the team to make its own motion and the group’s reach an agreement.At the same time,combined with the potential field function,the obstacles in the sensing range of the environment are autonomously avoided.The traditional pilot-following formation is poorly robust,and the lead-down of the line basically causes the task to fail.Therefore,the virtual leader is used to guide the movement of the drone formation.Finally,in view of the real obstacles in real life,the virtual speed barrier is calculated for the UAV based on the dynamic obstacles to enhance the obstacle avoidance success rate,and the virtual shell designing for the formation speeds up UAVs’ regrouping.This paper aims to improve the traditional flock behavior and artificial potential field algorithm for the large-scale scalable autonomous formation requirements in the future,innovatively combines the formation algorithm and obstacle avoidance algorithm,and optimizes the drone for the scene with dynamic obstacles.The formation process of 50 UAVs in free space and static/dynamic obstacles environments are simulated with MATLAB.Validit y and stability of the algorithm are verified.Random UAVs can successfully avoid the obstacles,and after exit the obstacle zone,a formation with the desired size is presented.更多还原