WO2018103242A1 - 一种基于运动学习的四旋翼无人机电塔巡检方法 - Google Patents

一种基于运动学习的四旋翼无人机电塔巡检方法 Download PDF

Info

Publication number
WO2018103242A1
WO2018103242A1 PCT/CN2017/079180 CN2017079180W WO2018103242A1 WO 2018103242 A1 WO2018103242 A1 WO 2018103242A1 CN 2017079180 W CN2017079180 W CN 2017079180W WO 2018103242 A1 WO2018103242 A1 WO 2018103242A1
Authority
WO
WIPO (PCT)
Prior art keywords
motion
electric tower
tower inspection
learning
obstacle
Prior art date
Application number
PCT/CN2017/079180
Other languages
English (en)
French (fr)
Inventor
吴怀宇
陈鹏震
牛洪芳
钟锐
刘友才
程果
Original Assignee
武汉科技大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 武汉科技大学 filed Critical 武汉科技大学
Publication of WO2018103242A1 publication Critical patent/WO2018103242A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
    • G05D1/10—Simultaneous control of position or course in three dimensions
    • G05D1/101—Simultaneous control of position or course in three dimensions specially adapted for aircraft
    • G05D1/106—Change initiated in response to external conditions, e.g. avoidance of elevated terrain or of no-fly zones
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
    • G05D1/10—Simultaneous control of position or course in three dimensions
    • G05D1/101—Simultaneous control of position or course in three dimensions specially adapted for aircraft
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
    • G05D1/0094—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot involving pointing a payload, e.g. camera, weapon, sensor, towards a fixed or moving target

Definitions

  • the invention relates to a four-rotor unmanned electromechanical tower inspection method based on motion learning, and belongs to the technical field of unmanned aerial vehicle trajectory planning.
  • the four-rotor UAV has become a hot spot in the field of UAV research with its simple mechanical structure and unique flight mode. It is a non-coaxial multi-rotor aircraft that, due to its special design structure, enables it to achieve a variety of flight attitudes.
  • the control of the flight attitude is achieved by adjusting the rotational speed of the four rotors symmetrically distributed.
  • the four rotors provide lift more uniform than a single rotor, resulting in a smoother flight and greater maneuverability.
  • the four-rotor UAV is small in size, concealed and stable in flight, and is especially suitable for civil and military applications such as near-investigation and surveillance.
  • the four-rotor UAV In the civilian field, the four-rotor UAV is mainly used for ground detection, disaster relief, high-altitude shooting, etc.
  • the military field it is mainly used for military investigation, battlefield monitoring, and intelligence gathering.
  • the electric tower In the power system, the electric tower is an important infrastructure for grid transmission. The maintenance and quality inspection of the electric tower is an important guarantee for the operation and development of modern power systems.
  • the electric tower was inspected by manual climbing, and the difficulty of climbing the tower was high. Different from the manual climbing inspection method, the quadrotor drone can easily reach the top of the tower more than 80 meters above the ground, achieving full-scale high-definition shooting, which greatly reduces the difficulty and danger of inspection. Therefore, the four-rotor UAV is widely used in electric tower power grid inspection.
  • the drone technology is intensive and synergistic, making its operation more and more complicated.
  • the flight operator With the increasing intensity and difficulty of the modern electric tower inspection mission, the flight operator will be affected physically and psychologically. It is becoming more and more difficult to rely on manual operation to complete complex electric tower inspection tasks; therefore, 3D trajectory planning is crucial.
  • the existing three-dimensional space trajectory planning algorithms add many constraints on the basis of environmental modeling.
  • the flight path is searched by various algorithms. The superiority of the search path depends on the accuracy of the environment model and the real-time and effective algorithm. Sexuality, such a planning method is not completely suitable for UAV trajectory planning tasks of electric tower inspection, and the efficiency of searching and planning paths is not high.
  • the technical problem to be solved by the present invention is to provide a four-rotor unmanned electromechanical tower inspection method based on motion learning, so that the quadrotor UAV has the capability of independently inspecting the electric tower, and Freed from complex and heavy remote control tasks, and improved the efficiency of the inspection of the four-rotor unmanned electromechanical tower.
  • a method for inspecting a four-rotor unmanned electromechanical tower based on motion learning which is characterized by the following steps:
  • Step S1 Exercise learning process: teaching a four-rotor drone through a remote control to complete a specific electric tower inspection task. Collecting a specific three-dimensional trajectory sequence of the electric tower inspection task as a motion learning sample, extracting a motion primitive based on the dynamic equation and the motion learning sample; the three-dimensional trajectory sequence is displacement, velocity, and acceleration information of three degrees of freedom;
  • Step S2 Generalization process: setting a starting point for the new electric tower inspection task, generalizing the three-dimensional trajectory sequence of the new inspection task according to the dynamic equation and the extracted motion primitive, and using the new three-dimensional trajectory sequence as navigation The path to the four-rotor UAV completes the electric tower inspection task. If there is no obstacle in the new electric tower inspection task, the planning ends after step S2; if there is an obstacle on the planned path of the new electric tower inspection task, then Go to step S3;
  • Step S3 first determine the approximate center position coordinate of the obstacle, and then re-plan the feasible three-dimensional obstacle avoidance trajectory by designing the coupling factor on the basis of the existing learning; finally, the planned feasible obstacle avoidance trajectory is given to the quadrotor drone to complete the autonomy.
  • Electric tower inspection mission first determine the approximate center position coordinate of the obstacle, and then re-plan the feasible three-dimensional obstacle avoidance trajectory by designing the coupling factor on the basis of the existing learning; finally, the planned feasible obstacle avoidance trajectory is given to the quadrotor drone to complete the autonomy. Electric tower inspection mission.
  • step S1 when the motion primitive is extracted in step S1, a linear differential equation with a constant coefficient is introduced, and the point-to-point motion of the quadrotor UAV is described as a dynamic system model with a nonlinear mandatory term;
  • the remote control teaches the four-rotor UAV to complete a specific electric tower inspection task.
  • the three degrees of freedom of the four-rotor UAV ascending and descending phases are obtained.
  • the sequence of displacement, velocity and acceleration based on the obtained learning samples and dynamic system models, respectively extracts the motion primitive sequences of the four-rotor UAV ascending and descending phases, and the extracted primitive sequences are used as the autonomous planning paths in the subsequent steps. basis.
  • step S2 a new electric tower inspection task starting point and an end point are set, and the learned motion primitive is brought into a dynamic system model with nonlinear mandatory items, and the same regular system is used to generalize. A discrete motion track point is obtained, which is the required electric tower inspection path. If the new inspection task is unobstructed, the planning is completed after step S2, and the planned track point is given to the quadrotor drone to complete the electric tower inspection. Inspection task
  • step S3 re-generates the feasible three-dimensional trajectory sequence based on the original motion primitive and dynamic equation by designing the coupling factor in the dynamic equation, and uses the three-dimensional trajectory sequence as the navigation path of the electric tower inspection.
  • the four-rotor UAV completes the task of autonomous electric tower inspection.
  • step S3 is as follows:
  • Step S32 in each of the three degrees of freedom dynamic equations, adding respective coupling factors to construct a dynamic system with obstacle avoidance function; setting a good starting point according to the motion primitives extracted in step S1 according to each degree of freedom And the end point, the kinetic system with obstacle avoidance function generalizes the trajectory sequence of each degree of freedom, so that the sequence of three degrees of freedom constitutes the three-dimensional obstacle avoidance trajectory of the quadrotor UAV, and the drone completes the electricity according to the planned path.
  • Tower inspection mission in each of the three degrees of freedom dynamic equations, adding respective coupling factors to construct a dynamic system with obstacle avoidance function; setting a good starting point according to the motion primitives extracted in step S1 according to each degree of freedom And the end point, the kinetic system with obstacle avoidance function generalizes the trajectory sequence of each degree of freedom, so that the sequence of three degrees of freedom constitutes the three-dimensional obstacle avoidance trajectory of the quadrotor UAV, and the drone completes the electricity according to the planned path.
  • the present invention discloses a four-rotor unmanned electromechanical tower inspection method based on motion learning, which is studied for the movement of a single tower inspection task, and introduces a learning framework for the three-dimensional motion of the quadrotor UAV.
  • the motion primitives are extracted based on the dynamic equation; then the learned motion primitives can be extended to the new electric tower inspection mission for new electricity.
  • the tower inspection task can generalize the feasible track points after setting the starting point and the end point through the learned motion primitives; if there are obstacles on the new electric tower inspection path, it can pass the original learning foundation.
  • the design coupling factor re-plans the feasible obstacle avoidance path; finally, the planned trajectory point is provided to the quadrotor UAV for the autonomous flight of the four-rotor unmanned flight to complete the electric tower inspection.
  • the method of the present invention is unique in that:
  • the quadrotor UAV can not only replicate the motion it learns, but also generalize the path to meet different target points to complete the three-dimensional space mission, and at the same time be able to avoid barrier.
  • FIG. 1 is a schematic diagram showing the teaching of the four-rotor unmanned electromechanical tower of the present invention
  • FIG. 2 is a schematic diagram of the electric tower inspection of the four-rotor self-planning three-dimensional trajectory of the four-rotor UAV of the present invention
  • FIG. 3 is a schematic diagram of an electric tower inspection of the four-rotor unmanned aerial vehicle autonomously planned three-dimensional obstacle avoidance trajectory according to the present invention
  • Fig. 5 is a schematic diagram of the multi-degree-of-freedom motion learning of the quadrotor UAV of the present invention (taking three degrees of freedom as an example).
  • FIG. 1 is a schematic diagram of the four-rotor unmanned electromechanical tower inspection and teaching.
  • the electric tower inspection process is divided into a rising process (realization) and a descending process (dashed line); the starting point is point A and the ending point is point B during the ascending process; the starting point is point B and the ending point is point C during the descending process.
  • the ascending process and the descending process are different from the starting point and the ending point, and are two different forms of motion, it is necessary to analyze the motion of the same nature separately.
  • This paper only elaborates the specific steps of the method by taking the ascending process as an example.
  • Figure 1 is a schematic diagram of electric tower inspection by controlling a four-rotor UAV through a remote control. Taking the ascending process (A to B) as an example, the four-rotor in the ascending process is collected based on the position information processing module inside the quadrotor UAV. The man-machine motion trajectory points are set and the motion primitives are extracted according to the dynamic equation.
  • Figure 2 is a schematic diagram of the four-dimensional space trajectory of the four-rotor UAV for electric tower inspection. According to the learned motion element and set the starting point and the end point of the new task, the corresponding electric tower patrols the three-dimensional trajectory. Then, the four-rotor drone completes the inspection mission.
  • Figure 3 is a schematic diagram of the electric turret inspection of the four-rotor UAV self-planning obstacle avoidance three-dimensional trajectory.
  • obstacles such as trees
  • the original trajectory planning method is not applicable.
  • the feasible three-dimensional obstacle avoidance trajectory is re-planned, and then the four-rotor drone completes the electric tower inspection task.
  • Step S1 The flow chart of the motion element for extracting the inspection process of the four-rotor unmanned electromechanical tower is as shown in FIG. 4, and includes the following steps:
  • Step S11 Introducing a linear differential equation with a constant coefficient and calling it a dynamic system. This system is used as a basis for motion learning. For a motion y with a degree of freedom, the equation of motion is:
  • ⁇ v and ⁇ v are normal numbers, ⁇ represents a time constant, and g is a constant point of attraction.
  • ⁇ ⁇ v /4 and ⁇ > 0
  • the system is critically damped, and y can avoid periodic vibration and converge to g at the fastest.
  • the process in which y converges to g in this dynamic system can be seen as a discrete point-to-point motion process.
  • a nonlinear mandatory term f is added to equation (1).
  • f is designed as a linear weighted sum of radial basis kernel functions, resulting in a more general point-to-point motion fit.
  • Equation (6) describes the general point-to-point motion as a dynamic system with nonlinear forcing terms.
  • Step S12 The motion primitive process of extracting the inspection task of the four-rotor unmanned electromechanical tower is as follows.
  • the weight w i in the nonlinear forcing term f(s) in equation (6) is the motion primitive.
  • the four-rotor UAV is taught by a remote controller to perform a specific electric tower inspection task from A to B (taking the ascending process as an example) to obtain a three-dimensional spatial motion trajectory sequence.
  • One of the sequences of degrees of freedom, the sequence of displacement, velocity, and acceleration is T ⁇ t, 2 ⁇ t, ..., n ⁇ t ⁇ , where ⁇ t represents the step size.
  • Equation (3) is rewritten to equation (9) for estimating the motion primitive, that is, the weight value w i .
  • the corresponding T and w are as in equations (10) and (11).
  • the w i of the linear equation (9) can be calculated by the least squares method, that is, the motion primitive for this degree of freedom; the motion extraction process of the other two degrees of freedom is similar. Through the above steps, the motion primitives of the three degrees of freedom at the rising stages A to B can be obtained.
  • Step S2 The quadrotor UAV first sets the starting point and the ending point of the task according to the extracted motion primitive for the new electric tower inspection task, and then generalizes the corresponding three-dimensional trajectory sequence.
  • the process is as follows. Taking a degree of freedom as an example, set the starting point y 0 and the end point g of the new electric tower inspection task, and bring the learned motion elementary formula (11) into the equation (6) to plan a discrete motion track point.
  • the learning and generalization process of the above one degree of freedom motion is shown in Flowchart 4.
  • the above process is for a degree of freedom of motion. For the three free movements in space, it is necessary to ensure the coupling of time, that is, the synchronization and consistency of time.
  • the regular system equation (5) in the respective dynamic equations of the three degrees of freedom must be guaranteed to be consistent, see Figure 5.
  • the starting point A' and the ending point B' in FIG. 2 are set. (taking the ascending motion process as an example), the trajectory sequence of each degree of freedom is calculated according to equation (6), and the path from A' to B' is formed, which is the required electric tower inspection path.
  • the trajectory planning is completed through step S2, and the three-dimensional trajectory is given to the four-rotor drone and the inspection is completed.
  • Step S3 When there is an obstacle in the new electric tower inspection task, the original planning method cannot be applied, and the re-feasible planning is needed.
  • the detailed steps are as follows:
  • the coupling factor C t is added to (6) to construct a dynamic system with obstacle avoidance function, namely equation (12).
  • C t [C t,1 C t,2 C t,3 ] T
  • C t is specifically solved as in equation (13), where ⁇ is the velocity vector
  • ⁇ is the velocity vector
  • R is the rotation matrix to determine the direction of rotation of the motion trajectory around the obstacle
  • a coupling motion perpendicular to the current velocity direction is added to the coupling factor as a function of the distance vector and the velocity vector.
  • Step S32 In each dynamic equation of each degree of freedom, a respective coupling factor is added, as in equation (12). According to the motion primitives extracted in step S1 for each degree of freedom, the coordinates of the starting point A" and the end point B" are set (take the ascending process as an example), The three-degree-of-freedom trajectory sequence is generalized by the formula (12), so that the three-degree-of-freedom sequence constitutes the three-dimensional obstacle avoidance trajectory of the quadrotor UAV, and the drone completes the electric tower inspection task according to the planned path, such as image 3.
  • the electric tower inspection method for the three-dimensional space of the four-rotor UAV proposed by the present invention is a trajectory planning method based on motion learning, and the learning of a specific electric tower inspection task is for a new one.
  • the electric tower inspection task can independently plan a feasible path for the four-rotor UAV, which has high engineering practical value.

Abstract

一种基于运动学习的四旋翼无人机电塔巡检方法，首先对四旋翼无人机的三维运动引入学习框架，将一次电塔巡检飞行任务轨迹作为运动学习的样本，基于动力学方程提取出其运动基元；进而基于学习到的运动基元可推广到新的电塔巡检飞行任务，泛化出相应的运动轨迹；当规划的飞行轨迹上有障碍物时，在已有学习基础上通过设计耦合因子从而规划出三维避障轨迹；最终可将得到的可行巡检轨迹用于四旋翼无人的自主飞行。该方法从四旋翼无人机运动学习的角度出发，基于学习得到的运动基元对新的电塔巡检飞行任务进行三维轨迹规划，完善了四旋翼无人机的自主轨迹规划的方法，有望提高电塔巡检效率。

Description

一种基于运动学习的四旋翼无人机电塔巡检方法 技术领域
本发明涉及一种基于运动学习的四旋翼无人机电塔巡检方法，属于无人机轨迹规划技术领域。
背景技术
近年来，四旋翼无人机以其简单的机械结构和独特的飞行方式而成为无人机研究领域中的热点。它是一种非共轴多旋翼式飞行器，由于其特殊的设计结构，使得其可以实现多种飞行姿态。通过调节对称分布的四个旋翼转速，实现对飞行姿态的控制。与常规直升机相比，四个旋翼提供升力比单旋翼更均匀，因而飞行更加平稳且机动性更强。四旋翼无人机体积小、隐蔽性好、飞行平稳，特别适合近侦查、监视等民用和军用领域。在民用领域，四旋翼无人机主要被应用于地面检测、抗灾救险、高空拍摄等；在军用领域，主要被用于军事侦查、战场监控、情报收集等。
在电力系统中，电塔是电网输电的重要基础设施，电塔的维护及质量检测是现代电力系统运行与发展的重要保障。过去通过人工攀爬检查电塔，登塔难度高危险性较大。与人工攀爬检查方式不同，四旋翼无人机可轻松到达距离地面80多米高的塔顶，实现全方位高清拍摄，从而大大降低了巡检难度和危险性。所以四旋翼无人机被广泛应用于电塔电网巡检。
但是，无人机技术密集及协同性强，使它的操纵越来越复杂；随着现代电塔巡检飞行任务的强度、难度的不断增加，飞行操作手在生理和心理上会受到影响，单纯依靠手控操作完成复杂的电塔巡检任务变得越来越困难；因此三维轨迹规划显得至关重要。然而现有的三维空间轨迹规划算法均在环境建模的基础上添加诸多的约束条件通过各种算法搜索飞行路径，搜索到路径的优越性依赖于环境模型的精确程度及算法的实时性和有效性，这样的规划方式不完全适合电塔巡检的无人机轨迹规划任务，搜索与规划路径的效率不高。
发明内容
针对上述现有技术问题，本发明要解决的技术问题是：提供一种基于运动学习的四旋翼无人机电塔巡检方法，使得四旋翼无人机具备自主巡检电塔的能力，将人从复杂和繁重遥控任务解脱出来，并提高四旋翼无人机电塔巡检的效率。
为解决上述技术问题，本发明采用如下技术方案：
一种基于运动学习的四旋翼无人机电塔巡检方法，其特征在于具体包括如下步骤：
步骤S1：运动学习过程：通过遥控器示教四旋翼无人机完成一次具体的电塔巡检任务， 收集该具体的电塔巡检任务三维轨迹序列作为运动学习样本，基于动力学方程和运动学习样本提取运动基元；所述三维轨迹序列即三个自由度的位移、速度、加速度信息；
步骤S2：泛化过程：为新的电塔巡检任务设定起点终点，根据动力学方程和提取的运动基元泛化出新巡检任务的三维轨迹序列，将新的三维轨迹序列作为导航路径给四旋翼无人机完成电塔巡检任务，若新的电塔巡检任务中没有障碍物则步骤S2后规划结束；若新的电塔巡检任务的规划路径上有障碍物，则进入步骤S3；
步骤S3：先确定障碍物大致的中心位置坐标，然后在已有学习基础上通过设计耦合因子重新规划出可行三维避障轨迹；最后将规划出的可行避障轨迹给四旋翼无人机完成自主电塔巡检任务。
上述技术方案中，步骤S1提取运动基元时，引入带有恒定系数线性微分方程，将四旋翼无人机的点到点运动描述为一种带有非线性强制项的动力学系统模型；通过遥控器示教四旋翼无人机完成一次具体的电塔巡检任务，基于四旋翼无人机内部的位置估计模块，分别获取四旋翼无人机上升阶段和下降阶段起点到终点三个自由度各自位移、速度和加速度的序列，基于获得的学习样本和动力学系统模型分别提取四旋翼无人机上升和下降阶段的运动基元序列，提取到的基元序列作为后续步骤中自主规划路径的基础。
上述技术方案中，步骤S2中，设定新的电塔巡检任务起点和终点，将学到的运动基元带入有非线性强制项的动力学系统模型，并使用同一正则系统，泛化出一条离散的运动轨迹点，即为所需的电塔巡检路径，若新的巡检任务无障碍物则步骤S2后规划结束，将规划的轨迹点给四旋翼无人机完成电塔巡检任务；
上述技术方案中，步骤S3通过在动力学方程中设计耦合因子，在原有的运动基元和动力学方程基础上重新泛化出可行三维轨迹序列，将三维轨迹序列作为电塔巡检的导航路径给四旋翼无人机完成自主电塔巡检任务。
上述技术方案中，步骤S3详细步骤如下：
步骤S31：设定在上升阶段和下降阶段，在三维笛卡尔空间中从起点到目标点的运动过程中有障碍物，首先确定障碍物大致中心为o＝[o1 o2 o3]T，在带有非线性强制项的动力学系统模型中加入耦合因子Ct从而构建带避障功能的动力学系统；对于三个自由度的运动，各自带避障功能的动力学系统有其各自的耦合因子Ct＝[Ct，1 Ct，2 Ct，3]T，Ct，j(j＝1，2，3)；耦合因子中加入了一个垂直于当前速度方向的运动，是距离矢量与速度矢量的函数；其中
Figure PCTCN2017079180-appb-000001
其中μ为速度向量
Figure PCTCN2017079180-appb-000002
与障碍物中心坐标和当前位置坐 标差向量(o-y)的夹角；R为旋转矩阵，决定运动轨迹绕障碍物的旋转方向，k和β为常量。
步骤S32：在三个自由度各自的动力学方程中，均添加各自的耦合因子构建带避障功能的动力学系统；根据各个自由度在步骤S1中所提取的运动基元，设定好起点和终点，由带避障功能的动力学系统泛化出各自自由度的轨迹序列，从而三个自由度的序列组成四旋翼无人机三维的避障轨迹，进而无人机根据规划路径完成电塔巡检任务。
综上所述，本发明公开了一种基于运动学习的四旋翼无人机电塔巡检方法，针对一次电塔巡检任务的运动进行学习，对四旋翼无人机的三维运动引入学习框架，将一次电塔巡检飞行任务轨迹作为运动学习的样本，基于动力学方程提取出其运动基元；进而基于学习到的运动基元可推广到新的电塔巡检飞行任务，对于新的电塔巡检任务，则可通过已学习的运动基元在设定起点和终点后泛化出可行的轨迹点；若新的电塔巡检路径上有障碍物，则可通过在原有的学习基础上设计耦合因子重新规划出可行的避障路径；最后将规划出的轨迹点提供给四旋翼无人机用于四旋翼无人的自主飞行完成电塔巡检。
对比传统的无人机路径规划方法如蚁群算法、粒子群算法、A*算法等，本发明方法特有之处在于：
(1)引入了运动学习机制，基于这样一种机制，四旋翼无人机不仅能够复制其学习到的运动，而且能够泛化出满足不同目标点的路径完成三维空间飞行任务，同时能够进行避障。
(2)传统的无人机路径规划方法由于未引入学习机制，面对相同或相似的环境原先的规划结果并没有对现有的规划有任何帮助。从而本方法摆脱传统的三维路径规划对环境建模与搜索算法性能的依赖，完善了四旋翼无人机的自主轨迹规划的方法，提高了四旋翼无人机的轨迹规划能力和电塔巡检效率。
附图说明
图1是本发明四旋翼无人机电塔巡检示教示意图；
图2是本发明四旋翼无人机自主规划三维轨迹进行电塔巡检示意图；
图3是本发明四旋翼无人机自主规划三维避障轨迹进行电塔巡检示意图；
图4是本发明基于运动学习的四旋翼无人机电塔巡检方法运动学习及泛化过程；
图5是本发明四旋翼无人机的多自由度运动学习原理图(以三个自由度为例)。
具体实施方式
下面结合附图和实例对本发明作更进一步的说明。
本发明所提出的方法首先需要对一次具体的电塔巡检任务进行学习，图1是四旋翼无人机电塔巡检示教示意图。如图1所示，电塔巡检过程分为上升过程(实现)和下降过程(虚线)；上升过程中起点为点A，终点为点B；下降过程中起点为点B，终点为点C。由于上升过程和下降过程是起点和终点不同，并且是两种不同的运动形式，因此需要单独对同种性质的运动进行分析，本文仅以上升过程为例详细阐述本方法的具体步骤。
图1为通过遥控器控制四旋翼无人机进行电塔巡检示意图，以上升过程(A到B)为例，基于四旋翼无人机内部的位置信息处理模块收集上升过程中的四旋翼无人机运动轨迹点集，并根据动力学方程提取运动基元。图2为四旋翼无人机自主规划三维空间轨迹进行电塔巡检示意图，根据学习到的运动基元并设定好新任务的起点和终点，自主泛化出相应的电塔巡检三维轨迹，然后给四旋翼无人机完成巡检任务。图3为四旋翼无人机自主规划避障三维轨迹进行电塔巡检示意图，当新的电塔巡检任务路径上有障碍物(如：树)时，则原有的轨迹规划方法不在适用，通过设计在动力学方程中加入耦合因子，重新规划出可行的三维避障轨迹，然后给四旋翼无人机完成电塔巡检任务。
上述技术方案整个过程的具体实现步骤如下：
步骤S1：提取四旋翼无人机电塔巡检过程的运动基元流程图如图4，包含了以下步骤：
步骤S11：引入带有恒定系数线性微分方程并称之为动力学系统，此系统作为对运动学习的基础，对于一个自由度的运动y，其运动方程为：
Figure PCTCN2017079180-appb-000003
Figure PCTCN2017079180-appb-000004
式(1)和(2)中αv和βv为正常数，τ表示时间常数，g为吸引点也为常数。选取合适的值如βv＝αv/4且τ＞0，则系统临界阻尼，y能够避免周期性振动而最快地收敛于g。此动力学系统中y收敛于g的过程可以看作是离散的点到点的运动过程。但是上述系统只能得到一种特定的临界阻尼运动，为推广到更广泛具有一般形式的运动，则在式(1)中加入非线性强制项f。f设计为一种径向基核函数线性加权和的形式，进而得到更一般的点到点运动拟合形式。
Figure PCTCN2017079180-appb-000005
ψi(s)＝exp(-hi(s-ci)2)       (4)
Figure PCTCN2017079180-appb-000006
式(3)为f函数的具体形式，式中N表示径向基核函数ψi(s)的个数；式(4)为径向基核函 数函数的具体表达式，hi＞0且决定核函数的宽，ci是径向基核函数的中心，其中
Figure PCTCN2017079180-appb-000007
式(5)称为正则系统，决定式(4)中变量s的动态特性，s的初始状态为s(0)＝1。进而改进动力学方程为：
Figure PCTCN2017079180-appb-000008
Figure PCTCN2017079180-appb-000009
式(6)将一般的点到点运动描述为一种带有非线性强制项的动力学系统，非线性强制项f(s)随时间衰减，最终系统收敛于(v，y)＝(0，g)，从而此动力学系统能够拟合不同的运动形式。
步骤S12：提取四旋翼无人机电塔巡检任务的运动基元过程如下。式(6)中非线性强制项f(s)中的权重wi即为运动基元。在图1中通过遥控器示教四旋翼无人机一次具体的电塔巡检任务即从A到B点(以上升过程为例)，得到其三维空间运动轨迹序列。其中一个自由度的序列，位移、速度和加速度的序列为
Figure PCTCN2017079180-appb-000010
t∈{Δt，2Δt，…，nΔt}，其中Δt表示步长。离散系统的运动的起点y0＝ydemo(0)，运动轨迹终点g＝ydemo(nΔt)，运动时间常数τ＝nΔt。将式(7)代入(6)，并将已知的运动序列带入得式：
Figure PCTCN2017079180-appb-000011
由式(8)可得通过学习得到的非线性强制项的序列，为找到式(6)中合适的权重wi，则将问题转化为函数逼近，即使得f尽可能接近fdemo。为估计运动基元即权值wi将式(3)改写成式(9)。对应的T和w如式(10)和(11)。
Tw＝f≈fdemo            (9)
Figure PCTCN2017079180-appb-000012
w＝[w1 … wN]T          (11)
通过最小二乘法可计算出线性方程(9)的wi，即为此个自由度的运动基元；其余两个自由度的运动基元提取过程类似。通过以上步骤可求得上升阶段A到B处三个自由度各自的运动基元。
步骤S2：四旋翼无人机根据提取的运动基元对于新的电塔巡检任务，首先设定任务的起 点和终点，然后泛化出相应的三维轨迹序列，过程如下。以一个自由度为例，设定新的电塔巡检任务起点y0和终点g，将学到的运动基元式(11)带入式(6)，规划出一条离散的运动轨迹点，从而达到泛化的过程，以上的一个自由度运动的学习和泛化过程见流程图4。上述过程是针对一个自由度的运动，对于空间中三个自由的运动还需保证时间的耦合性，即时间上的运功同步和一致。因此，三个自由度各自的动力学方程中的正则系统式(5)必须保证一致，见图5。至此，对于新的电塔巡检任务，依据步骤S1中学习到的三个自由度各自的运动基元，并使用同一正则系统，在设定好如图2中的起点A′和终点B′(以上升运动过程为例)后，根据式(6)计算出各个自由度的轨迹序列，形成由A′到B′的路径，即为所需的电塔巡检路径。对于新的电塔巡检任务中，若无障碍物则经过步骤S2完成轨迹规划，将三维轨迹点给四旋翼无人机后巡检结束。
步骤S3：当新的电塔巡检任务中有障碍物时，则原来的规划方法不能适用，需要重新可行规划。详细步骤如下：
步骤S31：在三维笛卡尔空间中，点到点的运动其位置向量为y＝[y1 y2 y3]T和对应的速度向量为
Figure PCTCN2017079180-appb-000013
，目标点为g＝[g1 g2 g3]T。在从起点到目标点的运动过程中有障碍物，设障碍物大致球心为o＝[o1 o2 o3]T，如图3。在(6)中加入耦合因子Ct从而构建带避障功能的动力学系统，即式(12)。
Figure PCTCN2017079180-appb-000014
对于三个自由度的运动，各自的动力学系统有其各自的耦合因子，写成向量形式为Ct＝[Ct，1 Ct，2 Ct，3]T，Ct，j(j＝1，2，3)即为规划避障路径的关键。Ct具体解算如式(13)，其中μ为速度向量
Figure PCTCN2017079180-appb-000015
与障碍物中心坐标和当前位置坐标差向量(o-y)的夹角，见式(14)，R为旋转矩阵决定运动轨迹绕障碍物的旋转方向，k和β为常量，这里取k＝1000，β＝20/π。耦合因子中加入了一个垂直于当前速度方向的运动，是距离矢量与速度矢量的函数。
Figure PCTCN2017079180-appb-000016
Figure PCTCN2017079180-appb-000017
步骤S32：在每个自由度各自的动力学方程中，均添加各自的耦合因子，如式(12)。根据各个自由度在步骤S1中所提取的运动基元，设定好起点A″和终点B″坐标(以上升过程为例)， 由式(12)泛化出三个自由度的轨迹序列，从而三个自由度的序列组成四旋翼无人机三维的避障轨迹，进而无人机根据规划路径完成电塔巡检任务，如图3。
综上所述，本发明所提出四旋翼无人机三维空间的电塔巡检方法，是一种基于运动学习的轨迹规划方法，通过对一次具体的电塔巡检任务的学习，对于新的电塔巡检任务能够自主规划出可行路径提供给四旋翼无人机，具有很高的工程实用价值。
本发明保护范围不仅局限于以上所述的较佳实施方式，凡在与本发明相同原理下的变更或润饰均应包含在本发明的保护范围之内。

Claims (5)

  1. 一种基于运动学习的四旋翼无人机电塔巡检方法，其特征在于具体包括如下步骤：步骤S1：运动学习过程：通过遥控器示教四旋翼无人机完成一次具体的电塔巡检任务，收集该具体的电塔巡检任务三维轨迹序列作为运动学习样本，基于动力学方程和运动学习样本提取运动基元；所述三维轨迹序列即三个自由度的位移、速度、加速度信息；步骤S2：泛化过程：为新的电塔巡检任务设定起点终点，根据动力学方程和提取的运动基元泛化出新巡检任务三维轨迹序列，将产生的三维轨迹序列作为导航路径给四旋翼无人机完成电塔巡检任务，若新的电塔巡检任务中没有障碍物则步骤S2后规划结束；步骤S3：若新的电塔巡检任务的规划路径上有障碍物，首先确定障碍物大致的中心位置坐标，然后在已有学习基础上通过设计耦合因子重新规划出可行三维避障轨迹；最后将规划出的可行避障轨迹给四旋翼无人机完成自主电塔巡检任务。
  2. 根据权利要求1所述的基于运动学习的四旋翼无人机电塔巡检方法，其特征在于：步骤S1提取运动基元时，引入带有恒定系数线性微分方程，将四旋翼无人机的点到点运动描述为一种带有非线性强制项的动力学系统模型；通过遥控器示教四旋翼无人机完成一次具体的电塔巡检任务，基于四旋翼无人机内部的位置估计模块，分别获取四旋翼无人机上升阶段和下降阶段起点到终点三个自由度各自位移、速度和加速度的序列，基于获得的学习样本和动力学系统模型分别提取四旋翼无人机上升和下降阶段的运动基元序列，提取到的基元序列作为后续步骤中自主规划路径的基础。
  3. 根据权利要求1所述的基于运动学习的四旋翼无人机电塔巡检方法，其特征在于：步骤S2中，设定新的电塔巡检任务起点和终点，将学到的运动基元带入有非线性强制项的动力学系统模型，并使用同一正则系统，泛化出一条离散的运动轨迹点，即为所需的电塔巡检路径，若新的巡检任务无障碍物则步骤S2后规划结束，将规划的轨迹点给四旋翼无人机完成电塔巡检任务；
  4. 根据权利要求1所述的基于运动学习的四旋翼无人机电塔巡检方法，其特征在于：步骤S3通过在动力学方程中设计耦合因子，在原有的运动基元和动力学方程基础上重新泛化出可行三维轨迹序列，将三维轨迹序列作为电塔巡检的导航路径给四旋翼无人机完成自主电塔巡检任务。
  5. 根据权利要求4所述的基于运动学习的四旋翼无人机电塔巡检方法，其特征在于：步骤S3详细步骤如下：
    步骤S31：设定在上升阶段和下降阶段，在三维笛卡尔空间中从起点到目标点的运动过程中有障碍物，首先确定障碍物大致中心为o＝[o1 o2 o3]T，在带有非线性强制项的动力学系统模型中加入耦合因子Ct从而构建带避障功能的动力学系统；对于三个自由度的运动，各自带避障功能的动力学系统有其各自的耦合因子Ct＝[Ct，1 Ct，2 Ct，3]T，Ct，j(j＝1，2，3)；耦合因子中加入了一个垂直于当前速度方向的运动，是距离矢量与速度矢量的函数；其中
    Figure PCTCN2017079180-appb-100001
    Figure PCTCN2017079180-appb-100002
    其中μ为速度向量
    Figure PCTCN2017079180-appb-100003
    与障碍物中心坐标和当前位置坐标差向量(o-y)的夹角；R为旋转矩阵，决定运动轨迹绕障碍物的旋转方向，k和β为常量。
    步骤S32：在三个自由度各自的动力学方程中，均添加各自的耦合因子构建带避障功能的动力学系统；根据各个自由度在步骤S1中所提取的运动基元，设定好起点和终点，由带避障功能的动力学系统泛化出各自自由度的轨迹序列，从而三个自由度的序列组成四旋翼无人机三维的避障轨迹，进而无人机根据规划路径完成电塔巡检任务。
PCT/CN2017/079180 2016-12-09 2017-04-01 一种基于运动学习的四旋翼无人机电塔巡检方法 WO2018103242A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201611126213.4A CN106774389A (zh) 2016-12-09 2016-12-09 一种基于运动学习的四旋翼无人机电塔巡检方法
CN201611126213.4 2016-12-09

Publications (1)

Publication Number Publication Date
WO2018103242A1 true WO2018103242A1 (zh) 2018-06-14

Family

ID=58877654

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/079180 WO2018103242A1 (zh) 2016-12-09 2017-04-01 一种基于运动学习的四旋翼无人机电塔巡检方法

Country Status (2)

Country Link
CN (1) CN106774389A (zh)
WO (1) WO2018103242A1 (zh)

Cited By (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109520504A (zh) * 2018-11-27 2019-03-26 北京航空航天大学 一种基于网格离散化的无人机巡查路径优化方法
CN110470305A (zh) * 2019-08-23 2019-11-19 沈阳航空航天大学 一种动态环境下无人机避障航迹规划方法
CN110825110A (zh) * 2019-11-13 2020-02-21 昆明能讯科技有限责任公司 一种电力线路可见光点云解算照片的采集飞行方法
CN111008467A (zh) * 2019-11-29 2020-04-14 天津大学 无人机吊挂飞行运输系统的减摆轨迹在线规划方法
CN111044044A (zh) * 2019-12-06 2020-04-21 国网安徽省电力有限公司淮南供电公司 一种电力无人机巡检路线规划方法和装置
CN111766897A (zh) * 2020-07-10 2020-10-13 广东电网有限责任公司 一种输电线路的通道巡视方法、无人机和系统
CN111982123A (zh) * 2020-08-26 2020-11-24 云南电网有限责任公司迪庆供电局 无人机巡检航线规划方法及装置
CN112179397A (zh) * 2019-07-01 2021-01-05 内蒙古中煤蒙大新能源化工有限公司 一种化工装置无人巡检方法
CN112230678A (zh) * 2020-10-29 2021-01-15 皖江工学院 基于粒子群算法的三维无人机路径规划方法及规划系统
CN112836292A (zh) * 2021-01-15 2021-05-25 中国航空工业集团公司沈阳飞机设计研究所 一种飞机通用质量特性论证方法
CN113205116A (zh) * 2021-04-15 2021-08-03 江苏方天电力技术有限公司 输电线路无人机巡检拍摄目标点自动提取及航迹规划方法
CN113343355A (zh) * 2021-06-08 2021-09-03 四川大学 基于深度学习的飞机蒙皮型面检测路径规划方法
CN113467469A (zh) * 2021-07-23 2021-10-01 中国核动力研究设计院 一种基于bim技术的物项吊运空间轨迹规划方法及系统
CN113554781A (zh) * 2021-08-02 2021-10-26 广东电网有限责任公司佛山供电局 架空输电线路远程巡检方法、装置和计算机设备
CN113741490A (zh) * 2020-05-29 2021-12-03 广州极飞科技股份有限公司 一种巡检方法、装置、飞行器及存储介质
CN113759977A (zh) * 2021-09-18 2021-12-07 西北工业大学 基于优化的绳系多无人机协同搬运的避障轨迹规划方法
CN114089770A (zh) * 2021-11-23 2022-02-25 广东电网有限责任公司 一种变电站巡检机器人的巡检点位生成方法及相关装置
CN114104334A (zh) * 2021-12-09 2022-03-01 中国民航大学 一种基于运动小车自动图像识别方式的飞机检查方法
CN114397909A (zh) * 2021-12-29 2022-04-26 西北工业大学 一种针对大型飞机的小型无人机自动巡检方法
CN115146882A (zh) * 2022-09-06 2022-10-04 南京信息工程大学 一种空地协同巡检方法及系统
CN115275870A (zh) * 2022-09-28 2022-11-01 合肥优晟电力科技有限公司 一种基于高空线路维护的巡检系统
CN115639840A (zh) * 2022-11-09 2023-01-24 海南斯兰低碳研究中心有限公司 一种基于光伏电站运维分析的无人机巡检方法
CN115826596A (zh) * 2022-09-19 2023-03-21 国家能源集团宝庆发电有限公司 基于多旋翼无人机的智能火电厂烟囱巡检方法及系统
CN117876624A (zh) * 2024-03-13 2024-04-12 纵古(江苏)智能科技有限公司 基于无人机遥感影像的复杂环境航迹规划方法
CN117784817B (zh) * 2024-02-27 2024-05-17 北京航空航天大学 一种陆空两栖无人平台一体化规划控制系统与方法

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109032175B (zh) * 2018-07-13 2021-08-06 辽宁东科电力有限公司 基于阿基米德螺旋线的无人机电力巡检路径计算方法
CN108955645A (zh) * 2018-07-16 2018-12-07 福州日兆信息科技有限公司 应用于通信铁塔智能巡检的三维建模方法及装置
CN110727271A (zh) * 2019-10-30 2020-01-24 北京科技大学 一种机器人运动基元确定方法及装置
CN114326776A (zh) * 2021-04-16 2022-04-12 贵州电网有限责任公司 一种具有防撞避障功能的智能化电力线巡检无人机

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102510011A (zh) * 2011-10-24 2012-06-20 华北电力大学 基于微型多旋翼无人直升机的电力杆塔智能巡检方法
US20120158280A1 (en) * 2008-01-14 2012-06-21 Ravenscroft Donald L Computing route plans for routing around obstacles having spatial and temporal dimensions
CN104865971A (zh) * 2015-05-26 2015-08-26 广西大学 一种输电线路巡检无人机的控制方法及无人机
US20150336669A1 (en) * 2014-05-20 2015-11-26 Verizon Patent And Licensing Inc. Unmanned aerial vehicle network-based recharging
CN105955291A (zh) * 2016-04-29 2016-09-21 深圳市哈博森科技有限公司 一种无人机飞行航线轨迹记录与自动飞行控制方式

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6490521B2 (en) * 2000-12-28 2002-12-03 Intel Corporation Voice-controlled navigation device utilizing wireless data transmission for obtaining maps and real-time overlay information
JP4169263B2 (ja) * 2003-04-07 2008-10-22 株式会社バンダイナムコゲームス 画像生成装置及び情報記憶媒体
CN104648665B (zh) * 2015-01-30 2016-12-07 上海交通大学 一种用于巡线的四旋翼无人机装置
CN104648685B (zh) * 2015-02-12 2016-08-24 武汉科技大学 基于智能手机的四旋翼飞行器指定路径航拍系统和方法
CN105045260A (zh) * 2015-05-25 2015-11-11 湖南大学 一种未知动态环境下的移动机器人路径规划方法
CN104978580B (zh) * 2015-06-15 2018-05-04 国网山东省电力公司电力科学研究院 一种用于无人机巡检输电线路的绝缘子识别方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120158280A1 (en) * 2008-01-14 2012-06-21 Ravenscroft Donald L Computing route plans for routing around obstacles having spatial and temporal dimensions
CN102510011A (zh) * 2011-10-24 2012-06-20 华北电力大学 基于微型多旋翼无人直升机的电力杆塔智能巡检方法
US20150336669A1 (en) * 2014-05-20 2015-11-26 Verizon Patent And Licensing Inc. Unmanned aerial vehicle network-based recharging
CN104865971A (zh) * 2015-05-26 2015-08-26 广西大学 一种输电线路巡检无人机的控制方法及无人机
CN105955291A (zh) * 2016-04-29 2016-09-21 深圳市哈博森科技有限公司 一种无人机飞行航线轨迹记录与自动飞行控制方式

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
CHEN, PENGZHEN ET AL.: "Path Planning for Micro Quadrotor UAVs Based on Dynamic Movement Primitives", CHINESE HIGH TECHNOLOGY LETTERS, vol. 26, no. 2, 15 February 2016 (2016-02-15), ISSN: 1002-0470 *

Cited By (40)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109520504B (zh) * 2018-11-27 2022-07-22 北京航空航天大学 一种基于网格离散化的无人机巡查路径优化方法
CN109520504A (zh) * 2018-11-27 2019-03-26 北京航空航天大学 一种基于网格离散化的无人机巡查路径优化方法
CN112179397A (zh) * 2019-07-01 2021-01-05 内蒙古中煤蒙大新能源化工有限公司 一种化工装置无人巡检方法
CN110470305A (zh) * 2019-08-23 2019-11-19 沈阳航空航天大学 一种动态环境下无人机避障航迹规划方法
CN110825110A (zh) * 2019-11-13 2020-02-21 昆明能讯科技有限责任公司 一种电力线路可见光点云解算照片的采集飞行方法
CN111008467A (zh) * 2019-11-29 2020-04-14 天津大学 无人机吊挂飞行运输系统的减摆轨迹在线规划方法
CN111044044A (zh) * 2019-12-06 2020-04-21 国网安徽省电力有限公司淮南供电公司 一种电力无人机巡检路线规划方法和装置
CN111044044B (zh) * 2019-12-06 2023-04-07 国网安徽省电力有限公司淮南供电公司 一种电力无人机巡检路线规划方法和装置
CN113741490A (zh) * 2020-05-29 2021-12-03 广州极飞科技股份有限公司 一种巡检方法、装置、飞行器及存储介质
CN111766897B (zh) * 2020-07-10 2024-02-27 广东电网有限责任公司 一种输电线路的通道巡视方法、无人机和系统
CN111766897A (zh) * 2020-07-10 2020-10-13 广东电网有限责任公司 一种输电线路的通道巡视方法、无人机和系统
CN111982123B (zh) * 2020-08-26 2023-10-31 云南电网有限责任公司迪庆供电局 无人机巡检航线规划方法及装置
CN111982123A (zh) * 2020-08-26 2020-11-24 云南电网有限责任公司迪庆供电局 无人机巡检航线规划方法及装置
CN112230678A (zh) * 2020-10-29 2021-01-15 皖江工学院 基于粒子群算法的三维无人机路径规划方法及规划系统
CN112230678B (zh) * 2020-10-29 2023-07-14 皖江工学院 基于粒子群算法的三维无人机路径规划方法及规划系统
CN112836292A (zh) * 2021-01-15 2021-05-25 中国航空工业集团公司沈阳飞机设计研究所 一种飞机通用质量特性论证方法
CN112836292B (zh) * 2021-01-15 2023-10-13 中国航空工业集团公司沈阳飞机设计研究所 一种飞机通用质量特性论证方法
CN113205116A (zh) * 2021-04-15 2021-08-03 江苏方天电力技术有限公司 输电线路无人机巡检拍摄目标点自动提取及航迹规划方法
CN113205116B (zh) * 2021-04-15 2024-02-02 江苏方天电力技术有限公司 输电线路无人机巡检拍摄目标点自动提取及航迹规划方法
CN113343355A (zh) * 2021-06-08 2021-09-03 四川大学 基于深度学习的飞机蒙皮型面检测路径规划方法
CN113467469B (zh) * 2021-07-23 2024-01-23 中国核动力研究设计院 一种基于bim技术的物项吊运空间轨迹规划方法及系统
CN113467469A (zh) * 2021-07-23 2021-10-01 中国核动力研究设计院 一种基于bim技术的物项吊运空间轨迹规划方法及系统
CN113554781A (zh) * 2021-08-02 2021-10-26 广东电网有限责任公司佛山供电局 架空输电线路远程巡检方法、装置和计算机设备
CN113759977B (zh) * 2021-09-18 2023-07-28 西北工业大学 基于优化的绳系多无人机协同搬运的避障轨迹规划方法
CN113759977A (zh) * 2021-09-18 2021-12-07 西北工业大学 基于优化的绳系多无人机协同搬运的避障轨迹规划方法
CN114089770A (zh) * 2021-11-23 2022-02-25 广东电网有限责任公司 一种变电站巡检机器人的巡检点位生成方法及相关装置
CN114089770B (zh) * 2021-11-23 2024-04-12 广东电网有限责任公司 一种变电站巡检机器人的巡检点位生成方法及相关装置
CN114104334B (zh) * 2021-12-09 2024-02-23 中国民航大学 一种基于运动小车自动图像识别方式的飞机检查方法
CN114104334A (zh) * 2021-12-09 2022-03-01 中国民航大学 一种基于运动小车自动图像识别方式的飞机检查方法
CN114397909A (zh) * 2021-12-29 2022-04-26 西北工业大学 一种针对大型飞机的小型无人机自动巡检方法
CN114397909B (zh) * 2021-12-29 2023-06-30 西北工业大学 一种针对大型飞机的小型无人机自动巡检方法
CN115146882A (zh) * 2022-09-06 2022-10-04 南京信息工程大学 一种空地协同巡检方法及系统
CN115826596B (zh) * 2022-09-19 2023-08-25 国家能源集团宝庆发电有限公司 基于多旋翼无人机的智能火电厂烟囱巡检方法及系统
CN115826596A (zh) * 2022-09-19 2023-03-21 国家能源集团宝庆发电有限公司 基于多旋翼无人机的智能火电厂烟囱巡检方法及系统
CN115275870B (zh) * 2022-09-28 2022-12-06 合肥优晟电力科技有限公司 一种基于高空线路维护的巡检系统
CN115275870A (zh) * 2022-09-28 2022-11-01 合肥优晟电力科技有限公司 一种基于高空线路维护的巡检系统
CN115639840A (zh) * 2022-11-09 2023-01-24 海南斯兰低碳研究中心有限公司 一种基于光伏电站运维分析的无人机巡检方法
CN117784817B (zh) * 2024-02-27 2024-05-17 北京航空航天大学 一种陆空两栖无人平台一体化规划控制系统与方法
CN117876624A (zh) * 2024-03-13 2024-04-12 纵古(江苏)智能科技有限公司 基于无人机遥感影像的复杂环境航迹规划方法
CN117876624B (zh) * 2024-03-13 2024-05-07 纵古(江苏)智能科技有限公司 基于无人机遥感影像的复杂环境航迹规划方法

Also Published As

Publication number Publication date
CN106774389A (zh) 2017-05-31

Similar Documents

Publication Publication Date Title
WO2018103242A1 (zh) 一种基于运动学习的四旋翼无人机电塔巡检方法
Kaufmann et al. Deep drone acrobatics
Bacik et al. Autonomous flying with quadrocopter using fuzzy control and ArUco markers
Sayre-McCord et al. Visual-inertial navigation algorithm development using photorealistic camera simulation in the loop
Nguyen et al. Autonomous navigation in complex environments with deep multimodal fusion network
Sonugür A Review of quadrotor UAV: Control and SLAM methodologies ranging from conventional to innovative approaches
Rodriguez-Ramos et al. Towards fully autonomous landing on moving platforms for rotary unmanned aerial vehicles
Zhao et al. Vision-based tracking control of quadrotor with backstepping sliding mode control
Fu et al. Autonomous landing of a quadrotor on an UGV
Giernacki et al. DJI tello quadrotor as a platform for research and education in mobile robotics and control engineering
Baldini et al. Learning pose estimation for UAV autonomous navigation and landing using visual-inertial sensor data
Cao et al. UAV path planning based on improved particle swarm algorithm
Wu et al. Vision-based target detection and tracking system for a quadcopter
Lin et al. Robust image-based landing control of a quadrotor on an unpredictable moving vehicle using circle features
Leong et al. Vision-based sense and avoid with monocular vision and real-time object detection for uavs
Luo et al. Biomimetic design for unmanned aerial vehicle safe landing in hazardous terrain
Erskine et al. Model predictive control for dynamic quadrotor bearing formations
Lee et al. Receding horizon-based RRT* algorithm for a UAV real-time path planner
Yubo et al. Survey of UAV autonomous landing based on vision processing
Shastry et al. Autonomous detection and tracking of a high-speed ground vehicle using a quadrotor UAV
Chen et al. Real-time active detection of targets and path planning using UAVs
Dai et al. Unsupervised feature fusion combined with neural network applied to UAV attitude estimation
Cao et al. Research on application of computer vision assist technology in high-precision UAV navigation and positioning
Kou et al. Autonomous Navigation of UAV in Dynamic Unstructured Environments via Hierarchical Reinforcement Learning
Xiao et al. Autonomous Tracking and Landing of QUAV Based on Air-Ground Cooperation

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17878210

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17878210

Country of ref document: EP

Kind code of ref document: A1