Rotorcraft unmanned aerial vehicle hoist system load state estimation method, apparatus, and medium

Abstract

This invention relates to a method, device, and medium for estimating the load state of a rotary-wing unmanned aerial vehicle (UAV) hoisting system. The method includes: Step S1, establishing a dynamic model of the rotary-wing UAV hoisting system based on Newton's laws of motion; Step S2, using a vision-based real-time tag detection algorithm to identify tags affixed to the load, and outputting the coordinates of the tags under the UAV system, i.e., the visual positioning result, based on a camera model and rope length constraints; Step S3, combining the visual positioning result output in Step S2 with the dynamic model of the rotary-wing UAV hoisting system in Step S1, and using a hybrid extended Kalman filter to perform real-time load state estimation. Compared with existing technologies, this invention has the advantage of high prediction accuracy.

Description

Technical Field

[0001] This invention relates to the field of air cargo transport status estimation, and in particular to a method, device and medium for estimating the load status of a rotary-wing unmanned aerial vehicle (UAV) lifting system. Background Technology

[0002] In recent years, multi-rotor drones have been widely used in the field of material transportation due to their more agile flight capabilities, more precise control, and vertical takeoff and landing characteristics. Moreover, unlike manned rotorcraft, multi-rotor drones have advantages in terms of manpower risks and transportation costs. When transporting materials, there are two ways to connect the payload to the drone: one is to fix the object to the drone body, and the other is to suspend the payload from below the drone using ropes. Generally speaking, the former imposes more restrictions on the shape, size, and mass of the payload, especially when transporting large components, often making the drone extremely cumbersome. While the former increases the system's degrees of freedom and underactuation, it has fewer restrictions on the shape, size, and mass of the payload. For large components, multiple drones can be used in a coordinated lifting manner to complete the transportation without affecting the drone's maneuverability, thus showing better development prospects.

[0003] Compared to existing fixed transport systems for rotary-wing UAVs, rotary-wing UAV sling load systems have relatively relaxed restrictions on payload weight and shape, and their application scenarios are also wider. However, sling load systems introduce new problems, one of which is how to effectively observe the payload's state. Currently, a series of methods for observing the payload's state have emerged both domestically and internationally. For example, some studies have used cables to connect to gimbals with angle encoders, and inertial measurement units to detect the relative attitude between the payload and the UAV. However, although these methods can solve the problem to some extent, they still have some limitations. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a more accurate method, device and medium for estimating the load status of a rotary-wing UAV hoisting system.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] According to a first aspect of the present invention, a method for estimating the load state of a rotary-wing unmanned aerial vehicle (UAV) lifting system is provided, the method comprising:

[0007] Step S1: Establish a dynamic model of the rotary-wing UAV hoisting system based on Newton's laws of motion;

[0008] Step S2: Use a vision-based real-time label detection algorithm to identify the labels pasted on the load, and output the coordinates of the labels under the UAV system, i.e., the visual positioning result, based on the camera model and rope length constraints.

[0009] Step S3: Combining the visual positioning results output in step S2 with the dynamic model of the rotary-wing UAV hoisting system in step S1, a hybrid extended Kalman filter is used to estimate the load state in real time.

[0010] Preferably, the assumptions made about the quadcopter drone and payload for a fixed rope length include:

[0011] The drone is a single rigid body, with its geometric center coinciding with its center of gravity.

[0012] The load is a single mass point, and its shape is not considered.

[0013] The rope is massless, inelastic, and taut, with the drone's center of mass and a point load at each end.

[0014] Preferably, in step S1, a dynamic model of the rotary-wing UAV hoisting system is established based on Newton's laws of motion, and the expression is:

[0015] p L =v L

[0016]

[0017] q = -[q] × ω

[0018]

[0019] R = R[ω] ×

[0020] in: For load location, For load speed, unit vector The direction of the rope, It is the thrust generated by the quadcopter motor. Let e3 be the rotation matrix from the UAV coordinate system to the inertial system, where e3 = [00 1]. T g is the acceleration due to gravity. It's the quality of the quadcopter drone. It is the quality of the load; It is the angular velocity of the load rotating around the center of mass of the UAV, and it is always perpendicular to q, satisfying q T ω=0, [ω] × This indicates the search for the antisymmetric matrix of the vectors inside the brackets.

[0021] Preferably, step S2 includes the following sub-steps:

[0022] Step S21: Fix a camera to the bottom of the drone, attach a label to the payload, capture images containing the payload label through the camera, and use a real-time label detection algorithm to perform feature detection on the images;

[0023] Step S22: Based on the feature positions in the image, and combined with the camera model and rope length constraints, calculate the coordinates of the load in the UAV body coordinate system, i.e., the visual positioning result.

[0024] Preferably, in step S22, the coordinates of the load in the UAV body coordinate system are calculated based on the feature positions in the image, combined with the camera model and rope length constraints, which is the visual positioning result. The specific calculation process is as follows:

[0025] 1) The transformation expression for point coordinates from the world coordinate system (X,Y,Z,1) to the image coordinate system (u,v,1) is:

[0026]

[0027] Where: the T matrix is ​​the extrinsic parameter matrix of the camera, describing the pose relationship between the camera and the world coordinate system, and its intrinsic parameter f x ,f y u0, v0 are calculated based on the relative pose of the camera and the UAV; the K matrix is ​​the camera's intrinsic parameter matrix, whose intrinsic parameters... Obtained through camera intrinsic parameter calibration;

[0028] 2) The position of the load relative to the camera coordinate system The expressions for the constraint on rope length l are as follows:

[0029]

[0030]

[0031] in: This represents the camera's position relative to the drone's coordinate system. Let be the rotation matrix from the camera coordinate system to the UAV coordinate system, and let be a known constant matrix. Solve simultaneously to find λ and .

[0032] 3) Calculate the coordinates of the mounting point in the UAV coordinate system.

[0033]

[0034] Preferably, step S3 specifically involves: obtaining the current state and control variables of the UAV from the UAV controller based on the coordinates of the tag under the UAV system obtained in step S2; constructing a hybrid extended Kalman filter by combining the dynamic model of the rotary UAV hoisting system in step S1; and using a specific numerical integration method of the state vector and error covariance matrix during the filtering process to complete the real-time estimation of the load state.

[0035] Preferably, the design process of the hybrid extended Kalman filter includes:

[0036] Define the state vector and system input of the hybrid extended Kalman filter:

[0037]

[0038] Define the state transition equation:

[0039]

[0040] Where: w and v are both Gaussian noise that follows a normal distribution;

[0041] Based on the dynamic model of the rotary-wing UAV hoisting system, the observation equations are determined as follows:

[0042]

[0043] Preferably, the specific numerical integration method for the state vector and the error covariance matrix during the filtering process includes: integrating the state vector using the second-order Runge-Kutta integration method, and integrating the error covariance matrix using the transition matrix method.

[0044] According to a second aspect of the present invention, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the program to implement any of the methods described above.

[0045] According to a third aspect of the invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements any of the methods described herein.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] 1) This invention obtains load state information through visual inspection, which can obtain high-precision position and velocity information. At the same time, the Kalman filter can filter noise, which improves the estimation accuracy of the method.

[0048] 2) By combining system dynamics, the load state estimation algorithm based on visual detection and Kalman filter in this invention can better understand the changing law of load state, has higher robustness and accuracy, and can better cope with various complex situations in practical applications.

[0049] 3) This invention uses non-contact measurement, which does not require installing sensors on the object being measured. The load state can be estimated simply by visual inspection, avoiding the trouble of sensor installation and wiring, as well as the impact of contact sensing on the suspended model.

[0050] 4) This method uses a Kalman filter to estimate the load state, which has the characteristics of strong real-time performance and can meet the real-time requirements in practical applications. Attached Figure Description

[0051] Figure 1 A diagram defining the coordinate system and vectors for a quadcopter UAV sling load system.

[0052] Figure 2 This is a diagram illustrating the imaging principle of a camera.

[0053] Figure 3 This is a flowchart of the method of the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0055] Example 1

[0056] like Figure 3 As shown in the figure, this embodiment provides a method for estimating the load state of a rotary-wing unmanned aerial vehicle (UAV) lifting system. The method includes the following steps:

[0057] Step S1: Establish a dynamic model of the quadcopter UAV hoisting system based on Newton's second law;

[0058] Step S2: Use a vision-based real-time label detection algorithm to identify the labels pasted on the load, and transform their coordinates to the UAV body coordinate system according to the camera model and rope length constraints, and output the visual positioning results.

[0059] Step S3: Combining the visual positioning results output in step S2 with the dynamic model of the rotary-wing UAV hoisting system in step S1, a hybrid extended Kalman filter is used to estimate the load state in real time.

[0060] In this embodiment, the following assumptions are made regarding the quadcopter drone and the load when the rope length is fixed: 1) The drone is a simple rigid body with its geometric center coinciding with its center of gravity. 2) The load is a single mass point, and its shape is not considered. 3) The rope is massless, inelastic, and taut, with the drone's center of mass and the load at its two ends, respectively.

[0061] Next, each step will be explained in detail.

[0062] 1. Establish a dynamic model of the UAV hoisting system.

[0063] like Figure 1 As shown, the modeling process uses three coordinate systems, including:

[0064] (1) A body coordinate system, which is fixed to the UAV body, and uses... express;

[0065] (2) Camera coordinate system, which is fixed to the camera, with its origin at the center of the sensor chip surface. The two axes are parallel to the horizontal and vertical alignment directions of the sensor's photosensitive units, and are determined by the right-hand rule. The coordinate system is represented as

[0066] (3) Inertial frame of reference, denoted as

[0067] From the relationship between vectors, we can conclude that:

[0068] lq = p L -p Q (1)

[0069] in, The load is located, and l is the rope length, a unit vector. Indicates the direction of the rope.

[0070] The kinematics of the load can be expressed as follows:

[0071]

[0072] in, It is the angular velocity of the load rotating around the drone's center of mass, and it is always perpendicular to q. Because Therefore, there is q. T ω=0, [ω] × This indicates the search for the antisymmetric matrix of the vectors inside the brackets.

[0073] Since the combined lift of the four motors is always along the z-axis of the UAV's body coordinate system, and the mounting point is at the UAV's center of mass, the following particle dynamics equations exist:

[0074]

[0075]

[0076] in, and These are the speeds of the quadcopter and the load, respectively. It's the quality of the quadcopter drone. This refers to the mass of the load, e3 = [0 0 1] T , It is the thrust generated by the quadcopter motor. It is the tension in the rope. Let g be the rotation matrix from the UAV coordinate system to the inertial frame, and g be the gravitational acceleration.

[0077] Solve using equations (1) and (2) respectively. By combining them, we can obtain:

[0078]

[0079] Adding equations (3) and (4), taking the first derivative of both sides of equation (1), and combining these two equations, we get:

[0080]

[0081] By using equations (3) and (4) Expressing this and substituting it into equation (5), we get:

[0082]

[0083] In summary

[0084]

[0085] At this point, the model of the UAV mounting system has been established, and the above equations will be used in the subsequent Kalman filter design.

[0086] 2. Visual inspection

[0087] The first part completed the mathematical modeling of the quadcopter UAV's payload system. It can be observed that, despite the position of the load... Three-dimensional coordinates are needed for representation, but due to the constraints of the rope, there are actually only two degrees of freedom. Therefore, in this embodiment, an efficient tag detection and localization algorithm was designed based on Whycon tags, realizing tag detection and localization functions with advantages such as high dynamic performance, high accuracy, and adaptive lighting.

[0088] In a camera system, there are three coordinate systems: the image coordinate system, the camera coordinate system, and the world coordinate system, such as... Figure 2As shown. The world coordinate system is a three-dimensional coordinate system used to represent the position and orientation of objects in the real world. In photography, the world coordinate system can be viewed as the coordinate system in which objects in the scene are located. The camera coordinate system is a coordinate system relative to the camera itself, with its origin usually located at the optical center of the camera (the intersection of the optical axes). The three axes of the camera coordinate system are usually defined as orthogonal axes; for example, the X-axis points to the right of the camera, the Y-axis points to the bottom of the camera, and the Z-axis points to the front of the camera. The image coordinate system is a two-dimensional coordinate system used to represent the pixel positions on the image plane after the camera has captured the image. The origin of the image coordinate system is usually located at the top left corner of the image, with the X-axis extending to the right and the Y-axis extending downwards. The imaging process of a camera can be described by two parts: rigid body transformation and pinhole imaging, representing the transformation of the representation of points from the world coordinate system to the camera coordinate system and from the camera coordinate system to the image coordinate system, respectively.

[0089] Considering the translation and rotation relationship between the camera coordinate system and the world coordinate system, such as... Figure 2 As shown, based on the relevant conclusions of rigid body transformation, we have

[0090]

[0091] in It is the rotation matrix from the world coordinate system to the camera coordinate system. This represents the vector pointing from the origin of the world coordinate system to the origin of the camera coordinate system in the camera coordinate system. The camera coordinate system has its origin at the pinhole position and its principal optical axis as the Z-axis. Based on the pinhole imaging principle and considering the principal point offset, we have the following projection equation:

[0092]

[0093] in, To account for the lens focal length after different scaling factors in two directions, It considers the coordinates of the point on the image plane after the principal point is offset. Offset of principal points in two directions, such as Figure 2 , is the scale factor.

[0094] According to equations (9) and (10), the transformation of point coordinates from the world coordinate system to the image coordinate system can be obtained, as shown in the following equation.

[0095]

[0096] In the formula, T is also called the camera's extrinsic parameter matrix, which describes the pose relationship between the camera and the world coordinate system. It can be calculated through the relative pose relationship between the camera and the UAV. The K matrix is ​​also called the camera's intrinsic parameter matrix. It contains information on focal length, principal point offset, and scaling factors in both directions. It describes the transformation from the camera coordinate system to the image coordinate system through pinhole imaging. It can be obtained through the calibration of the camera's intrinsic parameters.

[0097] According to equation (10), we can obtain

[0098]

[0099] in, This is the position of the load relative to the camera coordinate system. For example... Figure 1 As shown, there are

[0100]

[0101] in, This represents the camera's position relative to the drone's coordinate system. The coordinates of the mount point in the UAV coordinate system are: Let be the rotation matrix from the camera coordinate system to the UAV coordinate system, which is a known constant matrix.

[0102] Furthermore, since the load and the drone are connected via a cable of fixed length, they are constrained by the cable length.

[0103]

[0104] By combining equations (12) and (14), λ can be solved by solving a quadratic equation in one variable, and the result can be obtained. Substituting the result back into equation (13) will give the final result.

[0105] 3. Design of Kalman filters considering system dynamics.

[0106] Considering that outdoor lighting may change drastically and the load may exceed the camera's field of view as the rope swings, this paper uses a Kalman filter and combines system dynamics predictions with camera observations to give an estimate of the load state.

[0107] Multiply both sides of equation (7) by [q] × , can be obtained

[0108]

[0109] Next, we define the filter state and system input.

[0110]

[0111] According to equation (15), the state transition equation can be written in the following form.

[0112]

[0113] Where w and v are both Gaussian noises that follow a normal distribution.

[0114] Based on the UAV-borne model, the observation equations can be written in the following form.

[0115]

[0116] Observations reveal that the state transition equation is continuous, while the observations are discrete due to execution in each algorithm loop, and the system is highly nonlinear. For nonlinear systems, an extended Kalman filter can be used. This method linearizes the system dynamics model using the Jacobian matrix to approximate the nonlinear system and utilizes the prediction and correction steps of the Kalman filter for state estimation. For the case where the state transition is continuous but the observations are discrete, a hybrid Kalman filter can be used.

[0117] Therefore, the Kalman filter for the load observation system is constructed as follows:

[0118]

[0119]

[0120]

[0121]

[0122]

[0123]

[0124]

[0125]

[0126]

[0127]

[0128]

[0129] A 11 =diag(σ 11 ,σ 22 ,v 33 (30)

[0130]

[0131] A 21 =I 3×3 (32)

[0132]

[0133]

[0134] The state updates and covariance matrix updates of the prediction part of the dynamics section within the filter are given in continuous form, while the algorithm of this invention is calculated only once per time step. Therefore, the second-order Runge-Kutta integral method is used to update the state vector. Integrating, the update law is:

[0135]

[0136] Where h is the sampling time. Since P is a 6x6 matrix, direct numerical integration requires a lot of computation. In this embodiment, the transition matrix method is used for integration, and the update law is as follows:

[0137] P(t+h)=[φ 11 (h)P(t)+φ 12 (h)][φ 21 (h)P(t)+φ 22 (h)] -1 (36)

[0138] in, φ is the transition matrix.

[0139] In this embodiment, a quadcopter drone carries a load, and the drone controller controls the drone body to track the trajectory in three-dimensional space, without controlling the trajectory of the load.

[0140] The control law of the drone controller is as follows:

[0141]

[0142] M=K r (rr des )+K ω (ω-ω des )

[0143] in, Location of the drone. For the drone's linear velocity, For the UAV's Euler angles relative to the inertial frame, For the angular velocity of the drone, all four quantities can be obtained from simulation software.

[0144]

[0145] The desired motion trajectory determined in this embodiment is as follows:

[0146] x des = [sin(0.02t)cos(0.02t)-1 2] T ,ω des =v des =0

[0147] During the simulation, the load began to sway as the drone flew in space. The vision algorithm detected the labels pasted on the load in real time and output their positions. At each time step, the Kalman filter acquired the results of the vision algorithm, the state of the drone itself, and the control output of the controller, and performed the Kalman filtering process. Finally, it output an estimate of the load state and compared it with the load state obtained directly from the simulation environment. It was found that the waveforms of the two basically overlapped.

[0148] The electronic device of this invention includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in read-only memory (ROM) or loaded from a storage unit into random access memory (RAM). The RAM may also store various programs and data required for device operation. The CPU, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0149] Multiple components in the device are connected to the I / O interface, including: input units such as keyboards and mice; output units such as various types of displays and speakers; storage units such as disks and optical discs; and communication units such as network interface cards (NICs), modems, and wireless transceivers. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0150] The processing unit executes the various methods and processes described above, such as methods S1 to S3. For example, in some embodiments, methods S1 to S3 may be implemented as computer software programs tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via ROM and / or a communication unit. When the computer program is loaded into RAM and executed by the CPU, one or more steps of methods S1 to S3 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S3 by any other suitable means (e.g., by means of firmware).

[0151] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload programmable logic devices (CPLDs), and so on.

[0152] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0153] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0154] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for estimating the load state of a rotary-wing unmanned aerial vehicle (UAV) lifting system, characterized in that, The method includes: Step S1: Establish a dynamic model of the rotary-wing UAV hoisting system based on Newton's laws of motion; Step S2: Use a vision-based real-time label detection algorithm to identify the labels pasted on the load, and output the coordinates of the labels under the UAV tether, i.e., the visual positioning result, based on the camera model and rope length constraints. This includes the following sub-steps: Step S21: Fix a camera to the bottom of the drone, attach a label to the payload, capture images containing the payload label through the camera, and use a real-time label detection algorithm to perform feature detection on the images; Step S22: Based on the feature positions in the image, and combined with the camera model and rope length constraints, calculate the coordinates of the load in the UAV body coordinate system, i.e., the visual positioning result. In step S22, based on the feature positions in the image and combined with the camera model and rope length constraints, the coordinates of the load in the UAV body coordinate system are calculated, i.e., the visual positioning result. The specific calculation process is as follows: 1) Point coordinates from the world coordinate system To the image coordinate system The conversion expression is: ， in: The matrix is ​​the camera's extrinsic parameter matrix, which describes the pose relationship between the camera and the world coordinate system; its intrinsic parameters... It is calculated through the relative pose relationship between the camera and the drone; The matrix is ​​the intrinsic parameter matrix of the camera, and its intrinsic parameters... Obtained through camera intrinsic parameter calibration; Scale factor; 2) The position of the load relative to the camera coordinate system and rope length The constraint expressions are as follows: ， ， in: This represents the camera's position relative to the drone's coordinate system. Let be the rotation matrix from the camera coordinate system to the UAV coordinate system, and be a known constant matrix. Solve the system of equations simultaneously to obtain... and ; 3) Calculate the coordinates of the mounting point in the UAV coordinate system. : ； Step S3: Combining the visual positioning results output in step S2 with the dynamic model of the rotary-wing UAV hoisting system in step S1, a hybrid extended Kalman filter is used to estimate the load state in real time, specifically as follows: Based on the coordinates of the tag under the UAV system obtained in step S2, the current state and control variables of the UAV are obtained from the UAV controller. Combined with the dynamic model of the rotary UAV hoisting system in step S1, a hybrid extended Kalman filter is constructed. By using a specific numerical integration method of the state vector and error covariance matrix during the filtering process, the real-time estimation of the load state is completed.

2. The method for estimating the load state of a rotary-wing unmanned aerial vehicle (UAV) lifting system according to claim 1, characterized in that, The assumptions made for the quadcopter drone and payload with a fixed rope length include: The drone is a single rigid body, with its geometric center coinciding with its center of gravity. The load is a single mass point, and its shape is not considered. The rope is massless, inelastic, and taut, with the drone's center of mass and a point load at each end.

3. The method for estimating the load state of a rotary-wing unmanned aerial vehicle (UAV) lifting system according to claim 2, characterized in that, In step S1, a dynamic model of the rotary-wing UAV hoisting system is established based on Newton's laws of motion, and the expression is: ， in: For load location, For load speed, unit vector The direction of the rope, It is the thrust generated by the quadcopter motor. Let be the rotation matrix from the UAV coordinate system to the inertial frame. , It is gravitational acceleration. It's the quality of the quadcopter drone. It is the quality of the load; It is the angular velocity of the load rotating around the drone's center of mass, and it is always perpendicular to the axis of rotation. ,satisfy , This indicates the search for the antisymmetric matrix of the vectors inside the brackets.

4. The method for estimating the load state of a rotary-wing unmanned aerial vehicle (UAV) lifting system according to claim 1, characterized in that, The design process of the hybrid extended Kalman filter includes: Define the state vector and system input of the hybrid extended Kalman filter: ， Where: unit vector The direction of the rope, It is the thrust generated by the quadcopter motor. Let be the rotation matrix from the UAV coordinate system to the inertial frame. ; Define the state transition equation: ， in: For Gaussian noise that follows a normal distribution, It refers to the mass of the quadcopter drone; Based on the dynamic model of the rotary-wing UAV hoisting system, the observation equations are determined as follows: ， in: It is the angular velocity of the load rotating around the drone's center of mass, and it is always perpendicular to the axis of rotation. ,satisfy ; It is Gaussian noise that follows a normal distribution.

5. The method for estimating the load state of a rotary-wing unmanned aerial vehicle (UAV) lifting system according to claim 4, characterized in that, The specific numerical integration methods for the state vector and error covariance matrix during the filtering process include: integrating the state vector using the second-order Runge-Kutta integration method and integrating the error covariance matrix using the transition matrix method.

6. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 5.

Images