A robust control method and system for aerial contact robots

Abstract

This invention discloses a robust control method for an aerial contact operation robot, comprising: constructing a contact dynamics model of the aerial contact operation robot; constructing a visual servo controller based on image moments; constructing a visual impedance controller, which outputs a desired image moment based on the error vector between the tracked contact force and the desired contact force; the visual servo controller outputting a generalized control force generated by the aerial contact operation robot based on the desired image moment and the actual contact force; calculating a generalized torque generated by the aerial contact operation robot based on the generalized control force and the desired rotation matrix of the aerial contact operation robot using an attitude controller; and controlling the aerial contact operation robot based on the generalized control force and the generalized torque. This control method effectively improves the robot's operational stability and enables contact operations in areas with weak or absent outdoor GPS signals.

Description

Technical Field

[0001] This invention belongs to the field of aerial work robot technology, specifically relating to a robust control method and system for an aerial contact work robot. Background Technology

[0002] Traditional aerial contact inspection tasks, such as corrosion detection of oil and gas pipelines and crack detection of bridge piers—tasks targeting critical infrastructure—usually require technicians to use hand-operated lifts, cranes, or scaffolding and ropes to access specific inspection points. This is time-consuming, labor-intensive, costly, and dangerous for routine inspections and maintenance. Even a small oil refinery needs to inspect thousands of oil pipelines annually, with at least 30% located in highly inaccessible areas, further increasing the difficulty of manual work. Therefore, there is a strong demand for developing alternatives to manual inspection in aerial contact operations. Aerial robots, composed of rotary-wing drones carrying operational mechanisms, have been widely used in aerial contact inspection tasks due to their dexterity and lightweight nature.

[0003] Traditional underactuated drones require changes in their roll and pitch angles to achieve translational motion. When subjected to external disturbances (such as gusts of wind), their attitude changes to maintain stability, leading to swaying during end-effector operations. Therefore, the contact force of aerial contact robots is significantly affected by external disturbances, impacting the stability of aerial operations. Thus, improving the stability of aerial contact robots during contact operations is crucial. Summary of the Invention

[0004] The purpose of this invention is to provide a robust control method and system for aerial contact operation robots, which effectively improves the stability of aerial operation robots when performing contact operations. Specifically, the method of this invention tracks the contact force by constructing a visual impedance controller, transforming impedance control into contact force tracking control, thus ensuring the stability of the aerial operation robot based on contact force tracking. Furthermore, the method of this invention introduces visual servo control, applying image moments, which does not rely on external position information. Since the control method controls the robot using force and torque, without publishing position, it can perform contact operations in locations with weak or absent outdoor GPS signals (such as the bottom of bridge piers) or indoors (where there is no GPS location information).

[0005] Therefore, the present invention provides the following technical solution:

[0006] On one hand, the present invention provides a robust control method for an aerial contact operation robot, comprising the following steps:

[0007] Construct a contact dynamics model for aerial robots performing contact operations;

[0008] A visual servo controller based on image moments is constructed, wherein image moments representing the position and state information of the aerial robot are defined in the virtual image plane, and the movement of the aerial robot in the virtual image plane is mapped to the Cartesian coordinate system;

[0009] Construct a visual impedance controller to transform impedance control into control that tracks contact force;

[0010] The error between the tracked contact force and the desired contact force is input into the visual impedance controller to obtain the desired image moment s. z,d The visual servo controller is based on the desired image moments s z,d And the generalized control force generated by the aerial work robot is generated by the actual contact force output; based on the generalized control force and the expected rotation matrix of the aerial work robot, the attitude controller is introduced to calculate the generalized torque generated by the aerial work robot;

[0011] The aerial robot operates based on the generalized control force and the generalized torque.

[0012] Further, optionally, the image moment-based visual servo controller is represented as:

[0013]

[0014] In the formula, This refers to the generalized control force generated by aerial robots. All are positive gain coefficients, with errors. This is expressed as the visual servo control rate. for The first derivative, The desired linear velocity of the aerial robot. For image moment error, m u For the quality of aerial work robots, Representing gravitational acceleration, vector e3 =

[001] T , This refers to the contact force generated when an aerial robot comes into contact with its environment. The rotation matrix of the aerial robot relative to the world coordinate system is represented by T, where T is the matrix transpose symbol.

[0015] Among them, the visual servo control rate and the image moment error e t Represented as:

[0016]

[0017]

[0018] In the formula, It is an interaction matrix. This represents a positive gain diagonal matrix. The image moment s represents the image moment when the aerial robot reaches the desired pose. z,d Actual image dimensions s x s y s z For image moments s t The variable values ​​in the x, y, and z directions.

[0019] Further optionally, the generalized torque generated by the aerial robot in the attitude controller is expressed as follows:

[0020]

[0021] In the formula, This represents the generalized torque generated by the aerial robot. All are positive gain coefficients. For attitude error, This refers to the angular velocity error. Angular velocity, This represents the desired angular velocity. for The first derivative; Represents the Jacobian matrix of an aerial work robot; operators This means that it transforms a three-dimensional vector into an antisymmetric matrix; This represents the rotation matrix of the aerial robot relative to the world coordinate system. For the desired rotation matrix, This represents the torque generated when the aerial robot comes into contact with the environment, where T is the matrix transpose symbol.

[0022] Further optionally, the visual impedance controller is constructed based on a one-dimensional variable stiffness impedance model. When the optical axis of the camera of the aerial operation robot is installed parallel to the operation mechanism, the visual impedance controller based on the one-dimensional variable stiffness impedance model is represented as follows:

[0023]

[0024] In the formula, e f =F c -F c,d F represents the error vector between the actual contact force and the expected contact force in three dimensions. c F represents the actual contact force generated when an aerial robot interacts with its environment in three dimensions. c,dThe desired contact force, m, represents the force generated when an aerial robot interacts with its environment in three dimensions. d b represents the coefficient of inertia. d k represents the damping coefficient. d (t) represents the stiffness rate vector, which exhibits characteristics that change with time;

[0025] The stiffness rate of change vector is expressed as:

[0026]

[0027] Where k0, k1, and k2 are all positive real gain coefficients greater than zero. Representing vectors The reverse, e f The first derivative, for This represents the image moment command value in the camera's optical axis direction after correction by the controller. This indicates that after modification by the controller, it is relative to... The new value.

[0028] Operational safety is a critical indicator for aerial robots. Existing research on aerial interaction tasks employs impedance control to ensure system robustness and avoid direct decomposition of motion and force control. However, these studies all use fixed impedance parameters, resulting in low compliance and safety during contact. Because constant impedance control parameters are fixed, its tracking performance is poor for tasks requiring constant contact force, which is dangerous for robot-environmental interaction tasks. Compared to constant impedance control, variable impedance control is an effective method that can improve compliance and safety during contact by dynamically adjusting impedance parameters. However, how to implement variable impedance control on aerial contact robots remains to be studied. This invention provides a visual impedance controller based on a one-dimensional variable stiffness impedance model.

[0029] Further, optionally, the stability constraint condition of the visual impedance controller based on the one-dimensional variable stiffness impedance model is: Lyapunov function V p First differential It is always less than or equal to zero, and V p There exists an upper bound, which increases with time, V p It will eventually stabilize gradually;

[0030] The proof is as follows:

[0031] Construct the Lyapunov function as shown below:

[0032]

[0033] In the formula, s z For image moments s t The variable value in the z-direction, k t It is the linear stiffness coefficient;

[0034] Taking the first derivative of the Lyapunov function, we get V p The first-order differential expression is as follows:

[0035]

[0036] In the formula, s z,t This indicates the position where the end-effector just makes contact with the target, but the contact force is zero;

[0037] Error vector e of contact force f Taking the first-order differential, we get e f Its first-order differential expression as follows:

[0038] e f =k t (s z -s z,t )-F c,d ,

[0039] The first-order differential expression of Lyapunov can be further rewritten as:

[0040]

[0041] Among them, b d Indicates the damping coefficient;

[0042] Since as time t approaches infinity This holds true throughout, and therefore we can conclude that as time t approaches infinity... With e f The asymptotic stability of the proposed control method is proven because it tends towards zero. With e f As the force approaches zero, the system will eventually maintain a constant contact force, i.e., F. c =F c,d .

[0043] Alternatively, the contact dynamics model of the aerial robot for contact operations is represented as follows:

[0044]

[0045]

[0046]

[0047]

[0048] in, This indicates the position of the origin of the aerial robot's coordinate system in the world coordinate system. express The first derivative corresponds to the velocity of the aerial robot in the world coordinate system.

[0049] Indicates the mass of the aerial work robot. Representing gravitational acceleration, vector e3 =

[001] T , and These represent the generalized control force and torque generated by the aerial robot, respectively. This represents the torque generated when an aerial robot comes into contact with its environment; This represents the attitude vector of the aerial robot in the world coordinate system, expressed in Euler angles. To indicate, θ represents the roll angle, θ represents the pitch angle, and ψ represents the yaw angle. This represents the rotation matrix of the aerial robot relative to the world coordinate system. express The Jacobian matrix representing the aerial robot. This refers to the contact force generated when an aerial robot comes into contact with its environment. Represents the angular velocity of the aerial robot in the world coordinate system; Operator This means that it transforms a three-dimensional vector into an antisymmetric matrix, for a vector Its expression is as follows:

[0050]

[0051] Alternatively, in the process of mapping the movement of the aerial robot in the virtual image plane to the Cartesian coordinate system, the transformation relationship between the virtual camera coordinate system and the Cartesian coordinate system, as well as the image moment relationship in the virtual image plane, are as follows:

[0052] Point P o In the camera coordinate system The three-dimensional position variables are represented as follows: Point P o In the virtual camera coordinate system The three-dimensional position variable is represented as The transformation relationship between the virtual camera coordinate system and the Cartesian coordinate system is as follows: Represents the virtual camera coordinate system Relative to camera coordinate system rotation matrix;

[0053] Then point P o The normalized form on the virtual camera plane is: x v ,y v Point P o The xy coordinates on the virtual camera plane;

[0054] Define the image moment of order i+j as:

[0055] Furthermore, the image moments s t =[s x ,s y ,s z ] T Represented as image moments:

[0056] in, For image moments m ij The coordinates of the center of the outline, This indicates when the aerial robot reaches the desired pose, m 00 The corresponding value.

[0057] Secondly, the present invention provides a control system based on the above control method, which includes at least: a visual servo controller, a visual impedance controller, an attitude controller, and a force / torque sensor;

[0058] Among them, the force / torque sensor is used to sense the contact force and torque generated when the aerial robot comes into contact with the environment;

[0059] The visual servo controller is built based on image moments;

[0060] The error between the tracked contact force and the desired contact force is input to the visual impedance controller, thereby obtaining the desired image moment s. z,d The visual servo controller is based on the desired image moments s z,d The generalized control force generated by the aerial work robot is generated by the actual contact force output; based on the generalized control force and the expected rotation matrix of the aerial work robot, an attitude controller is introduced to calculate the generalized torque generated by the aerial work robot; the obtained generalized control force and the generalized torque are used to control the operation of the aerial work robot.

[0061] Thirdly, the present invention provides an aerial work robot, which includes a memory and a processor. The memory stores a computer program, and the processor invokes the computer program to perform the following:

[0062] The error between the tracked contact force and the desired contact force is input into the visual impedance controller to obtain the desired image moment s. z,d ;

[0063] Using the visual servo controller and based on the desired image moments s z,d And the generalized control force generated by the aerial robot is obtained from the actual contact force;

[0064] Based on the generalized control force and the expected rotation matrix of the aerial robot, the attitude controller is introduced to calculate the generalized torque generated by the aerial robot.

[0065] The aerial work robot is controlled using the generalized control force and the generalized torque.

[0066] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that is invoked by a processor to implement:

[0067] The error between the tracked contact force and the desired contact force is input into the visual impedance controller to obtain the desired image moment s. z,d ;

[0068] Using the visual servo controller and based on the desired image moments s z,d And the generalized control force generated by the aerial robot is obtained from the actual contact force;

[0069] Based on the generalized control force and the expected rotation matrix of the aerial robot, the attitude controller is introduced to calculate the generalized torque generated by the aerial robot.

[0070] The aerial work robot is controlled using the generalized control force and the generalized torque.

[0071] Beneficial effects

[0072] Compared with existing methods, the advantages of the present invention are:

[0073] This invention provides a control technology for aerial work robots performing contact operations, ensuring safe and compliant operation when transitioning from non-contact to contact modes. Specifically, by constructing a visual impedance controller, impedance control is transformed into contact force tracking control; that is, by tracking the contact force, the stability of the aerial work robot during contact operations is effectively improved. Furthermore, the control method provided by this invention uses visual servo control, employing image moments as position information, independent of external position information. Since the control method controls the robot using forces and moments, without publishing position, contact operations can be performed in locations with weak or absent outdoor GPS signals (such as the bottom of bridge piers) or indoors (where there is no GPS location information).

[0074] A further preferred embodiment of the present invention designs and proposes a visual impedance controller based on a one-dimensional variable stiffness impedance model. Unlike existing control methods using fixed impedance parameters in aerial interaction missions, the present invention achieves one-dimensional variable stiffness impedance control. By dynamically adjusting the impedance parameters, it realizes constant contact force control of the end-effector on the object, resulting in better compliance and stability during the contact process.

[0075] The technical solution of this invention also designed a Lyapunov function, which further proves the asymptotic stability of the control method of this invention. Attached Figure Description

[0076] Figure 1 This is a control block diagram provided by the technical solution of the present invention;

[0077] Figure 2 This is a control flowchart provided by the technical solution of the present invention. Detailed Implementation

[0078] This invention provides a robust control method and system for aerial contact operation robots. It proposes image moments and uses visual servo control. Because the control method uses force and torque to control the robot, without publishing position and relying on external position information, this invention is also applicable to contact operations in areas with weak or absent outdoor GPS signals (such as the bottom of bridge piers) or indoors (where there is no GPS location information). Furthermore, this invention proposes a visual impedance controller based on a one-dimensional variable stiffness impedance model. By dynamically adjusting the impedance parameters, the compliance and safety during the contact process are improved. The invention will be further described below with reference to embodiments.

[0079] based on Figure 1 The control block diagram shown illustrates the robust control method for an aerial contact robot provided in this embodiment of the invention, where the visual impedance controller is based on the tracked contact force F.c The error between the desired contact force and the actual contact force is processed to obtain the desired image moment s. z,d The visual servo controller is based on the desired image moments. z,d and the actual contact force F c Output generalized control force T generated by the aerial work robot cmd Based on generalized control force T cmd And the expected rotation matrix of the aerial work robot The generalized torque τ generated by the aerial robot is calculated by introducing an attitude controller. cmd Finally, based on the generalized control force T cmd and the generalized torque τ cmd Control the aerial work robot to perform its tasks.

[0080] The reasoning process for this invention is as follows:

[0081] (I) Based on the Newton-Euler equations, construct the dynamic equations of the aerial robot system under contact mode:

[0082]

[0083]

[0084]

[0085]

[0086] Wherein, formulas (1) and (2) represent the translational dynamics of the aerial work robot system, and formulas (3) and (4) represent the rotational dynamics of the aerial work robot system; the following is a description of the variables in formulas (1)-(4):

[0087] This indicates the position of the origin of the aerial robot's coordinate system in the world coordinate system. express The first derivative corresponds to the velocity of the aerial robot in the world coordinate system. Indicates the mass of the aerial work robot. Representing gravitational acceleration, vector e3 =

[001] T , and These represent the generalized control force and torque generated by the aerial robot, respectively. This refers to the torque generated when an aerial robot comes into contact with its environment. This represents the attitude vector of the aerial robot in the world coordinate system, expressed in Euler angles. To indicate, θ represents the roll angle, θ represents the pitch angle, and ψ represents the yaw angle. This represents the rotation matrix of the aerial robot relative to the world coordinate system. express The first derivative; The Jacobian matrix representing the aerial robot. This refers to the contact force generated when an aerial robot comes into contact with its environment. This represents the angular velocity of the aerial robot in the world coordinate system, where T is the matrix transpose symbol. Operators This means that it transforms a three-dimensional vector into an antisymmetric matrix, for a vector Its expression is as follows:

[0088]

[0089] (ii) Define image moments and introduce the transformation relationship between the virtual camera coordinate system and the Cartesian coordinate system; wherein, the movement of image moments in the virtual image plane is mapped to the Cartesian coordinate system through the defined transformation relationship.

[0090] This embodiment uses a pinhole camera model to represent each observed feature point, and defines an image matrix representing the position and state information of the aerial robot. s x s y s z These are the variable values ​​of the image moments in the x, y, and z directions.

[0091] Consider a point P o On the target plane to be detected, point P o In the camera coordinate system The three-dimensional position variables are represented as follows:

[0092] Normalizing its representation on the camera plane, we get:

[0093]

[0094] Introduce a virtual camera model, whose coordinate system is the virtual camera coordinate system. The corresponding camera plane is a virtual camera plane. It is assumed that the virtual camera plane remains parallel to the plane containing the target point, meaning the roll and pitch angles are always zero. Point P o In the virtual camera coordinate system The coordinates below are represented as P. v Camera coordinate system With virtual camera coordinate system The origins of points P and P coincide, therefore point P... o coordinates Pv With coordinate P c The transformation relationship is as follows:

[0095]

[0096] in, Represents the virtual camera coordinate system Relative to camera coordinate system The rotation matrix is ​​then used to obtain point P. o Normalized form on the virtual camera plane:

[0097]

[0098] The image moments are applied to the detection plane, and the i+j order image moments are defined as follows:

[0099]

[0100] Furthermore, the image moment m defined in formula (9) is calculated. ij The center of the contour, the coordinates of the center of the contour It is expressed as follows:

[0101]

[0102] Where, m 00 Let the area of ​​the image moment be denoted as , and thus the image moment can be expressed in the following form:

[0103] s t =[s x ,s y ,s z ] T (11)

[0104]

[0105]

[0106] in, This indicates when the aerial robot reaches the desired pose, m 00 The corresponding value.

[0107] Based on the selected image features, s z It can be represented as:

[0108]

[0109] make Represents the virtual camera coordinate system The desired depth to the target plane is obtained from the characteristics of image moments, which yields the image moment s. t Image dynamics:

[0110]

[0111] From formula (15), it can be seen that the designed image moment s t Independent of the required depth information Z v This image dynamics is very simple and decouples the rotation and translation of the aerial robot. Compared with discrete point feature sets, the circle-based moment features defined in a non-rotating virtual image plane have angle invariance and are more robust to disturbances from textureless or unreliable textures in real-world scenarios.

[0112] (III) Defining the Visual Servo Control Rate Used for calculations in Formula 26.

[0113]

[0114]

[0115] in, This represents the image moments when the aerial robot reaches the desired pose. The error is the image moment. It is an interaction matrix used to correlate changes in image features at the camera with camera speed. This represents a positive gain diagonal matrix.

[0116] (iv) Consider the expected value of the rotation matrix representing the attitude described in formulas (3) and (4). Expected value of angular velocity Define the expected value of the rotation matrix The expression is as follows:

[0117]

[0118] u1=u2×u3 (19)

[0119]

[0120]

[0121] β=[cos(ψ d ),sin(ψ d ),0] T (twenty two)

[0122] Where, ψ dThis represents the yaw angle when the aerial robot reaches the desired pose. It should be understood that this embodiment preferably calculates the expected value of the rotation matrix according to the above formula; however, other feasible embodiments that can calculate the expected value of the rotation matrix also fall within the protection scope of this invention.

[0123] To make the rotation matrix equal Define attitude error With angular velocity error Used for subsequent calculations in Formula 25. This embodiment uses the following formula to calculate the attitude error and angular velocity error:

[0124]

[0125]

[0126] in, The ω represents the desired angular velocity vector. The operator vex(·) is the inverse operation of the operator sk(·), which means transforming the antisymmetric matrix sk(a) into vector a, where ω is the angular velocity.

[0127] Combining the above formulas (16)-(24), the generalized control force and torque generated by the aerial operation robot are involved. and The calculation is as follows:

[0128]

[0129]

[0130] in, The desired linear velocity of the aerial work robot can be obtained from the aerial work robot itself. This represents the positive gain coefficient.

[0131] (V) Design of a visual impedance controller based on a one-dimensional variable stiffness impedance model:

[0132] In existing research on aerial interaction tasks, impedance control methods have been employed to ensure system robustness and avoid direct decomposition of motion and force control. However, these studies all use control methods with fixed impedance parameters, resulting in low compliance and safety during contact. Compared to constant impedance control, variable impedance control is an effective method that can improve compliance and safety during contact by dynamically adjusting impedance parameters. Furthermore, since constant impedance control parameters are fixed, its tracking performance is poor for tasks requiring constant contact force, which is dangerous for robot environmental interaction tasks. Therefore, this invention designs a variable stiffness vision-impedance control method.

[0133] When an aerial robot comes into contact with an object, it exhibits a specific mass, spring, and damping model. The impedance model of the mechanism can be equivalent to:

[0134]

[0135] Where, p cd =p c -p d p d p represents the desired position trajectory of the end effector generated by the impedance controller. c It is the command position trajectory of the end effector after being corrected by the impedance controller. They represent p respectively cd The first and second derivatives of . f =F c -F c,d F represents the error vector between the actual contact force and the expected contact force in three dimensions. c F represents the actual contact force vector in three dimensions. c,d Let m represent the desired contact force vector in three dimensions. d b represents the coefficient of inertia. d k represents the damping coefficient. d (t) represents the stiffness rate vector, which exhibits characteristics that change with time.

[0136] At the same time, define the following variables and Let L represent the relative position of the target with respect to the end effector and the camera, and thus obtain the variable L. e rate of change ΔL e With variable L c rate of change ΔL c The relationship is as follows:

[0137]

[0138] in, Indicates from the camera coordinate system To the coordinate system where the end effector is located The rotation matrix.

[0139] In the designed robot, combining formula (14), we obtain:

[0140]

[0141] Regardless of whether it's a fully driven or underdriven robot, as long as the camera's optical axis is mounted parallel to the working mechanism, this means that in the direction of the camera's optical axis, Δs z =ΔL e ;

[0142] Substituting equation (29) into (27), the impedance equation is rewritten as:

[0143]

[0144] in, s z,c This represents the image moment command value (s) in the camera's optical axis direction after correction by the controller. z,d Represents image moments s z The expected value.

[0145] Stiffness change rate vector k d The expression for (t) is as follows:

[0146]

[0147] Where k0, k1, and k2 are all positive real gain coefficients greater than zero. Representing vectors The reverse. The error vector e represents the difference between the actual contact force and the expected contact force. f The first derivative. This means that when the desired contact force jumps from zero to a fixed value, the contact force will gradually track the desired contact force over a period of time and eventually reach the desired value, rather than being a fixed contact force value. Meanwhile, this invention considers that the environmental model has a linear stiffness coefficient k. t Therefore, the contact force F between the end-effector and the environment c It can be represented as F c =k t (s z -s z,t ), s z,t This indicates the position where the end effector just makes contact with the target, but the contact force is zero. For the contact force F... c Taking the first derivative yields Therefore, the impedance model in formula (27) can be rewritten as follows:

[0148]

[0149] Thus, based on Figure 1 The control logic shown, utilizing the various models and formulas constructed through the above reasoning, can achieve stability control of the aerial robot. To analyze the contact force tracking-based control method proposed in this invention, consider designing the following Lyapunov function:

[0150]

[0151] The objective of this invention is to prove the Lyapunov function V. p First differential It is always less than or equal to zero, and V p If there is an upper bound, then as time gradually increases, V p Eventually, it will asymptotically stabilize. Based on formula (33), further performing a first-order differential yields V. p The first-order differential expression is as follows:

[0152]

[0153] Based on the expression for contact force above, the error e of the contact force is... f Taking the first-order differential, we get e f Its first-order differential expression as follows:

[0154]

[0155] Substituting formula (6) into formula (5), Lyapunov's first-order differential expression is rewritten as:

[0156]

[0157] Since as time t approaches infinity This holds true throughout, and therefore we can conclude that as time t approaches infinity... With e f The asymptotic stability of the proposed control method is proven because it tends towards zero. With e f As the force approaches zero, the system will eventually maintain a constant contact force, i.e., F. c =F c,d .

[0158] In summary, such as Figure 2 As shown, the design concept of the control method of this invention is as follows:

[0159] Step 1: Construct the contact dynamics model of the aerial robot for contact operations (Equations 1-4); and propose a visual servo controller based on image moments, wherein image moments representing the position and state information of the aerial robot are defined in the virtual image plane, and the movement of the aerial robot in the virtual image plane is mapped to the Cartesian coordinate system.

[0160] Step 2: Construct a visual impedance controller based on a one-dimensional variable stiffness impedance model to transform impedance control into control that tracks contact force;

[0161] Step 3: Design the Lyapunov function to verify the stability of the control method;

[0162] It should be understood that this verifies the stability of the control method proposed in this invention, and then control is carried out according to the control logic in step 4.

[0163] Step 4: The visual impedance controller outputs an error vector based on the difference between the tracked contact force and the desired contact force. This leads to the desired image moment s z,d The visual servo controller is based on the desired image moments. z,d The system outputs the generalized control force generated by the aerial work robot based on the actual contact force; based on the generalized control force and the desired rotation matrix of the aerial work robot, an attitude controller is introduced to calculate the generalized torque generated by the aerial work robot; finally, the aerial work robot is controlled to operate based on the generalized control force and the generalized torque.

[0164] This invention also provides a control system based on the above control method, which includes at least: a visual servo controller, a visual impedance controller, an attitude controller, and a force / torque sensor.

[0165] Among them, force / torque sensors are used to sense the contact force and torque generated when the aerial robot comes into contact with the environment; the visual servo controller is built based on image moments. For example... Figure 1 As shown, the visual impedance controller obtains the desired image moment s based on the input of the error between the tracked contact force and the desired contact force. z,d The visual servo controller is based on the desired image moments s z,d The generalized control force generated by the aerial work robot is generated by the actual contact force output; based on the generalized control force and the expected rotation matrix of the aerial work robot, an attitude controller is introduced to calculate the generalized torque generated by the aerial work robot; the obtained generalized control force and the generalized torque are used to control the operation of the aerial work robot.

[0166] For details on the implementation process of each module, please refer to the methods described above; they will not be repeated here. It should be understood that the above division of functional modules is merely a logical functional division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the integrated units described above can be implemented in hardware or as software functional units.

[0167] This invention also provides an aerial work robot, which includes a memory and a processor. The memory stores a computer program, and the processor calls the computer program to perform the following:

[0168] By utilizing a visual impedance controller and based on the input of the error between the tracked contact force and the desired contact force, the desired image moment s is obtained. z,d ;Utilizing the visual servo controller and based on the desired image moments s z,d The generalized control force generated by the aerial work robot is obtained from the actual contact force; based on the generalized control force and the expected rotation matrix of the aerial work robot, an attitude controller is introduced to calculate the generalized torque generated by the aerial work robot; the aerial work robot is controlled to operate using the generalized control force and the generalized torque.

[0169] The memory may include high-speed RAM and may also include a non-volatile defibrillator, such as at least one disk storage device. If the memory and processor are implemented independently, the memory, processor, and communication interface can be interconnected via a bus to communicate with each other. The bus can be an industry-standard architecture bus, an external device interconnect bus, or an extended industry-standard architecture bus, etc. The bus can be categorized as an address bus, data bus, control bus, etc.

[0170] Optionally, in a specific implementation, if the memory and processor are integrated on a single chip, the memory and processor can communicate with each other through an internal interface.

[0171] It should be understood that, in the embodiments of the present invention, the processor may be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor. The memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory may also include non-volatile random access memory. For example, the memory may also store device type information.

[0172] This invention also provides a computer-readable storage medium storing a computer program that is invoked by a processor to: utilize a visual impedance controller and, based on an input of the error between the tracked contact force and the desired contact force, obtain a desired image moment s. z,d ;Utilizing the visual servo controller and based on the desired image moments s z,d The generalized control force generated by the aerial work robot is obtained from the actual contact force; based on the generalized control force and the expected rotation matrix of the aerial work robot, an attitude controller is introduced to calculate the generalized torque generated by the aerial work robot; the aerial work robot is controlled to operate using the generalized control force and the generalized torque.

[0173] Please refer to the explanation of the method above for the specific implementation process of each step.

[0174] The readable storage medium is a computer-readable storage medium, which can be an internal storage unit of the controller described in any of the foregoing embodiments, such as the controller's hard drive or memory. The readable storage medium can also be an external storage device of the controller, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the controller. Further, the readable storage medium can include both the controller's internal storage unit and external storage devices. The readable storage medium is used to store the computer program and other programs and data required by the controller. The readable storage medium can also be used to temporarily store data that has been output or will be output.

[0175] Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0176] It should be emphasized that the examples described in this invention are illustrative rather than limiting. Therefore, this invention is not limited to the examples described in the specific embodiments. Any other embodiments derived by those skilled in the art based on the technical solutions of this invention, without departing from the spirit and scope of this invention, whether modifications or substitutions, are also within the protection scope of this invention.

Claims

1. A robust control method for an aerial contact operation robot, characterized in that: Includes the following steps: Construct a contact dynamics model for aerial robots performing contact operations; A visual servo controller based on image moments is constructed, wherein image moments representing the position and state information of the aerial robot are defined in the virtual image plane, and the movement of the aerial robot in the virtual image plane is mapped to the Cartesian coordinate system; Construct a visual impedance controller to transform impedance control into control that tracks contact force; The error between the tracked contact force and the desired contact force is input into the visual impedance controller to obtain the desired image moment. The visual servo controller is based on the desired image moments. And the generalized control force generated by the aerial work robot is generated by the actual contact force output; based on the generalized control force and the expected rotation matrix of the aerial work robot, the attitude controller is introduced to calculate the generalized torque generated by the aerial work robot; The aerial work robot is controlled to perform its operations based on the generalized control force and the generalized torque. The representation of a visual servo controller based on image moments is as follows: ； In the formula, This refers to the generalized control force generated by aerial robots. All are positive gain coefficients, with errors. , This is expressed as the visual servo control rate. for The first derivative, The desired linear velocity of the aerial robot. For image moment error, For the quality of aerial work robots, Represents gravitational acceleration, a vector. , This refers to the contact force generated when an aerial robot comes into contact with its environment. The rotation matrix of the aerial robot relative to the world coordinate system is represented by T, where T is the matrix transpose symbol. Among them, visual servo control rate and image moment error Represented as: ； ； In the formula, It is an interaction matrix. This represents a positive gain diagonal matrix. This represents the image moments when the aerial robot reaches the desired pose, i.e., the desired image moments. Actual image dimensions , For image moments The variable values ​​in the x, y, and z directions.

2. The method according to claim 1, characterized in that: The generalized torque generated by the aerial robot in the attitude controller is expressed as follows: ； In the formula, This represents the generalized torque generated by the aerial robot. All are positive gain coefficients. For attitude error, This refers to the angular velocity error. Angular velocity, This represents the desired angular velocity. for The first derivative; Represents the Jacobian matrix of an aerial work robot; operators This means that it transforms a three-dimensional vector into an antisymmetric matrix; This represents the rotation matrix of the aerial robot relative to the world coordinate system. For the desired rotation matrix, This refers to the torque generated when an aerial robot comes into contact with its environment.

3. The method according to claim 1, characterized in that: The visual impedance controller is constructed based on a one-dimensional variable stiffness impedance model. When the optical axis of the camera of the aerial operation robot is installed parallel to the operation mechanism, the visual impedance controller based on the one-dimensional variable stiffness impedance model is represented as follows: ； In the formula, This represents the error vector between the actual contact force and the expected contact force in three dimensions. This represents the actual contact force generated when an aerial robot interacts with its environment in three dimensions. This represents the expected contact force generated when an aerial robot interacts with its environment in three dimensions. Represents the coefficient of inertia. Indicates the damping coefficient. This represents the stiffness rate of change vector, which exhibits characteristics that change with time. The stiffness rate of change vector is expressed as: ； in, , , All are positive real gain coefficients greater than zero. Representing vectors The reverse, express The first derivative, for , This represents the image moment command value in the camera's optical axis direction after correction by the controller.

4. The method according to claim 3, characterized in that: The stability constraint of the visual impedance controller based on the one-dimensional variable stiffness impedance model is: Lyapunov function. First differential Always less than or equal to zero, and There exists an upper bound, which increases gradually over time. It will eventually stabilize gradually; The proof is as follows: Construct the Lyapunov function as shown below: ； In the formula, For image moments The variable value in the z-direction, It is the linear stiffness coefficient; Taking the first derivative of the Lyapunov function, we get The first-order differential expression is as follows: ； Error vector of contact force Taking the first-order differential, we get Its first-order differential expression as follows: ； In the formula, This indicates the position where the end-effector just makes contact with the target, but the contact force is zero; Lyapunov's first-order differential expression can be further rewritten as: ； in, Indicates the damping coefficient; Since as time t approaches infinity This holds true throughout, and therefore we can conclude that as time t approaches infinity... and The asymptotic stability of the proposed control method is proven because it tends towards zero. and As the force approaches zero, the system will eventually maintain a constant contact force, i.e. .

5. The method according to claim 1, characterized in that: The contact dynamics model of an aerial robot performing contact-based operations is represented as follows: ； ； ； ； in, This indicates the position of the origin of the aerial robot's coordinate system in the world coordinate system. express The first derivative corresponds to the velocity of the aerial robot in the world coordinate system. ; express The first derivative; Indicates the mass of the aerial work robot. Represents gravitational acceleration, a vector. , and These represent the generalized control force and torque generated by the aerial robot, respectively. This represents the torque generated when an aerial robot comes into contact with its environment; This represents the attitude vector of the aerial robot in the world coordinate system, expressed in Euler angles. To indicate, Indicates the roll angle. Indicates pitch angle, Indicates the yaw angle; This represents the rotation matrix of the aerial robot relative to the world coordinate system. express The first derivative, The Jacobian matrix representing the aerial robot. This refers to the contact force generated when an aerial robot comes into contact with its environment. Represents the angular velocity of the aerial robot in the world coordinate system; Operator This means that it transforms a three-dimensional vector into an antisymmetric matrix, for a vector Its expression is as follows: 。 6. The method according to claim 1, characterized in that: In the process of mapping the movement of the aerial robot in the virtual image plane to the Cartesian coordinate system, the transformation relationship between the virtual camera coordinate system and the Cartesian coordinate system, as well as the image moment relationship in the virtual image plane, are as follows: point In the camera coordinate system The three-dimensional position variables are represented as follows: ;point In the virtual camera coordinate system The three-dimensional position variable is represented as The transformation relationship between the virtual camera coordinate system and the Cartesian coordinate system is as follows: ; Represents the virtual camera coordinate system Relative to camera coordinate system rotation matrix; Further point The normalized form on the virtual camera plane is: , For point The xy coordinates on the virtual camera plane; definition The first-order image moments are: ; Furthermore, image moments Represented as image moments: , , ; in, For image moments The coordinates of the center of the outline This indicates when the aerial robot reaches the desired pose. The corresponding value, This is an intermediate variable that is set.

7. A system based on the method of any one of claims 1-6, characterized in that: It includes at least: a visual servo controller, a visual impedance controller, an attitude controller, and a force / torque sensor; Among them, the force / torque sensor is used to sense the contact force and torque generated when the aerial robot comes into contact with the environment; The visual servo controller is built based on image moments; The error between the tracked contact force and the desired contact force is input to the visual impedance controller, thereby obtaining the desired image moment. The visual servo controller is based on the desired image moments. The generalized control force generated by the aerial work robot is generated by the actual contact force output; based on the generalized control force and the expected rotation matrix of the aerial work robot, an attitude controller is introduced to calculate the generalized torque generated by the aerial work robot; the obtained generalized control force and the generalized torque are used to control the operation of the aerial work robot.

8. An aerial work robot, characterized in that: The system includes a memory and a processor. The memory stores a computer program, and the processor calls the computer program to implement the following: The error between the tracked contact force and the desired contact force is input into the visual impedance controller, thereby obtaining the desired image moment. ; Using a visual servo controller and based on the desired image moments And the generalized control force generated by the aerial robot is obtained from the actual contact force; Based on the generalized control force and the expected rotation matrix of the aerial robot, the attitude controller is introduced to calculate the generalized torque generated by the aerial robot. The aerial work robot is controlled to perform its tasks using the generalized control force and the generalized torque. The representation of a visual servo controller based on image moments is as follows: ； In the formula, This refers to the generalized control force generated by aerial robots. All are positive gain coefficients, with errors. , This is expressed as the visual servo control rate. for The first derivative, The desired linear velocity of the aerial robot. For image moment error, For the quality of aerial work robots, Represents gravitational acceleration, a vector. , This refers to the contact force generated when an aerial robot comes into contact with its environment. The rotation matrix of the aerial robot relative to the world coordinate system is represented by T, where T is the matrix transpose symbol. Among them, visual servo control rate and image moment error Represented as: ； ； In the formula, It is an interaction matrix. This represents a positive gain diagonal matrix. This represents the image moments when the aerial robot reaches the desired pose, i.e., the desired image moments. Actual image dimensions , For image moments The variable values ​​in the x, y, and z directions.

9. A computer-readable storage medium, characterized in that: The computer program is stored and is invoked by the processor to implement: The error between the tracked contact force and the desired contact force is input into the visual impedance controller, thereby obtaining the desired image moment. ; Using a visual servo controller and based on the desired image moments And the generalized control force generated by the aerial robot is obtained from the actual contact force; Based on the generalized control force and the expected rotation matrix of the aerial robot, the attitude controller is introduced to calculate the generalized torque generated by the aerial robot. The aerial work robot is controlled to perform its tasks using the generalized control force and the generalized torque. The representation of a visual servo controller based on image moments is as follows: ； In the formula, This refers to the generalized control force generated by aerial robots. All are positive gain coefficients, with errors. , This is expressed as the visual servo control rate. for The first derivative, The desired linear velocity of the aerial robot. For image moment error, For the quality of aerial work robots, Represents gravitational acceleration, a vector. , This refers to the contact force generated when an aerial robot comes into contact with its environment. The rotation matrix of the aerial robot relative to the world coordinate system is represented by T, where T is the matrix transpose symbol. Among them, visual servo control rate and image moment error Represented as: ； ； In the formula, It is an interaction matrix. This represents a positive gain diagonal matrix. This represents the image moments when the aerial robot reaches the desired pose, i.e., the desired image moments. Actual image dimensions , For image moments The variable values ​​in the x, y, and z directions.

Images