A method for anti-interference pointing control at the end of a flying robot arm

Abstract

The present invention provides an anti-interference pointing control method for the end of a flying manipulator arm to solve the kinematic interference effect of the floating base of a UAV on the flying manipulator arm when performing tasks. First, the kinematic equation of the end position of the flying manipulator arm actuator in the inertial coordinate system is constructed according to the principle of homogeneous transformation. Then, the kinematic interference caused by the floating base of the UAV on the end of the flying manipulator arm actuator is quantitatively analyzed. Next, a neural network with the state of the UAV and the manipulator arm as input is designed to predict the motion information of the UAV base, and the reference trajectory of the manipulator arm in the inertial system is converted to the trajectory in the coordinate system of the manipulator arm base through homogeneous transformation. Finally, a model prediction controller is designed to plan the expected acceleration of the end of the manipulator arm actuator and convert it to the angle of each servo of the manipulator arm. The present invention can significantly improve the actuator end accuracy of the flying manipulator arm when the base of the UAV is floating, and can be used for high-precision special operation tasks such as high-altitude sculpture detection.

Description

Technical Field

[0001] The present invention belongs to the field of flying robot control, and in particular relates to an anti-interference pointing control method for the end of a flying robot arm, which is suitable for a flying robot arm control system that needs to perform high-precision active operation tasks. Background Art

[0002] In recent years, small drones with multi-rotor as driving mechanism have been widely used in disaster warning, geological survey, emergency rescue, defense, electricity, aerial photography and other fields, but most of them are limited to the field of non-contact information perception. At present, with the continuous development of mechanical design, high-precision sensors and advanced control theory, a flying robot composed of a multi-rotor drone and a robot arm that can physically interact with the environment has become a very hot field, such as Figure 2 As shown in the figure. The flying manipulator has the ability to interact with the environment, and its application has been extended to various active operation tasks. When physically interacting with the environment, the application field of the flying manipulator mainly depends on the accuracy of its actuator end. Among various active operation tasks, there are some special scenarios that have very strict requirements on the accuracy of the actuator end of the flying manipulator, such as high-voltage tower maintenance; bridge and high-altitude sculpture maintenance; high-frequency cargo pick-up and placement in warehouses, etc. Therefore, ensuring the high accuracy of the actuator end of the flying manipulator is a key issue that needs to be solved urgently. However, compared with the traditional fixed-base manipulator, the base of the flying manipulator is a drone, and the manipulator will be affected by the kinematic interference caused by the floating of the drone base. At the same time, due to the multi-link characteristics of the manipulator, the floating interference of the drone base is transmitted to the actuator end of the flying manipulator, which will have an error amplification effect, seriously deteriorating the actuator end accuracy of the flying manipulator. In addition, the dynamic coupling interference generated by the manipulator during the operation task will cause additional disturbances to the drone base, increase the floating of the drone base, and thus affect the accuracy of the overall system. Compared with ordinary flying robots, the flying robotic arm consists of two parts: a rotorcraft drone and a multi-joint robotic arm. Therefore, when performing operating tasks, it is necessary to avoid collision between the robotic arm and the drone body during movement, which puts more requirements on the trajectory movement of the flying robotic arm.

[0003] Therefore, in order to ensure the high precision of the actuator end of the flying robotic arm, improve the success rate of active operation tasks, and enable the flying robotic arm to complete precise interactive tasks such as high-voltage tower maintenance, the anti-interference pointing control algorithm of the actuator end of the flying robotic arm must solve the above-mentioned UAV floating base and robotic arm actuator end trajectory planning problems during the design process.

[0004] Chinese invention patent CN201610389286.6 designed a control algorithm that combines differential geometry with LQR to deal with the interference of space manipulators, but there are two problems: (1) the established dynamic model does not take gravity into account, and the controller is only applicable in a weightless environment; (2) the manipulator has few degrees of freedom, which greatly limits its operational capabilities; Chinese invention patent CN201810094313.6 proposed a rotor flight manipulator system and algorithm based on dynamic center of gravity compensation, but there are also two problems: (1) the effect of rotor UAV floating on the mechanical Kinematic interference caused by the end of the arm actuator; (2) The dynamic coupling interference caused by the movement of the mechanical arm to the base of the UAV is not considered; Chinese invention patent CN201810010602.3 proposes an arm-carrying UAV equipped with a parallel mechanical arm, and a parallel mechanical arm is provided under the rotorcraft, but the device proposed has a small working space and can only work under the rotorcraft; the rotor flight grasping mechanical arm proposed by Chinese invention patent CN202123017099.0 also has a small working space and cannot complete the active operation task. Chinese invention patent CN202110994016.9 proposes a control algorithm that combines an anti-saturation controller and an external force estimator, but does not process the floating interference of the UAV base, resulting in a large control error at the end of the flight mechanical arm, which cannot meet the requirements of high-precision operation tasks; Chinese invention patent CN202010801707.8 designs a disturbance observer to estimate the amplitude of the base floating interference, but requires that the base floating interference must be periodic interference. Chinese invention patent CN202110148055.7 proposes a quad-rotor UAV system with arms and a control method based on pigeon flock optimization control, but there are two similar problems: (1) the robotic arm has only two degrees of freedom, which greatly limits the types of operation tasks; (2) the kinematic interference caused by the floating of the rotor UAV to the end of the robotic arm actuator is not considered;

[0005] Therefore, none of the above methods considers realizing anti-interference pointing control of the end of the flight manipulator in the presence of base floating interference to complete high-precision active operation tasks. Summary of the invention

[0006] In order to overcome the defects of the existing methods, for a flying robot system composed of a multi-rotor UAV and a multi-degree-of-freedom robotic arm, the present invention provides an anti-interference pointing control method for the end of a flying robotic arm, which can ensure the high precision of the end of the flying robotic arm actuator under the floating interference of the base, so as to complete various active operation tasks.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is:

[0008] A method for anti-interference pointing control of a flight robot arm end, comprising the following steps:

[0009] The first step is to construct the kinematic equation of the end position of the flight robot arm actuator in the inertial coordinate system based on the principle of homogeneous transformation; and to quantitatively analyze the kinematic interference caused by the floating base of the UAV to the end of the flight robot arm actuator;

[0010] In the second step, the reference trajectory of the manipulator in the inertial system is redesigned into the motion trajectory in the UAV coordinate system through homogeneous transformation, and a neural network with the UAV and manipulator states as input is designed to predict the motion state of the UAV base;

[0011] The third step is to design a model predictive controller to plan the expected acceleration of the end of the robotic arm actuator and convert it into the angle of each servo of the robotic arm; complete the anti-interference pointing control of the end of the flying robotic arm actuator under the interference of the UAV floating base and achieve high-precision grasping tasks.

[0012] Furthermore, the first step specifically includes the following steps:

[0013] The kinematic model of the end position of the flight robot arm actuator is established according to the principle of homogeneous transformation, which is expressed as follows:

[0014]

[0015] Where P e represents the position of the end of the flight robot arm actuator in the inertial system, R e is the rotation matrix of the end of the flight robot arm actuator in the inertial system; P b represents the position of the center of mass of the drone base in the inertial system, R b is the rotation matrix of the center of mass of the drone in the inertial system; Indicates the position of the end of the robotic arm actuator in the drone coordinate system, Represents the rotation matrix of the end of the manipulator actuator in the drone coordinate system; R b The calculation formula is as follows:

[0016]

[0017] In the formula, They respectively represent the roll angle, pitch angle and attitude angle of the flying manipulator in the inertial system;

[0018] Based on the kinematic equation established above, a qualitative analysis is performed to quantify the kinematic interference caused by the floating of the UAV base to the end of the robotic arm actuator.

[0019] Furthermore, the second step includes the following specific steps:

[0020] The motion trajectory of the robotic arm is redesigned through the principle of homogeneous transformation, and a neural network is designed to predict the motion state of the drone base:

[0021] The motion state of the end of the flight robot arm actuator is expressed under SE(3) (SE(3) represents the transformation motion of the rigid body, including rotation and translation) in the following form:

[0022]

[0023] in, Represents the homogeneous transformation matrix of the end of the flight robot arm actuator in the inertial coordinate system, is the homogeneous transformation matrix of the center of mass of the drone base in the inertial coordinate system, is the homogeneous transformation matrix of the end of the flight manipulator actuator in the drone coordinate system, I represents the 1×1 unit matrix; the next motion state of the drone base The trajectory of the manipulator in the inertial system is converted to its own base coordinate system, i.e., the drone coordinate system, by using the principle of homogeneous transformation to offset the floating interference of the drone base. Here, t is the current control moment and i is the step length of the forward prediction.

[0024] The homogeneous transformation takes the following form:

[0025]

[0026] At this time, the trajectory tracking problem of the flying robot arm in the inertial system is converted to the trajectory tracking problem of the robot arm in the drone coordinate system. By making full use of the high precision of the robot arm itself, the high precision of the actuator end of the flying robot arm is guaranteed.

[0027] The motion state of the drone base at the next moment is expressed by the following nonlinear function:

[0028]

[0029] Where x(t) represents the current motion state of the drone base, u(t) represents the current control input of the drone base, Represents the collective interference to the drone base, including the model uncertainty interference related to its own state x(t) and the robotic arm angle q(t), angular velocity Angular acceleration The relevant dynamic coupling interferes with the two parts;

[0030] A neural network is established that takes the state of the drone and the robotic arm as input to predict the next movement state of the drone. The form of the neural network is as follows:

[0031] η(t+1)=c(-aη(t)+g(W inκ(t)+Wη(t)))

[0032] f out =Π(W in κ(t)+W out η(t+1))

[0033] Among them, η(t) represents the state value of the intermediate neuron of the neural network; κ(t) represents the input of the neural network, including the motion state of the drone base and the robotic arm; W,W in and W out They represent the connection weights between the intermediate neurons, input neurons and output neurons of the neural network respectively; c and a are the learning parameters in the neural network; g and Π represent the function mapping relationship in the neural network respectively.

[0034] Furthermore, the specific steps of the third step are as follows:

[0035] Combining the UAV motion state output by the neural network with the kinematic homogeneous transformation relationship, and considering the collision avoidance between the UAV and the robotic arm and the physical limitations of the robotic arm actuator, the motion trajectory of the robotic arm actuator end is mapped to the trajectory of each joint through the Jacobi matrix to solve the anti-interference pointing problem of the actuator end of the flying robotic arm. The model predictive controller adjusts the control amount of the robotic arm under various constraints to achieve the optimal actuator control input while achieving the control target.

[0036] The kinematic state space modeling of the end of the flight robot arm actuator is as follows:

[0037]

[0038] in, represents the position (3D) and velocity (3D) of the end of the robot arm actuator at time j; h j+1 represents the position (3D) and velocity (3D) of the end of the robot arm actuator at time j+1; u j represents the acceleration of the end of the robot arm actuator at time j; y j+1 Represents the output of the state space at time j+1;

[0039] The values ​​of the system matrix A, input matrix B, and output matrix C are as follows:

[0040]

[0041] In the formula, σt represents the sampling time interval, I3 represents the 3×3 unit matrix, and 03 represents the 3×3 zero matrix. The cost function of the model prediction controller is further designed, and various constraints such as collision avoidance between the UAV frame and the robotic arm and acceleration constraints at the end of the robotic arm actuator are considered. The form is as follows:

[0042]

[0043] Where, J represents the total cost function, u represents the input matrix of model predictive control; J1, J2, and J3 are the tracking error cost function, the control input cost function, and the cost function of avoiding collision at the end of the manipulator actuator, respectively; W1, W2, and W3 are the weight matrices of the tracking error cost function, the control input cost function, and the cost function of avoiding collision at the end of the manipulator actuator, respectively, and are adjustable parameters;

[0044] min means finding the minimum value of function J, where the optimization input u is the independent variable of function J.

[0045] The tracking error cost function J1 at the end of the flight manipulator actuator is expressed as:

[0046]

[0047] in, Represents the homogeneous transformation matrix The position part; y(t+i) represents the actual position of the end of the robot arm actuator; N is the designed control domain length; t is the current control moment; i is the step size of the forward prediction; ||·|| represents the 2-norm of the vector.

[0048]

[0049] The system input is smoothed by adding the control input cost function J2 to reduce the drastic change of the acceleration at the end of the robot arm actuator;

[0050] The cost function J3 of the flight manipulator actuator end collision avoidance is expressed as:

[0051]

[0052] in, and They represent the safety boundaries of the end motion of the robot arm actuator, Indicates selection and The larger value of It means to find the maximum value of each element in the vector.

[0053] Furthermore, the constraints satisfied by the model predictive control are as follows:

[0054] u min ≤u(t+i)≤u max

[0055] In the formula, u min and u maxIndicates the limit of the acceleration of the end of the robot arm actuator, so as to ensure the stability of the flight robot system.

[0056] The advantages of the present invention compared with the prior art are:

[0057] The present invention relates to a predictive planning control method for a flying manipulator for base floating interference, which is mainly aimed at a flying manipulator system composed of a multi-rotor UAV and a multi-degree-of-freedom manipulator. Compared with the traditional fixed base manipulator, the flying manipulator has the advantages of a wider range of action and stronger air interaction ability. However, the kinematic interference caused by the floating of the UAV base will seriously deteriorate the end precision of the flying manipulator actuator, resulting in the failure of the active operation task. This method aims at the floating interference of the UAV base. First, the motion trajectory of the flying manipulator in the inertial system is converted to the motion trajectory in the coordinate system of the UAV base by using the homogeneous transformation principle, and the motion state information of the UAV base is predicted by the neural network. The kinematic interference caused by the floating of the UAV base can be effectively offset by the homogeneous transformation principle. At the same time, considering the actuator limitation and trajectory tracking ability of the manipulator itself, a model predictive controller is designed to optimize the acceleration of the end of the manipulator actuator, which significantly improves the end precision of the flying manipulator under the floating interference of the base, and ensures the accurate completion of the active operation task. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is a design flow chart of an anti-interference pointing control method for the end of a flying mechanical arm of the present invention;

[0059] Figure 2 This is the structural diagram of the flight robotic arm system. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0061] The present invention takes a general flying robotic arm system consisting of a multi-rotor UAV and a multi-degree-of-freedom robotic arm as an example to illustrate the specific implementation of the system and method. When the flying robotic arm performs high-precision active operation tasks, it has very high requirements on the accuracy of the actuator end and its own safety.

[0062] like Figure 1 As shown, the specific implementation steps of the anti-interference pointing control method of the flight robot arm end of the present invention are as follows:

[0063] The first step is to establish the kinematic equation of the end position of the flight robot arm actuator in the inertial coordinate system and quantify the impact of the floating base interference on the end:

[0064] In view of the multi-degree-of-freedom characteristics of the flying robot arm, and considering the position offset between the robot arm base and the center of mass of the drone, the kinematic model of the end position of the flying robot arm actuator is established according to the principle of homogeneous transformation, which is expressed as follows:

[0065]

[0066] Where P e represents the position of the end of the flight robot arm actuator in the inertial system, R e is the rotation matrix of the end of the flight robot arm actuator in the inertial system; P b represents the position of the center of mass of the drone base in the inertial system, R b is the rotation matrix of the center of mass of the drone in the inertial system; Indicates the position of the end of the robotic arm actuator in the drone coordinate system, represents the rotation matrix of the end of the manipulator actuator in the drone coordinate system; in addition, R b The calculation formula is as follows:

[0067]

[0068] In the formula, They respectively represent the roll angle, pitch angle and attitude angle of the flying robot arm in the inertial system.

[0069] Due to the multi-link characteristics of the robotic arm, the floating interference of the drone is transmitted to the end of the flying robotic arm actuator, which will have an error amplification effect. In order to quantify the kinematic interference caused by the floating of the drone base on the end of the robotic arm actuator, a qualitative analysis is performed based on the kinematic equation established above. The drone has an error of ±2cm in position along the x, y, and z axes, and a deviation of ±5° in attitude. The position offset of the end of the flight robot arm actuator is calculated. The results are as follows:

[0070]

[0071] From the data in the table, it can be seen that under the multi-link action of the robotic arm, the floating interference of the drone base is transmitted to the end of the actuator, which will cause a serious error amplification effect and deteriorate the end-actuator accuracy of the flying robotic arm. Therefore, the floating base interference of the drone needs to be solved urgently to ensure that the flying robotic arm can complete high-precision operation tasks.

[0072] The second step is to redesign the motion trajectory of the robotic arm through the principle of homogeneous transformation, and design a neural network to predict the motion state of the drone base:

[0073] The motion state of the end of the flight robot arm actuator is expressed under SE (3) in the following form:

[0074]

[0075] Where SE(3) represents the transformation motion of the rigid body, including rotation and translation;

[0076] in, Represents the homogeneous transformation matrix of the end of the flight robot arm actuator in the inertial coordinate system, is the homogeneous transformation matrix of the center of mass of the drone base in the inertial coordinate system, is the homogeneous transformation matrix of the end of the flight robot arm actuator in the drone coordinate system. I represents a 1×1 unit matrix. If the next motion state of the drone base can be obtained The homogeneous transformation principle can be used to transform the trajectory of the robotic arm in the inertial system into its own base coordinate system (i.e., the drone coordinate system) to offset the floating interference of the drone base.

[0077] The homogeneous transformation takes the following form:

[0078]

[0079] Among them, t is the current control time and i is the step size of forward prediction.

[0080] Therefore, the trajectory tracking problem of the flying robot arm in the inertial system is converted into the trajectory tracking problem of the robot arm in the UAV coordinate system. The high precision of the actuator end of the flying robot arm can be guaranteed by making full use of the high precision of the robot arm itself.

[0081] When the flying manipulator completes high-precision active operation tasks, it usually requires the UAV base to be in a quasi-static hovering state. However, due to the model uncertainty interference caused by the internal center of mass offset of the UAV and the dynamic coupling interference caused by the movement of the manipulator, the UAV base will continue to shake near the hovering point. Based on the above analysis, the motion state of the UAV base at the next moment can be expressed by the following nonlinear system:

[0082]

[0083] Among them, x(t) represents the current motion state of the UAV base, u(t) represents the current control input of the UAV base, Represents the collective interference to the drone base, including the model uncertainty interference related to the current motion state x(t) of the drone base and the angle of the robotic arm Angular velocity Angular acceleration Since the next motion state of the drone base depends on the current motion state of the drone and the robotic arm, a neural network with the drone and robotic arm states as input is established to predict the next motion state of the drone. The neural network form is as follows:

[0084] η(t+1)=c(-aη(t)+g(W in κ(t)+Wη(t)))

[0085] f out =Π(W in κ(t)+W out η(t+1))

[0086] Where η(t) represents the state value of the intermediate neuron of the neural network. κ(t) represents the input of the neural network, including the motion state of the drone base and the robotic arm. W,W in and W out They represent the connection weights between the intermediate neurons, input neurons and output neurons of the neural network respectively. c and a are the learning parameters in the neural network. g and Π represent the function mapping relationship in the neural network respectively.

[0087] Step 3: Design a model predictive control planner to optimize the acceleration of the robot end and map the control signal to the angle of each joint:

[0088] Combining the UAV motion state and kinematic homogeneous transformation relationship output by the neural network, and considering the collision avoidance between the UAV and the robotic arm and the physical limitations of the robotic arm actuator, the main focus is on the anti-interference pointing problem of the actuator end of the flying robotic arm actuator: the motion trajectory of the robotic arm actuator end is mapped to the trajectory of each joint through the Jacobi matrix. The model predictive controller adjusts the control amount of the robotic arm under various constraints, and achieves the optimal actuator control input while achieving the control target.

[0089] The kinematic state space modeling of the end of the flight robot arm actuator is as follows:

[0090]

[0091] in, represents the position (3D) and velocity (3D) of the end of the robot arm actuator at time j; h j+1 represents the position (3D) and velocity (3D) of the end of the robot arm actuator at time j+1; uj represents the acceleration of the end of the robot arm actuator at time j; y j+1 Represents the output of the state space at time j+1;

[0092] In addition, the values ​​of the system matrix A, input matrix B, and output matrix C are as follows:

[0093]

[0094] In the formula, σt represents the sampling time interval, I3 represents the 3×3 unit matrix, and 03 represents the 3×3 zero matrix. The cost function of the model prediction controller is further designed, while considering various constraints such as collision avoidance between the drone frame and the robotic arm and acceleration constraints at the end of the robotic arm actuator. The form is as follows:

[0095]

[0096] Where J represents the total cost function, u represents the input matrix of model predictive control; J1, J2, and J3 are the tracking error cost function, the control input cost function, and the cost function for avoiding collision at the end of the manipulator actuator, respectively, which will be introduced in detail below. W1, W2, and W3 are the weight matrices of the tracking error cost function, the control input cost function, and the cost function for avoiding collision at the end of the manipulator actuator, respectively, and are adjustable parameters. min represents the minimum value of function J, where the optimization input u is the independent variable of function J.

[0097] The tracking error cost function J1 at the end of the flight manipulator actuator is expressed as:

[0098]

[0099] in, Represents the homogeneous transformation matrix The position part; y(t+i) represents the actual position of the end of the robot arm actuator; N is the designed control domain length; t is the current control moment; i is the step size of the forward prediction; ||· represents the 2-norm of the vector.

[0100] The flight manipulator actuator end control input cost function J2 is expressed as:

[0101]

[0102] By adding the control input cost function J2, the system input can be smoothed to reduce the drastic changes in the acceleration of the end of the robot arm actuator.

[0103] The cost function J3 of avoiding collision at the end of the flight manipulator actuator is expressed as:

[0104]

[0105] in, and They represent the safety boundaries of the end motion of the robot arm actuator, Indicates selection and The larger value of It means to find the maximum value of each element in the vector.

[0106] Furthermore, the constraints satisfied by the model predictive control are as follows:

[0107] u min ≤u(t+i)≤u max

[0108] In the formula, u min and u max Indicates the limit of the acceleration of the end of the robot arm actuator, so as to ensure the stability of the flight robot system.

[0109] The contents not described in detail in the specification of the present invention belong to the prior art known to the professional and technical personnel in the field and are easily understood by the technical personnel in the field. The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for anti-interference pointing control of the end of a flight robot arm, characterized in that: The following steps are involved: The first step is to construct the kinematic equation of the end position of the flight robot arm actuator in the inertial coordinate system based on the principle of homogeneous transformation; and to quantitatively analyze the kinematic interference caused by the floating base of the UAV to the end of the flight robot arm actuator; In the second step, the reference trajectory of the manipulator in the inertial system is redesigned into the motion trajectory in the UAV coordinate system through homogeneous transformation, and a neural network with the UAV and manipulator states as input is designed to predict the motion state of the UAV base; Step 3: Design a model predictive controller to plan the expected acceleration of the end of the manipulator actuator and convert it into the angle of each manipulator servo; complete the anti-interference pointing control of the end of the flying manipulator actuator under the interference of the UAV floating base and achieve high-precision grasping tasks; Combining the UAV motion state output by the neural network with the kinematic homogeneous transformation relationship, and considering the collision avoidance between the UAV and the robotic arm and the physical limitations of the robotic arm actuator, the motion trajectory of the robotic arm actuator end is mapped to the trajectory of each joint through the Jacobi matrix to solve the anti-interference pointing problem of the actuator end of the flying robotic arm. The model predictive controller adjusts the control amount of the robotic arm under various constraints to achieve the optimal actuator control input while achieving the control target. The kinematic state space modeling of the end of the flight robot arm actuator is as follows: ； in, Indicates that the end of the robot arm actuator is 3D position and 3D velocity at a given moment; Indicates that the end of the robot arm actuator is 3D position and 3D velocity at a given moment; Indicates that the end of the robot arm actuator is The acceleration of the moment; Indicates that the state space is Output at the moment; System Matrix , input matrix And the output matrix The values ​​of are as follows: ； In the formula, represents the sampling time interval, express The identity matrix of express The zero matrix of the model prediction controller is further designed, taking into account the various constraints of collision avoidance between the drone frame and the robotic arm and the acceleration constraint of the end of the robotic arm actuator, in the following form: ； In the formula, represents the total cost function, represents the input vector of model predictive control; They are the tracking error cost function, the control input cost function and the cost function for avoiding collision at the end of the manipulator actuator; They are the weight matrices of the tracking error cost function, the control input cost function, and the robot actuator end collision avoidance cost function, which are adjustable parameters; Represents the function The minimum value of For function The independent variable; Tracking error cost function of the end effector of a flying robot arm It is expressed as: ； in, Represents the homogeneous transformation matrix Middle position part; Indicates the actual position of the end of the robot arm actuator; is the designed control domain length; is the current control moment; is the step size of forward prediction; It represents the 2-norm of a vector; Cost function of control input of the end effector of the flight manipulator It is expressed as: ； By adding the control input cost function Smoothing system inputs to reduce drastic changes in acceleration at the end of the robot arm actuator; Cost function for avoiding collision at the end of a flying robot arm actuator It is expressed as: ； in, and They represent the safety boundaries of the end motion of the robot arm actuator, Indicates selection and The larger value of It means to find the maximum value of each element in the vector.

2. The anti-interference pointing control method of the end of a flying robot arm according to claim 1, characterized in that: The specific steps of the first step are as follows: The kinematic model of the end position of the flight robot arm actuator is established according to the principle of homogeneous transformation, which is expressed as follows: In the formula, Indicates the position of the end of the flight robot arm actuator in the inertial system, is the rotation matrix of the end of the flight robot arm actuator in the inertial system; Indicates the position of the center of mass of the drone base in the inertial system, is the rotation matrix of the center of mass of the drone in the inertial system; Indicates the position of the end of the robotic arm actuator in the drone coordinate system, Represents the rotation matrix of the end of the robotic arm actuator in the drone coordinate system; The calculation formula is as follows: ； In the formula, They respectively represent the roll angle, pitch angle and attitude angle of the flying manipulator in the inertial system; Based on the kinematic equation established above, a qualitative analysis is performed to quantify the kinematic interference caused by the floating of the UAV base to the end of the robotic arm actuator.

3. The anti-interference pointing control method of the end of a flight robot arm according to claim 2, characterized in that: The specific steps of the second step are as follows: The motion trajectory of the robotic arm is redesigned through the principle of homogeneous transformation, and a neural network is designed to predict the motion state of the drone base: The motion state of the end of the flight robot arm actuator is expressed under the transformation motion of the rigid body including rotation and translation, as follows: ； in, Represents the homogeneous transformation matrix of the end of the flight robot arm actuator in the inertial coordinate system, is the homogeneous transformation matrix of the center of mass of the drone base in the inertial coordinate system, is the homogeneous transformation matrix of the end of the flight robot arm actuator in the drone coordinate system, express The identity matrix of The next movement state of the drone base , the trajectory of the robot arm in the inertial system is converted to its own base coordinate system, that is, the drone coordinate system, using the principle of homogeneous transformation to offset the floating interference of the drone base. is the current control moment, is the step size of forward prediction; The homogeneous transformation takes the following form: ； At this time, the trajectory tracking problem of the flying robot arm in the inertial system is converted to the trajectory tracking problem of the robot arm in the drone coordinate system. By making full use of the high precision of the robot arm itself, the high precision of the actuator end of the flying robot arm is guaranteed. The motion state of the drone base at the next moment is expressed by the following nonlinear function: ； in, Indicates the current motion status of the drone base Indicates the current control input of the drone base. Indicates the collective interference to the drone base, including the interference with its own status The model uncertainty related to the robot arm angle , angular velocity , angular acceleration The relevant dynamic coupling interferes with the two parts; A neural network is established that takes the state of the drone and the robotic arm as input to predict the next movement state of the drone. The form of the neural network is as follows: ； in, Represents the state value of the intermediate neuron of the neural network; Represents the input of the neural network, including the motion state of the drone base and the robotic arm; and Respectively represent the connection weights between the neural network intermediate neurons, input neurons and output neurons; and is the learning parameter in the neural network; and They represent the function mapping relationship in the neural network respectively.

4. The anti-interference pointing control method of the end of a flying robot arm according to claim 1, characterized in that: The constraints satisfied by the model predictive control are as follows: ； In the formula, and Indicates the limit of the acceleration of the end of the robot arm actuator, so as to ensure the stability of the flight robot system.

Images