A control method for flying robots based on characteristic models

Abstract

The flight robot control method based on the characteristic model solves the problem that the existing flight robot anti-disturbance control method requires a large amount of real-time status data of drones and the control method is complex, and belongs to the field of new aerial work drone control. The method of the present invention includes establishing an error characteristic model of the flight robot's attitude dynamics and identifying the characteristic parameters of the channels in the model. The identification method is a recursive least squares identification algorithm based on a dual-source error model. The identification process can be completed using input-output data. The attitude controller is designed based on the characteristic parameter vector obtained from the characteristic modeling part to achieve anti-disturbance control of the flight robot. There is no need to design additional adaptive algorithms, state observers, or other estimation methods based on the actual situation of the system itself to compensate for the torque changes caused by interference, and there is no need to adjust hyperparameters to improve the interference estimation accuracy under different environments.

Description

Technical Field

[0001] The present invention relates to a flight robot control method based on a characteristic model, which is applied to a rotor flight manipulator system equipped with a four-degree-of-freedom serial manipulator and requiring high-precision posture control, and belongs to the field of new aerial operation UAV control. Background Art

[0002] In recent years, with the continuous development of robotics technology, drones have been widely used in civilian applications such as aerial photography, industrial surveying, agricultural plant protection, and patrol inspections. Among these applications, flying robots, a new type of aerial intelligent system, have also attracted increasing attention. Flying robots typically consist of a multi-rotor drone and various types of attachments. A typical example is a flying robot arm composed of a drone and a multi-degree-of-freedom robotic arm. This structure enables them to perform a variety of tasks with high flexibility and the ability to move rapidly in three-dimensional space. Therefore, flying robot systems are well-suited for tasks such as grasping, transportation, and inspection.

[0003] High-precision control of flying robots is essential for achieving a variety of complex tasks. However, in actual operational scenarios, aircraft are inevitably subject to various interferences. Interferences can be divided into three categories: torques exerted on the aircraft by the external environment, such as wind disturbances; model uncertainties inherent in the aircraft, such as the effects of inaccurate dynamic model calculations; and coupled disturbances caused by the relative motion of the manipulator, if present. The presence of various interferences makes stable control of flying robots during aerial operations increasingly difficult. Therefore, how to suppress these various interferences and thus improve the control performance of flying robots during actual operations is a major focus of flying robot system research.

[0004] In existing research, there are two main approaches to dealing with multi-source interference. One is to treat all types of interference as external interference and utilize strategies such as adaptive algorithms to suppress it. For example, patent application number CN202211589931.0 proposes a UAV attitude control method based on an adaptive terminal sliding mode. To address the problem of unknown interference in quadrotor trajectory tracking, an adaptive law is designed to effectively estimate the upper bound of the interference. Patent application number CN202211681877.2 proposes a firefighting UAV attitude control method based on an improved active disturbance rejection controller. An extended state observer is designed to estimate the total disturbance of the UAV, thereby improving control effectiveness. Patent application number CN202211722767.6 proposes a quadrotor attitude control method based on a fast non-singular terminal sliding mode. This method achieves fast and robust control without requiring bounded interference. However, these estimation algorithms often require a large amount of UAV state information, such as angular velocity and angular acceleration. The excessive number of hyperparameters leads to a cumbersome parameter tuning process and limited algorithm versatility. The second approach is to distinguish coupling disturbances from other disturbances, model them, and feed them forward into the basic controller to achieve precise anti-disturbance. For example, patent application number CN202211319418.X proposes a method for accurately modeling coupling disturbances based on variable inertia parameters and introduces this method into the controller, effectively reducing the interference caused by the robotic arm on the drone. Patent application number CN202010801707 designs an anti-saturation attitude controller to address issues such as center of mass offset and base floating in arm-mounted rotor drones, effectively suppressing coupling disturbances. Similarly, patent application number CN202211165741.6 proposes a method for modeling coupling disturbances for rapid motion of a robotic arm with variable inertia parameters and feeds this method forward into an adaptive neural network controller for compensation. However, this method for accurately modeling coupling disturbances is overly complex, making it unsuitable for deployment on actual platforms. It is also unknown whether this method can maintain high modeling accuracy when the robotic arm is moving at high speeds. Furthermore, it is questionable whether the excessive number of cross-estimates in the modeling process will affect the accuracy of the disturbance modeling. In addition, the above-mentioned anti-disturbance control strategies are highly model-dependent, and an accurate dynamic model is essential, which is usually complex and difficult to obtain in practical applications. Summary of the Invention

[0005] In view of the problems that the existing anti-disturbance control method of flying robots requires the use of a large amount of real-time status data of drones and requires many hyperparameters to be debugged, which makes the control method complicated, the present invention provides a flying robot control method based on a characteristic model.

[0006] A flying robot control method based on a feature model of the present invention comprises:

[0007] Establish the error characteristic model of the flying robot's posture dynamics and identify

[0008]

[0009] in, represents the three characteristic parameters of channel i, i = φ, θ or ψ, φ, θ, ψ represent the roll signal, pitch signal and yaw signal respectively, e i (k) is the attitude angle tracking error of channel i, u i (k) is the control quantity output by the attitude controller to channel i, represents the estimated value of the attitude angle tracking error of channel i;

[0010] The discrete time sliding surface based on attitude control and the obtained Design attitude controller u i (k) To achieve control of the flying machine.

[0011] Preferably, identify The methods include:

[0012] Establish a dual-source error model based on model estimation error and control error:

[0013]

[0014] in, e ei (k) is the model estimation error obtained based on the true tracking error and the output of the error characteristic model, e ui (k) is the system control error obtained based on the true tracking error, E i (k) is the dual-source error output by the dual-source error model, φ i (k-1)=[e i (k-1),e i (k-2),u i (k-1)] T ,λ1,λ2,λ ui are all constants greater than 0;

[0015] Combining the dual-source error model and the error characteristic model of the flight robot's attitude dynamics, the least squares identification algorithm is used to identify the To identify.

[0016] As a preference, the least squares identification algorithm is:

[0017]

[0018] Among them, K i (k) represents the gain matrix, P i(k) represents the covariance matrix, λ is the forgetting factor, * T represents the transpose and I is the identity matrix.

[0019] As a preference, the posture controller u i (k):

[0020] u i (k)=u eqi (k)+u eli (k)+u sti (k);

[0021] Among them, the three control quantities u eqi (k),u eli (k),u sti (k) are:

[0022]

[0023] ξ 3i ,ξ 1i ,ξ 2i ,M i are all constants greater than 0, ΔT is the sampling time, tanh(·) is the hyperbolic tangent function, k si and α i are all constants greater than 0, γ i ∈(0,1),s i (k-1) represents the discrete-time sliding surface.

[0024] As an optimal solution, the discrete-time sliding surface is:

[0025]

[0026] Among them, k si and α i are all constants greater than 0, γ i ∈(0,1).

[0027] The beneficial effects of the present invention are that the feature modeling control method involved in the present invention is different from the model-based control method. As a data-driven strategy, it does not require precise dynamic analysis of the flying robot, nor does it require strict assumptions that the dynamic parameters of the system remain unchanged during flight. It is more suitable for aircraft control in complex aerial working conditions.

[0028] In addition, the control method of the present invention is divided into two parts: feature modeling and controller design. For the feature modeling part, its core is a recursive least squares identification algorithm based on a dual-source error model. Compared with other model identification methods, only input-output data (flight robot attitude angle data and control input data) is used to complete the identification process, and the algorithm is simple and uses fewer hyperparameters, which can greatly reduce the computational complexity. In addition, based on the input-output framework, the identified feature parameter vector contains dynamic information generated by various factors, such as the model uncertainty of the flight robot itself, external interference, and coupling disturbance of the robotic arm.

[0029] For the controller design part, compared with the existing flight platform anti-disturbance strategy, the attitude controller is designed based on the characteristic parameter vector obtained in the characteristic modeling part to realize the anti-disturbance control of the flying robot. There is no need to design additional adaptive algorithms, state observers and other estimation methods according to the actual situation of the system itself to compensate for the torque changes caused by interference, and there is no need to adjust hyperparameters to improve the interference estimation accuracy in different environments. Therefore, the control method based on the characteristic model has better platform applicability and more robust anti-disturbance performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 Schematic diagram of the control method of the present invention;

[0031] Figure 2 Schematic diagram of the control system based on the control method of the present invention. DETAILED DESCRIPTION

[0032] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0033] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0034] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0035] The flying robot control method of this embodiment includes:

[0036] Step 1: Identify the parameters in the error characteristic model of the flying robot's posture dynamics represents the three characteristic parameters of channel i, i = φ, θ or ψ, φ, θ, ψ represent the roll signal, pitch signal and yaw signal respectively;

[0037] Considering that the existing least squares algorithm only uses the model estimation error to complete the identification process, but the control error also has a significant impact on the identification accuracy, this step designs a dual-source error model based on the model estimation error and the control error, as follows:

[0038]

[0039] in, is the characteristic parameter vector used for controller design, Represents the characteristic parameters of the roll signal channel, Represents the characteristic parameters of the pitch signal channel, represents the characteristic parameter of the yaw signal channel, e i (k) is the attitude angle tracking error, e ei (k) is the model estimation error obtained based on the true tracking error and the output of the error characteristic model, e ui (k) is the system control error, φ i (k-1)=[e i (k-1),e i (k-2),u i (k-1)] T ,u i (k) is the control quantity output by the attitude controller, λ1, λ2, λ ui are all constants greater than 0, k-* represents different sampling times, E i (k) is the dual-source error output by the dual-source error model, u i (k) is the control quantity output by the attitude controller to channel i.

[0040] Based on the obtained dual-source error model, a least squares identification algorithm is designed to output a characteristic parameter vector containing all the dynamic information of the attitude loop, as follows:

[0041]

[0042] Among them, K i (k) represents the gain matrix, P i (k) represents the covariance matrix, λ is the forgetting factor, which is usually set to 0.97, T represents the transpose and I is the identity matrix.

[0043] The error characteristic model of the flying robot's attitude dynamics is:

[0044]

[0045] in, The output quantity for feature modeling is an estimate of the attitude angle tracking error.

[0046] Combining the dual-source error model and the error characteristic model of the flight robot's attitude dynamics, the least squares identification algorithm is used to identify

[0047] This embodiment realizes the modeling of the flying robot attitude control system by designing a recursive least squares method based on dual-source errors (model estimation error and control error). This method does not require the use of a large amount of real-time status data of the drone, has a small number of hyperparameters, greatly reduces the difficulty of debugging, and has strong versatility for different platforms. In addition, compared with other intelligent algorithms, such as neural networks and reinforcement learning, this method has a simple framework, greatly reduces the computational complexity, and is more conducive to deployment on flying robots and application in actual flight scenarios.

[0048] Step 2: Based on the feature modeling and the subsequent discrete-time sliding surface design, design the attitude controller to achieve high-precision control of the flying robot as follows:

[0049] First, the discrete time sliding surface is designed as:

[0050]

[0051] Among them, k si and α i are all constants greater than 0, γ i ∈(0,1).

[0052] Then, based on the sliding surface and feature modeling, the attitude controller is designed as follows:

[0053] u i (k)=u eqi (k)+u eli (k)+u sti (k)

[0054] Among them, the three control quantities u eqi (k),u eli (k),u sti (k) are:

[0055]

[0056] ξ 3i ,ξ 1i ,ξ 2i ,M i are all constants greater than 0, ΔT is the sampling time, and tanh(·) is the hyperbolic tangent function.

[0057] A discrete-time characteristic model attitude controller is constructed based on the designed discrete-time sliding surface and the characteristic parameters obtained in the first part. The proposed controller does not require additional observers or adaptive algorithms to estimate and compensate for disturbances, and can improve the high-precision attitude control performance of the rotorcraft under multi-source disturbances.

[0058] So far, the present invention has completed a method for anti-disturbance control of flying robots based on characteristic models. It is used to describe in detail the specific implementation scheme of a flying robot that performs aerial work tasks in response to multi-source interference in a real environment. The overall control system framework diagram of this method can be found in Figure 2 , where φ d ,θ d Obtained from the output of the basic position controller, ψ d is given and used to determine the relevant error in the attitude loop, Φ b Represents the three-dimensional real-time attitude angle, that is, Φ b =[φθψ] T , p b Represents the three-dimensional real-time position, that is, p b =[xyz],F d represents the desired lift force, output by the base position controller.

[0059] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely illustrative of the principles and applications of the invention. It should be understood that many modifications may be made to the illustrative embodiments, and that other arrangements may be devised, without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in ways other than those described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be employed in conjunction with other described embodiments.

Claims

1. A flying robot control method based on a feature model, characterized in that: The method comprises: Establish the error characteristic model of the flying robot's posture dynamics and identify in, represents the three characteristic parameters of channel i, i = φ, θ or ψ, φ, θ, ψ represent the roll signal, pitch signal and yaw signal respectively, e i (k) is the attitude angle tracking error of channel i, u i (k) is the control quantity output by the attitude controller to channel i, represents the estimated value of the attitude angle tracking error of channel i; The discrete time sliding surface based on attitude control and the obtained Design attitude controller u i (k) achieving control of the flying machine; The attitude controller u i (k): in i (k)=u eqi (k)+u eli (k)+u sti (k); Among them, the three control quantities u eqi (k),u eli (k),u sti (k) are: ξ 3i ,ξ 1i ,ξ 2i ,M i are all constants greater than 0, ΔT is the sampling time, tanh(·) is the hyperbolic tangent function, k si and α i are all constants greater than 0, γ i ∈(0,1),s i (k-1) represents the discrete-time sliding surface.

2. The flying robot control method based on feature model according to claim 1, characterized in that: Identify The methods include: Establish a dual-source error model based on model estimation error and control error: in, e ei (k) is the model estimation error obtained based on the true tracking error and the output of the error characteristic model, e ui (k) is the system control error obtained based on the true tracking error, E i (k) is the dual-source error output by the dual-source error model, φ i (k-1)=[e i (k-1),e i (k-2),u i (k-1)] T ,λ1,λ2,λ ui are all constants greater than 0; Combining the dual-source error model and the error characteristic model of the flight robot's attitude dynamics, the least squares identification algorithm is used to identify the To identify.

3. The flying robot control method based on feature model according to claim 2, characterized in that: The least squares identification algorithm is: Among them, K i (k) represents the gain matrix, P i (k) represents the covariance matrix, λ is the forgetting factor, * T represents the transpose and I is the identity matrix.

4. The flying robot control method based on feature model according to claim 1, characterized in that: The discrete-time sliding surface is: Among them, k si and α i are all constants greater than 0, γ i ∈(0,1).

5. A computer-readable storage device storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the feature-model-based flying robot control method according to any one of claims 1 to 4 are implemented.

6. A flying robot control device based on a feature model, comprising a storage device, a processor, and a computer program stored in the storage device and executable on the processor, characterized in that: The processor executes the computer program to implement the steps of the feature-model-based flying robot control method according to any one of claims 1 to 4.

7. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the flying robot control method based on the feature model as claimed in any one of claims 1 to 4 are implemented.

Images