Residual vibration suppression method for robot arm, control device, storage medium, and robot arm component

Abstract

The present invention provides a method for suppressing residual vibration of a robotic arm, a control device, a storage medium, and a robotic arm component. The method for suppressing residual vibration of the robotic arm includes: obtaining the dynamic characteristics of the robotic arm under a predetermined position and orientation and a predetermined load, and constructing a data set in which the dynamic characteristics correspond to the predetermined position and orientation and the predetermined load; training a deep neural network model based on the data set; predicting the target dynamic characteristics of the robotic arm using the trained deep neural network model based on the target position and orientation and the target load of the robotic arm; designing a vibration suppressor based on the target dynamic characteristics, and cooperating with a motion controller to control the movement of the robotic arm to the target position and orientation while suppressing residual vibration. By realizing online real-time prediction of dynamic characteristics at a low computational cost and adaptively designing a vibration suppressor, vibration suppression in an open work scenario can be realized.

Description

[Technical Field]

[0001] The present invention relates to the field of motion control technology, and in particular to a method for suppressing residual vibration of a robot arm, a control device, a storage medium, and a robot arm component. [Background technology]

[0002] Residual vibration is caused by the elasticity of the robot arm's joints and structural inertia forces and is typical of the step vibration phenomenon in dynamic systems. The vibration frequency is the system's natural frequency, and the maximum amplitude is usually positively correlated with the robot arm's motion acceleration. This means that sudden start-up or braking of the robot arm will cause large vibrations in the end effector. In efficiency-critical scenarios, residual vibrations limit the robot arm's motion speed, underutilizing the actuator's capabilities, and a lot of time is wasted waiting for the vibrations to subside. In scenarios requiring high precision from the end effector, such as collaborative surgery, overshooting the end effector position due to vibrations is unacceptable.

[0003] However, most current vibration suppression methods for robot arms only address a single frequency. Changes in the robot arm's own posture and tip load directly affect the system's inertia distribution, causing the system's dynamic characteristics to change over time. There is a strong nonlinear relationship between the robot arm's inertia distribution and the posture angles of each joint, making it difficult to accurately express the mapping relationship between the two using a simple function. Continuously formulating and solving complex dynamic equations for robot arms with different loads during real-time operation significantly increases the demands on controller hardware, generates a large amount of computation, and can significantly vary the computation time consumed by different postures, negatively impacting robot arm control. Therefore, a novel method for suppressing vibration in robot arms is urgently needed to address these issues. Summary of the Invention

[0004] The present invention addresses the shortcomings of the related art and provides a robot arm residual vibration suppression method, a control device, a storage medium, and a robot arm component to solve the problem in the related art that vibration suppression is only possible for a single frequency or the adaptability of frequencies corresponding to vibration suppression is low.

[0005] The present invention provides acquiring dynamic characteristics of the robot arm under a predetermined position and posture and a predetermined load, and constructing a data set in which the dynamic characteristics correspond to the predetermined position and posture and the predetermined load; training a deep neural network model based on the dataset; predicting target dynamic characteristics of the robot arm using the trained deep neural network model based on a target position, posture, and target load of the robot arm; designing a vibration suppressor based on the target dynamic characteristics, and controlling movement of the robot arm to a target position and posture while suppressing residual vibrations in cooperation with a motion controller; A method for suppressing vibrations in a robot arm is provided.

[0006] According to the above embodiment, the present invention constructs a dataset corresponding to dynamic characteristics, a predetermined position, posture, and load, trains a deep neural network model based on the dataset, and performs online prediction of the dynamic characteristics of a robot arm system with time-varying characteristics using the trained deep neural network model. By combining the trained deep neural network model with a vibration suppressor, residual vibration at the tip of the robot arm can be adaptively suppressed with very low computational cost. The present invention replaces the complex process of determining the characteristic frequencies of the dynamics of the robot arm with a deep neural network (DNN) model, thereby enabling rapid online prediction of the time-varying frequencies of the robot arm. Here, the dataset used to train the deep neural network model in the present invention is obtained by simulation analysis of dynamic parameters without relying on a physical prototype. Meanwhile, the number of samples in the sample space of the training dataset can be determined based on requirements such as the design range of each parameter of the robot arm and the vibration suppression effect, making it easy to operate and implement.

[0007] Furthermore, by combining a pre-trained deep neural network model with the vibration suppressor design, the target dynamic characteristics of the robot arm can be quickly predicted based on the target position, posture, and target load of the robot arm, allowing the robot arm to move smoothly to the target position. The required computational cost is extremely low, which can fully meet the requirements for rapid online execution of the robot arm.

[0008] In one embodiment, the step of acquiring dynamic characteristics of the robot arm under a predetermined position / posture and a predetermined load, and constructing a data set in which the dynamic characteristics correspond to the predetermined position / posture and the predetermined load, comprises: obtaining dynamic parameters of the robot arm joints; constructing a robot arm dynamics model based on a structural design scheme of the robot arm and dynamics parameters of the robot arm joints; and analyzing and acquiring dynamic characteristics of the robot arm under a predetermined position / posture and a predetermined load based on the robot arm dynamic model, and constructing a data set in which the dynamic characteristics correspond to the predetermined position / posture and the predetermined load.

[0009] In one embodiment, the step of obtaining the robot arm joint dynamics parameters comprises: obtaining an impulse response curve of the robot arm joint; and processing the impulse response curve to obtain equivalent stiffness, equivalent damping and equivalent inertia information for each joint of the robot arm.

[0010] In one embodiment, the step of constructing a robot arm dynamics model based on the dynamics parameters of the robot arm joints comprises: The method includes a step of combining the dynamics parameters of the robot arm joints with a three-dimensional model of the robot arm to construct the robot arm dynamics model.

[0011] In one embodiment, the step of analyzing and acquiring dynamic characteristics of the robot arm under a predetermined position / posture and a predetermined load based on the robot arm dynamic model, and constructing a data set in which the dynamic characteristics correspond to the predetermined position / posture and the predetermined load, comprises: constructing an independent variable data set using a predetermined position and posture of the robot arm and a predetermined load as independent variables; and using the robot dynamics model to obtain dynamic characteristics of the robot arm under conditions corresponding to combinations of each independent variable in the independent variable dataset, and constructing a response dataset.

[0012] In one embodiment, the step of training a deep neural network model based on the dataset comprises: training neural networks with different numbers of layers to determine an optimal number of layers; and training a neural network by different training set selection methods to obtain a mapping relationship between the dynamic characteristics and the data set corresponding to the predetermined position and posture and the predetermined load.

[0013] In one embodiment, the step of predicting target dynamic characteristics of the robot arm using the trained deep neural network model based on the target position and posture of the robot arm and the target load includes: The method includes a step of introducing the trained deep neural network model into a control device of the robot arm, and inputting information on a target position, posture, and target load of the robot arm into the trained deep neural network model to determine a target dynamic characteristic of the robot arm.

[0014] In one embodiment, the robot arm dynamic model is a robot arm simulation dynamic model, and the step of analyzing and obtaining dynamic characteristics of the robot arm under a predetermined position posture and a predetermined load based on the robot arm dynamic model includes: Obtaining actual dynamic characteristics of the robot arm prototype at a predetermined position and under a predetermined load through experiments; obtaining a simulated dynamics characteristic of the robot arm in the same state as an experiment under a predetermined position, posture, and load based on the robot arm simulation dynamics model; and modifying the robot arm simulation dynamics model to determine the dynamics characteristics.

[0015] In one embodiment, the step of experimentally obtaining actual dynamic characteristics of the robot arm prototype at a predetermined position and a predetermined load comprises: The method includes determining, through experimentation, the natural frequencies, modal damping ratios, and vibration modes of the robot arm prototype at multiple positional orientations and under multiple loads.

[0016] In one embodiment, the step of modifying the robot arm simulation dynamics model to determine the dynamics characteristics comprises: The method includes the steps of comparing the actual dynamic characteristics with the simulated dynamic characteristics, determining sensitive parameters by sensitivity analysis, and modifying the sensitive parameters to determine the dynamic characteristics.

[0017] The present invention further provides a robot arm component including an arm body and a controller, wherein the controller is connected to the arm body and controls movement of the arm body to a target position and posture, thereby implementing the above-mentioned method for suppressing vibration of a robot arm. Characterized by

[0018] The present invention further provides a control device including a processor, a memory, and at least one program, wherein the memory is communicatively connected to the processor, the at least one program is stored in the memory, and the at least one program, when executed by the processor, performs the aforementioned method for suppressing vibration of a robot arm.

[0019] The present invention further provides a computer-readable storage medium storing a computer program, which, when executed by a control device, performs the above-mentioned method for suppressing vibration of a robot arm.

[0020] Additional aspects and advantages of the invention will be set forth in part in the description that follows, and in part will be apparent from the description, or may be learned through practice of the invention. [Brief explanation of the drawings]

[0021] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present invention and, together with the description, serve to explain the principles of the invention. [Figure 1]FIG. 2 is a schematic diagram showing the flow of a method for suppressing residual vibration of a robot arm provided in an embodiment of the present invention. [Figure 2] FIG. 2 is a schematic diagram showing the flow of step S100 of the residual vibration suppression method for a robot arm provided in an embodiment of the present invention. [Figure 3] FIG. 4 is a schematic diagram showing the flow of step S110 of the residual vibration suppression method for a robot arm provided in an embodiment of the present invention. [Figure 4] FIG. 10 is a schematic diagram showing the flow of step S130 of another method for suppressing residual vibration of a robot arm provided in an embodiment of the present invention. [Figure 5] FIG. 10 is a schematic diagram showing the flow of step S200 of another method for suppressing residual vibration of a robot arm provided in an embodiment of the present invention. [Figure 6] FIG. 4 is a schematic diagram showing the flow of step S130 of the residual vibration suppression method for a robot arm provided in an embodiment of the present invention. [Figure 7] FIG. 1 is a schematic diagram showing a three-dimensional model of a robot arm provided in an embodiment of the present invention. [Figure 8] 8 shows the vibration response spectrum of the tip of the three-dimensional model of the robot arm in FIG. 7. [Figure 9] FIG. 10 is a comparison diagram of test results before and after vibration suppression using the vibration suppression method for a robot arm provided in an embodiment of the present invention. [Figure 10] FIG. 2 is a schematic diagram showing the structure of a control device provided in an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0022] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. Where the following description refers to the drawings, like numbers in different drawings refer to the same or similar elements unless otherwise noted. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatus, devices, and methods consistent with certain aspects of the present invention, as detailed in the appended claims.

[0023] The terms used in the present invention are merely for the purpose of describing particular embodiments and are not intended to limit the present invention. As used in the present invention and the appended claims, the singular forms "a," "the," and "said" are intended to include the plural forms unless the context clearly indicates otherwise. Also, as used herein, the term "and / or" should be understood to refer to and encompass any or all possible combinations of one or more associated listed items.

[0024] The dynamic model of residual vibrations, typically observed in robot arms with flexible joints, can be simplified to a multi-degree-of-freedom mass-spring-damper system, a typical structural dynamics model. The system's natural frequency is jointly determined by spring stiffness, damping, and mass distribution, and the dominant frequency of residual vibrations is the system's lower-order natural frequency. Related research has shown that the reduction gears within the joints are the primary contributors to elasticity and damping characteristics, and their equivalent stiffness and damping can be considered constants. However, changes in the robot arm's posture and tip load directly affect the system's inertia distribution, resulting in time-varying dynamic characteristics. The relationship between the robot arm's inertia distribution and the posture angles of each joint is highly nonlinear, based on the sequential multiplication of coordinate transformation matrices. Given the domains of the joint posture angles and tip load inertia, and the range of the robot arm's natural frequency, it is difficult to express the mapping between the two in a concise functional form. Related research has attempted to suppress vibration by analyzing response data and extracting system features. However, this approach suffers from natural hysteresis, and its application only shortens the vibration damping time. Its suppression of maximum overshoot is relatively limited, making it incapable of supporting fine-grained operation. Another solution involves introducing a model for prediction and designing a real-time controller (feedback or feedforward) based on the prediction results. This approach can resolve hysteresis by extracting system features based on the response. However, this requires high hardware costs and is not applicable to high-speed process systems. A better solution is yet to be found. Continuously formulating and solving complex dynamics equations for robot arms with different loads during real-time operation of a robot arm significantly increases the demands on controller hardware, generates a large amount of computation, and can significantly vary the computation time consumed by different postures, negatively impacting robot arm control.

[0025] The present invention aims to provide a method and control device for suppressing residual vibration of a robot arm based on a deep neural network, thereby solving the above-mentioned technical problems of the related art.

[0026] Hereinafter, a method and control device for suppressing residual vibration of a robot arm based on a deep neural network according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings. Note that the features in the following embodiments can be complemented or combined with each other unless they are inconsistent.

[0027] FIG. 1 is a schematic diagram showing the flow of a method for suppressing residual vibration of a robot arm provided in an embodiment of the present invention, and includes the following steps S100 to S400.

[0028] In S100, the dynamic characteristics of the robot arm under a predetermined position / posture and a predetermined load are acquired, and a data set is constructed in which the dynamic characteristics correspond to the predetermined position / posture and the predetermined load.

[0029] In this embodiment, the dynamic characteristics are the natural frequency f i , modal damping ratio ζ i , and vibration mode φ i The given position and posture are the joint angle coordinates q j (j=1,...,7), the predetermined load is m, and the data set {(q j ,m)|(f i ,ζ i ,φ i )}.

[0030] In some embodiments, the dynamic characteristic is a natural frequency f i , modal damping ratio ζ i Those skilled in the art can flexibly set it.

[0031] The robot arm and integrated joints of this embodiment can be applied to, for example, industrial robots, service robots, or specialized robots. As an example, the robot arm can be applied to a piano-playing robot.

[0032] In this embodiment, the robot arm has n degrees of freedom, where n is a positive integer. For example, the robot arm has 7 degrees of freedom.

[0033] In this embodiment, the method for acquiring the dynamic characteristics of the robot arm under a predetermined position and posture and a predetermined load may be a conventional calculation method such as an analytical method or dynamic modeling, and the present invention does not impose any specific limitations.

[0034] The S200 trains a deep neural network model based on the dataset.

[0035] In this embodiment, the data set {(q j ,m)|(f i ,ζ i ,φ i )} to train a deep neural network, and calculate the domain {q j ,m} and construct the natural frequency f i , modal damping ratio ζ i , vibration mode φ i is used as a response parameter and the range {f i ,ζ i ,φ i} and construct the design factor (q j ,m) and dynamic property parameters (f i ,ζ i ,φ i ) to obtain a mapping relationship. This embodiment uses a pre-trained deep neural network model to perform online prediction of the time-varying dynamic characteristics of the robot arm system, which can avoid the computational costs of solving the robot arm dynamics through online simulation, is more advantageous for the online adaptive design of the vibration suppressor, and can fully meet the needs of rapid online execution of the robot arm.

[0036] In S300, a trained deep neural network model is used to predict the target dynamic characteristics of the robot arm based on the target position, posture, and target load of the robot arm.

[0037] In this embodiment, a trained deep neural network model is introduced into the robot arm control device, and after receiving the information of the target position, posture and target load of the robot arm, the target dynamic characteristic parameters {f i ,ζ i ,φ i To predict and determine}, information on the target position, posture and load of the robot arm is input into a trained deep neural network model.

[0038] In S400, a vibration suppressor is designed based on the target dynamic characteristics, and in cooperation with the motion controller, it controls the movement of the robot arm to the target position and posture while suppressing residual vibrations.

[0039] In the vibration suppressor of this embodiment, an input shaper may be selected that shapes the initial motion control command input to the robot arm so as to suppress residual vibration at the tip of the robot in the process of controlling the movement of the robot arm to the target position and posture. The input shaper shapes the target dynamics characteristic parameter {f i ,ζ i ,φ i Specifically, the input parameters of each joint are convolved with the corresponding input shaper to obtain the final input parameters of each joint, which then work in cooperation with the motion controller to drive the robot arm to complete the planned motion and smoothly reach the target position, thereby suppressing vibration at the tip.

[0040] In some embodiments, the dynamic characteristic is a natural frequency f i , modal damping ratio ζ i Those skilled in the art can flexibly set it.

[0041] It should be noted that different vibration suppressors may be used to achieve vibration suppression, and although the present invention shows the embodiment of the input shaper, it is not limited thereto, and those skilled in the art can flexibly configure it according to the actual situation.

[0042] Furthermore, methods for suppressing residual vibrations in robot arms using control in this field can be broadly divided into two types: feedback control and feedforward control. Feedback control, typified by proportional-integral-derivative (PID) control based on an inverse dynamics model, is easy to implement and robust, but is sensitive to the dynamic parameters of the robot arm, cannot be applied to variable loads or high-speed motion control, and is only effective at suppressing maximum overshoot of residual vibrations. Furthermore, delays occur in the feedback loop, and sensors are required to detect the feedback controller and the tip position and orientation, which increases the complexity and hardware costs of the system.

[0043] Common feedforward control methods include optimal trajectory planning and input shaping. These methods do not require feedback sensors, but input control also fails to account for system variability. Currently, robot arm trajectory planning is based on cubic polynomials or cubic B-spline curve interpolation, but this generates trajectories with rapidly changing derivatives at the start and stop points, inducing vibrations during startup and shutdown. Trajectories based on non-uniform rational B-spline (NURBS) curves can better eliminate vibrations, but solving NURBS curves is complex and computationally intensive, a common issue with trajectory planning methods using higher-order polynomials and other complex curves. Most of these algorithms are currently researched at the algorithm level, and some of them lack generality. When simultaneously considering vibration suppression and comprehensive optimization of factors such as time and energy consumption, weighting factors are commonly used, but the selection of weighting factors is highly subjective. Furthermore, optimal trajectories often contradict each other in terms of operating time and maximum drive output.

[0044] Input shaping uses feedforward control to remove specific frequency components from the system input and suppress the corresponding frequency components in the response. Suppressing specific frequency components in the system response using feedforward control essentially involves filtering out the frequency components from the frequency characteristics of the system input. Compared to a full notch filter, input shaping technology uses only multiple pulses with a specified delay and amplitude ratio to achieve filtering. It has advantages such as a simple shape, small delay, and no complex ripples. It has a signal smoothing effect in the time domain and an effect equivalent to a comb filter in the frequency domain. Input shaping is highly suitable for suppressing residual vibrations in robot arms. However, because the notch frequency of the pulse sequence is directly determined by the time interval between pulses, it is necessary to solve the problem of predicting the target frequency while the robot arm's natural frequency varies over time. Therefore, this invention introduces a deep neural network model into a robot arm vibration suppression method and combines it with an input shaper to achieve accurate and rapid suppression of residual vibrations at the tip of the robot arm.

[0045] The present invention constructs a dataset that corresponds dynamic characteristics to a predetermined position, posture, and load. A deep neural network model is then trained based on the dataset. The trained deep neural network model is used to perform online prediction of the dynamic characteristics of a robot arm system with time-varying characteristics. By combining this with a vibration suppressor, residual vibrations at the tip of the robot arm can be adaptively suppressed with very low computational cost. The present invention replaces the complex process of determining the characteristic frequencies of the robot arm's dynamics with a deep neural network (DNN) model, enabling rapid online prediction of the time-varying frequencies of the robot arm. The dataset used to train the deep neural network model in the present invention is obtained by simulating and analyzing dynamic parameters without relying on a physical prototype. Meanwhile, the number of samples in the sample space of the training dataset can be determined based on requirements such as the design range of each parameter of the robot arm and the vibration suppression effect, making it simple to operate and easy to implement.

[0046] Furthermore, by combining a pre-trained deep neural network model with the vibration suppressor design, the target dynamic characteristics of the robot arm can be quickly predicted based on the target position, posture, and target load of the robot arm, allowing the robot arm to move smoothly to the target position. The required computational cost is extremely low, which can fully meet the requirements for rapid online execution of the robot arm.

[0047] In some embodiments, step S100 includes the following steps S110 to S130 as shown in FIG.

[0048] In S110, the dynamic parameters of the robot arm joints are acquired.

[0049] The dynamics parameters in this embodiment include equivalent stiffness, equivalent damping and equivalent inertia of rotating parts.

[0050] In this embodiment, the equivalent stiffness, equivalent damping, and equivalent inertia of rotating parts of each joint of different models of robot arms (physical prototypes) are measured by dynamics experiments on each joint individually.

[0051] The equivalent stiffness, equivalent damping, and equivalent inertia of the rotating parts in this embodiment are determined by measurement and calculation methods based on conventional vibration theory, and the present invention does not impose any specific limitations thereon.

[0052] In S120, a robot arm dynamics model is constructed based on the structural design scheme of the robot arm and the dynamics parameters of the robot arm joints.

[0053] In this embodiment, a robot arm dynamics model, specifically a dynamics simulation model of the robot arm, is constructed based on the dynamics parameters of the robot arm and the three-dimensional model of the robot arm, where the structural design scheme includes geometric information of the robot arm, specifically, the geometric information includes the length of each arm segment between each joint of the robot arm, and the relative positions and angles of each joint.

[0054] In S130, based on the robot arm dynamics model, the dynamics characteristics of the robot arm under a predetermined position / posture and a predetermined load are analyzed and obtained, and a data set corresponding to the dynamics characteristics and the predetermined position / posture and the predetermined load is constructed.

[0055] In some embodiments, the dynamics model of the robot arm is constructed by MATLAB® / Simulink (or ADAMS) multibody system dynamics simulation software, the structural model of each arm segment of the robot arm is introduced by the CAD model, and the dynamics parameters obtained in step 100 are introduced into the original dynamics model of the robot arm to obtain a final dynamics model of the robot arm with the same or similar end residual vibration behavior as the physical prototype through simplification.

[0056] In some embodiments, step S110 includes the following steps S111 to S112 as shown in FIG.

[0057] In S111, the impulse response curve of the robot arm joint is obtained.

[0058] In this embodiment, a robot arm equipped with a counterweight is fixedly attached to the movable end of the joint, an impulse input is applied to the joint, and the response parameters of the counterweight are measured to obtain an impulse response curve.

[0059] In S112, the impulse response curve is processed to obtain equivalent stiffness, equivalent damping and equivalent inertia information for each joint of the robot arm.

[0060] In this embodiment, the dynamics parameters of each joint are obtained by processing the response signal of the impulse response curve.

[0061] In some embodiments, step S130 includes the following steps S131 to S132, as shown in FIG.

[0062] In S131, an independent variable data set is constructed using a predetermined position and posture of the robot arm and a predetermined load as independent variables.

[0063] In this embodiment, each joint angle coordinate q j (j=1,…,7), construct an 8-dimensional vector {q1,…,q7,m} as the independent variable vector with tip load m. Each element is normalized by the upper and lower bounds of each angle and the range of the load limit. To construct the independent variable dataset, we use the Latin Hypercube method to create a vector of size n. DOE A ×8 normalized DOE (Design of Experiments) matrix is ​​constructed, where each row is a random normalized independent variable combination, n DOEThe value of must meet the requirements for deep neural network model training in step S400. Here, each joint angle coordinate corresponds to the broad displacement of each degree of freedom of the robot arm, i.e., the rotation angle of a revolute joint or the expansion and contraction of a sliding joint.

[0064] In S132, the robot dynamics model is used to obtain the dynamic characteristics of the robot arm under conditions corresponding to the combinations of the independent variables in the independent variable dataset, and a response dataset is constructed.

[0065] In this embodiment, the combinations of independent variables are extracted row by row from the DOE matrix in step S310, denormalized, and input as initial state parameters into the robot arm dynamics model. A third-order polynomial quasi-trapezoidal wave pulse signal of appropriate magnitude is applied to drive each joint to start and stop rapidly in sequence, with the pulse interval length required to damp the tip vibration to a sufficiently low level. The dynamics transfer function between each joint and the tip vibration response is analyzed segment by segment, and the corresponding dynamics characteristics are extracted. A dynamics characteristic parameter matrix of the robot arm is constructed, and a response data set is created. Here, the dynamics characteristic parameters include {f i ,ζ i ,φ i}, where i is the mode number of interest.

[0066] In some embodiments, the dynamic characteristic is a natural frequency f i , modal damping ratio ζ i Those skilled in the art can flexibly set it.

[0067] In some embodiments, step S200 includes the following steps S210 to S220, as shown in FIG.

[0068] In S210, neural networks with different numbers of layers are trained to determine the optimal number of layers.

[0069] The optimum number of layers in this embodiment is the minimum number of layers that satisfies the accuracy requirements.

[0070] In S220, the neural network is trained by different training set selection methods to obtain the mapping relationship of the data set corresponding to the dynamic characteristics and the predetermined position and posture and the predetermined load.

[0071] In this embodiment, after determining the optimal number of layers in step S210, the neural network is trained using different training set selection methods to obtain the mapping relationship between the dynamic characteristics and the data sets corresponding to a predetermined position, posture, and load. The obtained deep neural network has the ability to predict the dynamic characteristic parameters of the robot arm based on the combination of independent variables.

[0072] In some embodiments, the robot arm dynamics model is a robot arm simulation dynamics model, and as shown in FIG. 6 , in step S130, the step of analyzing and obtaining the dynamics characteristics of the robot arm under a predetermined position / posture and a predetermined load based on the robot arm dynamics model includes the following steps S130 to S133:

[0073] In S1301, the actual dynamic characteristics of the robot arm prototype at a given position, posture and under a given load are obtained through an experiment.

[0074] In this embodiment, a physical prototype of the robot arm is tested to obtain the actual dynamic characteristics of the robot arm prototype. Specifically, the dynamic characteristics of the physical prototype are tested under multiple position postures and multiple loads, and the dynamic characteristics of the robot arm prototype are obtained by calculating the natural frequency f i , modal damping ratio ζ i , and vibration mode φ iIn one example, taking into consideration complex and variable working conditions in the actual operation of the robot arm, several typical position and postures are selected and combined with a plurality of different loads to form test states, and the dynamic characteristics of the robot arm are tested in each test state.

[0075] In S1302, based on the robot arm simulation dynamics model, the simulation dynamics characteristics of the robot arm in the same state as in the experiment under a predetermined position, posture and load are obtained.

[0076] In this embodiment, in step S1301, the initialization parameters of each test state are used to obtain the simulated dynamic characteristics of the robot arm in the corresponding state based on the robot arm dynamic model obtained in step S120.

[0077] In S1303, the robot arm simulation dynamic model is modified to determine the dynamic characteristics.

[0078] It should be noted that steps S1301 to S1303 of the present invention may be performed before step S131, or may be performed after step S131 and before step S132. The present invention is not particularly limited.

[0079] In this embodiment, the actual dynamic characteristics are compared with the simulated dynamic characteristics, sensitive parameters are determined through sensitivity analysis, and the sensitive parameters are modified to determine the dynamic characteristics. Here, the sensitive parameters are simulation parameters that significantly affect the deviation between the actual dynamic characteristics and the simulated dynamic characteristics as a result of comparing the two. By modifying the sensitive parameters within a reasonable range, the dynamic model can more accurately simulate the dynamic behavior of a physical prototype, and can be used to analyze the dynamic characteristics of a robot arm.

[0080] In this embodiment, the dynamics model is further modified by combining a physical prototype and a three-dimensional model of the robot arm, and the final dynamics characteristics are closer to the kinematic behavior of the physical prototype and therefore have higher accuracy.

[0081] It should be noted that the modification of the dynamic model in the embodiment of the present invention depends on a physical prototype, and if a physical prototype is not available, the step can be skipped without affecting the implementation of the entire method.

[0082] To facilitate understanding, this invention presents a specific application scenario. For example, when a piano-playing robot plays the piano, its seven-degree-of-freedom bionic arm moves between different keys. The rapid start and stop of the arm induces hand vibration. Due to the high stiffness of the bionic arm, this phenomenon is a typical residual vibration problem of robotic arms with flexible joints. This problem can be simplified as a typical vibration phenomenon formed by a multi-degree-of-freedom mass-spring-damper system under step excitation in structural dynamics. In practical applications, the vibration state of the tip of a robotic arm is typically required to stabilize and stop when the end effector, such as a hand or tool, reaches the target position without significant overshoot or vibration.

[0083] Two aspects must be considered in addressing the above problem. First, the analysis of the dynamic behavior of a robot arm system reveals that the system's natural frequency—the frequency of the system's step vibration—is jointly determined by spring stiffness, damping, and mass distribution. Research in related literature has shown that the reduction gears within the joints are the main contributors to their elasticity and damping characteristics, and their equivalent stiffness and damping can be considered constants. However, changes in the robot arm's posture and tip load directly affect the system's inertia distribution, resulting in time-varying vibration characteristics. Furthermore, the relationship between the inertia distribution and the joint's posture angle is highly nonlinear, making it difficult to express the mapping from joint posture angle and tip load to the robot arm's natural frequency in a simple functional form. The complexity of this mapping relationship is one of the challenges in residual vibration suppression. Second, when high-speed, stable stopping is required after the tip reaches the target position, feedforward control is clearly the preferred method. Other vibration suppression methods based on feedback control or response measurement cannot avoid hysteresis. Combining these two aspects, a versatile residual vibration suppression method must be able to more accurately predict the time-varying natural frequency of a robot arm. Therefore, the present invention can solve the problems arising from the above application scenarios by providing a method for vibration suppression of a robot arm that more accurately predicts the time-varying natural frequencies of the robot arm.

[0084] Figure 7 shows a simulation model of a robot arm used in the above application scenario for a piano-playing robot. The robot arm simulation model in Figure 7 is a seven-joint robot arm, consisting of a first joint 1, a second joint 2, a third joint 3, a fourth joint 4, a fifth joint 5, a sixth joint 6, and a seventh joint 7, which are connected in sequence for rotation. Trapezoidal pulses are applied sequentially to the seven joints in a certain posture (maintaining equal jerk magnitudes during acceleration and deceleration). The response spectrum of the tip vibration is shown in Figure 8. The response spectrum reveals that there are at least two dominant vibration frequencies in the arm. This embodiment of the present invention solves the problem of residual vibration at the tip of a piano-playing robot's robot arm and meets the high posture adjustment time requirements for piano-playing tasks.

[0085] As an example, the residual vibration of a robot arm was tested according to the application scenario described above. Figure 9 shows a comparison of the test results of a robot arm dynamic model before and after vibration suppression according to this embodiment. Comparing the results before and after vibration suppression, among the 1,000 groups of test cases, the average reduction rate was approximately 82.5%, and the total vibration amplitude was reduced by approximately 86.8%. The amplitude reduction rates were over 95% in 107 groups, 90%-95% in 273 groups, and 80%-90% in 377 groups. It can be seen that the robot arm vibration suppression method provided by this embodiment of the present invention can significantly improve the vibration suppression effect.

[0086] Based on the same inventive idea, the present invention further provides a robot arm component, including an arm body and a controller, wherein the controller is connected to the arm body, controls the movement of the arm body to a target position and posture, and implements the above-mentioned method for suppressing vibration of a robot arm.

[0087] Based on the same inventive idea, the present invention further provides a control device 500 including a processor 510, a memory 520, and at least one program 530 as shown in Figure 10, wherein the memory 520 is communicatively connected to the processor 510, the at least one program 530 is stored in the memory and executed by the processor 510, and the at least one program 530 is arranged to implement the above-mentioned method for suppressing vibration of a robot arm.

[0088] Based on the same inventive idea, the present invention further provides a computer-readable storage medium storing a computer program, which, when executed by a control device, implements the aforementioned method for suppressing vibration of a robot arm.

[0089] The above-described embodiments of the present invention can complement each other unless they are inconsistent.

[0090] As will be understood by those skilled in the art, steps, means, and methods in the various operations, methods, and flows already discussed in the present invention may be replaced, modified, combined, or deleted. Furthermore, other steps, means, and methods having the same characteristics as those in the various operations, methods, and flows already discussed in the present invention may also be replaced, modified, rearranged, decomposed, combined, or deleted. Furthermore, steps, means, and methods in related art having the same characteristics as those in the various operations, methods, and flows disclosed in this application may also be replaced, modified, rearranged, decomposed, combined, or deleted.

[0091] In the present specification, terms such as "center," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," etc., indicate orientations or positional relationships based on those shown in the accompanying drawings, and are intended merely to facilitate and simplify the description of the present specification, and do not indicate or imply that the referenced devices or elements must be configured or operated in a particular orientation, and therefore should not be construed as limitations on the present specification.

[0092] The terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance, nor do they implicitly designate the number of technical features depicted. Thus, a feature defined by the term "first" or "second" may explicitly or implicitly include one or more of such features. In the present specification, unless otherwise specified, "plurality" means two or more.

[0093] In the specification of the present invention, unless otherwise expressly specified and limited, the terms "mounted," "connected," and "coupled" should be broadly understood as, for example, a fixed connection, a detachable connection, an integral connection, a direct connection, an indirect connection through an intermediate medium, or a connection within two elements. Those skilled in the art can understand the specific meaning of the above terms in the present invention based on specific cases.

[0094] In the description herein, the particular features, structures, materials, or characteristics may be combined in any suitable manner in any one or more embodiments or examples.

[0095] Although the individual steps in the flowcharts of the accompanying drawings are shown sequentially, as indicated by the arrows, it should be understood that the steps are not necessarily performed sequentially in the order indicated by the arrows. Unless expressly stated herein, there is no strict order restriction on the performance of these steps, and they may be performed in other orders. Furthermore, at least some of the steps in the flowcharts of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily performed to completion at the same time, but may be performed at different times, and the order of execution is not necessarily sequential, but may be performed in order or alternating with other steps or at least some of the sub-steps or stages of other steps.

[0096] It should be noted that the above are only some of the embodiments of the present invention, and those skilled in the art may make multiple improvements and modifications without departing from the principle of the present invention, which shall also be deemed to be within the protection scope of the present invention.

Claims

1. acquiring dynamic characteristics of the robot arm under a predetermined position and posture and a predetermined load, and constructing a data set in which the dynamic characteristics correspond to the predetermined position and posture and the predetermined load; training a deep neural network model based on the dataset; predicting target dynamic characteristics of the robot arm using the trained deep neural network model based on a target position, posture, and target load of the robot arm; designing a vibration suppressor based on the target dynamic characteristics, and controlling movement of the robot arm to a target position and posture while suppressing residual vibrations in cooperation with a motion controller; The step of acquiring dynamic characteristics of the robot arm under a predetermined position / posture and a predetermined load, and constructing a data set in which the dynamic characteristics correspond to the predetermined position / posture and the predetermined load, includes: obtaining dynamic parameters of the robot arm joints; constructing a robot arm dynamics model based on a structural design scheme of the robot arm and dynamics parameters of the robot arm joints; analyzing and acquiring dynamic characteristics of the robot arm under a predetermined position / posture and a predetermined load based on the robot arm dynamic model, and constructing a data set in which the dynamic characteristics correspond to the predetermined position / posture and the predetermined load; A method for suppressing vibration of a robot arm.

2. The step of obtaining dynamic parameters of the robot arm joints includes: obtaining an impulse response curve of the robot arm joint; and processing the impulse response curve to obtain equivalent stiffness, equivalent damping, and equivalent inertia information for each joint of the robot arm.

2. The method for suppressing vibration of a robot arm according to claim 1.

3. The step of constructing a robot arm dynamics model based on the dynamics parameters of the robot arm joints includes: constructing the robot arm dynamics model by combining the dynamics parameters of the robot arm joints with a three-dimensional model of the robot arm; 2. The method for suppressing vibration of a robot arm according to claim 1.

4. The step of analyzing and acquiring dynamic characteristics of the robot arm under a predetermined position / posture and a predetermined load based on the robot arm dynamic model, and constructing a data set in which the dynamic characteristics correspond to the predetermined position / posture and the predetermined load, includes: constructing an independent variable data set using a predetermined position and posture of the robot arm and a predetermined load as independent variables; using the robot arm dynamics model to acquire dynamic characteristics of the robot arm under conditions corresponding to combinations of each independent variable in the independent variable dataset, and constructing a response dataset; 2. The method for suppressing vibration of a robot arm according to claim 1.

5. training a deep neural network model based on the dataset, training neural networks with different numbers of layers to determine an optimal number of layers; training a neural network by different training set selection methods to obtain a mapping relationship between the dynamic characteristics and the data set corresponding to the predetermined position and posture and the predetermined load; 2. The method for suppressing vibration of a robot arm according to claim 1.

6. predicting a target dynamic characteristic of the robot arm using the trained deep neural network model based on a target position and posture of the robot arm and a target load, introducing the trained deep neural network model into a control device of the robot arm, and inputting information of a target position and posture and a target load of the robot arm into the trained deep neural network model to determine a target dynamic characteristic of the robot arm; 2. The method for suppressing vibration of a robot arm according to claim 1.

7. The robot arm dynamic model is a robot arm simulation dynamic model, and the step of analyzing and obtaining dynamic characteristics of the robot arm under a predetermined position posture and a predetermined load based on the robot arm dynamic model includes: Obtaining actual dynamic characteristics of the robot arm prototype at a predetermined position and under a predetermined load through experiments; obtaining a simulated dynamics characteristic of the robot arm in the same state as an experiment under a predetermined position, posture, and load based on the robot arm simulation dynamics model; and modifying the robot arm simulation dynamics model to determine the dynamics characteristics.

2. The method for suppressing vibration of a robot arm according to claim 1.

8. The step of experimentally obtaining actual dynamic characteristics of the robot arm prototype under a predetermined position and a predetermined load includes: determining, by experiment, natural frequencies, modal damping ratios, and vibration modes of the robot arm prototype at multiple positional attitudes and under multiple loads; 8. The method for suppressing vibration of a robot arm according to claim 7.

9. The step of modifying the robot arm simulation dynamics model to determine the dynamics characteristics includes: comparing the actual dynamic characteristics with the simulated dynamic characteristics, determining sensitive parameters by sensitivity analysis, and modifying the sensitive parameters to determine the dynamic characteristics; 8. The method for suppressing vibration of a robot arm according to claim 7.

10. An arm body; a controller connected to the arm body, for controlling movement of the arm body to a target position and posture, and for implementing the method for suppressing vibration of a robot arm according to any one of claims 1 to 9; 1. A robotic arm component comprising:

11. a processor; a memory communicatively connected to the processor; At least one program stored in the memory, which, when executed by the processor, implements the method for suppressing vibration of a robot arm according to any one of claims 1 to 9; A control device characterized in that:

12. A computer-readable storage medium storing a computer program, the computer program being executed by a control device to implement the method for suppressing vibration of a robot arm according to any one of claims 1 to 9. A computer-readable storage medium comprising:

Images