High-precision end path control method for improving multi-axis synchronization performance of robot

Abstract

The invention relates to a high-precision tail end path control method for improving the multi-axis synchronization performance of a robot, and the method comprises the steps: building an incidence matrix of a Cartesian space path error and a joint space tracking error of each axis, and constructing a collaborative coupling error which is linearly combined by the joint space tracking error and a multi-axis synchronization error, according to the joint space tracking errors of all the shafts, balance scaling matrixes of all the shaft systems under different characteristics are established and used for adjusting control commands of all the shafts, the cooperative coupling errors are corrected, and the cooperative coupling errors containing the characteristics of all the shafts are obtained; three-ring cascade control is designed to track target torque, speed and position set by the robot; and a synchronous controller of an RBF neural network dynamic model is designed, so that a robot multi-axis cooperative system is established, and high-precision tail end path control of multi-axis synchronization performance is realized. According to the method, the robot path error problem is converted into the motion synchronization problem, multi-axis coordination control is achieved, and therefore the contour precision is improved.

Description

Technical Field

[0001] The present invention relates to the field of robot control technology, and in particular to a high-precision terminal path control method for improving multi-axis synchronization performance of a robot. Background Art

[0002] In the field of aerospace manufacturing and maintenance, high-precision robotic systems are widely used in key scenarios such as automatic wire laying of aircraft skin composite materials, precision machining of engine blades, high-precision assembly of spacecraft components, and extravehicular maintenance of space stations. These tasks place stringent demands on the robot's trajectory tracking accuracy, multi-axis collaborative stability, and adaptability to dynamic environments. Although robots exhibit the advantages of high flexibility and low cost in the processing process compared to traditional machine tools, the accuracy of open-chain serial robots is relatively low, which makes improving the robot's positioning accuracy the key to meeting the product's processing accuracy requirements. Traditional compensation methods mainly focus on the robot's positioning accuracy. However, due to the existence of control delays, the improvement of trajectory accuracy is often difficult to achieve the expected results.

[0003] Since the contour accuracy or shape accuracy of the part to be processed corresponds to the path accuracy during the robot processing process, precise control of the robot's motion path is a practical solution to meet the above application requirements. In existing robot control systems, the control loop of each actuator usually relies solely on its own feedback information. This control method leads to uncoordinated motion between the axes, resulting in errors in the robot's end-path. In order to improve the robot's path accuracy, synchronous control of each axis must be achieved. Through multi-axis collaborative control, not only can the coordinated movements of each joint be ensured, but the motion state of each axis can also be adjusted in real time in complex processing environments to adapt to dynamically changing processing conditions. This can improve the path accuracy of the robot's motion. Therefore, there is an urgent need for a high-precision end-path control method that improves the multi-axis synchronization performance of the robot.

[0004] Patent publication number CN113037150A discloses a method for multi-axis control of industrial robots based on a DSP+FPGA servo system. By calculating the position and velocity control PI loops, the set current is obtained, which is then used for current control, improving control performance and achieving faster current control response. This method primarily addresses the limitations of the PWM and ADC peripherals in the DSP. However, this control method is described solely from the perspective of cascaded control of electromechanical systems, without considering the coupling effects and implementation effects between multiple axes.

[0005] Patent publication number CN119937446A discloses a cross-coupled multi-axis synchronous interpolation control method based on PLCopen. This method uses an S-shaped acceleration and deceleration algorithm to interpolate the end effector and a cross-coupling control algorithm for real-time compensation of the interpolation process. This method can compensate for the contour errors generated by the synchronous motion of the axes of a multi-axis robot. However, this method requires calculating the contour errors of lines and arcs, and then using a PID controller to compensate for them. This process requires converting the robot's joint tracking error into a Cartesian space tracking error, and then further converting it into a Cartesian space contour error, which increases the calculation time.

[0006] Patent publication number CN112698623B discloses a multi-axis contour control method for multi-axis contouring applications. This method uses high-order splines to describe the contour, performs velocity planning in the time domain to satisfy path constraints, and then shapes the time-domain interpolated data planned for each axis, effectively improving corner bulging. This method improves interpolated contour accuracy during the motion planning phase and does not address the reduction of contour or path errors caused by robot errors during motion.

[0007] The above patents design multi-axis collaborative planning to achieve the robot's contour accuracy or path accuracy during the robot motion planning process, but do not take into account the impact of the robot's actual motion error on the robot's end accuracy, do not directly consider the correlation between path accuracy and tracking accuracy, and do not establish the overall architecture and compensation implementation plan of the robot system. Summary of the Invention

[0008] In response to the shortcomings of the existing technology, the present invention provides a high-precision end-path control method that improves the multi-axis synchronization performance of the robot, which solves the problem of the uncoordinated motion between the axes of the traditional robot, resulting in the end-path error of the robot and the relatively low control accuracy. In the application of robot processing, the shape accuracy of the product corresponds to the path accuracy of the robot motion. Therefore, a coupling error control method is provided with the purpose of path error control and multi-axis synchronous control as the means. By converting the robot path error problem into a motion synchronization problem, coordinated control of multiple axes is achieved, thereby improving the contour accuracy.

[0009] To solve the above technical problems, the present invention provides the following technical solution: a high-precision terminal path control method for improving the multi-axis synchronization performance of a robot, comprising the following steps:

[0010] S1, tracking error e through Cartesian space pose p The transformation matrix S between the Cartesian space path error δ and the robot Jacobian matrix J is used to establish the Cartesian space path error δ and the spatial tracking error e of each axis joint. q The correlation matrix δ=SJeq , through the correlation coefficient c i Achieve robot Cartesian space path error convergence;

[0011] S2, the relationship between the multi-axis is regulated by the multi-axis synchronization error ε of the robot, and the joint space tracking error e is constructed. q The cooperative coupling error E is a linear combination of the multi-axis synchronization error ε, and the degree of coupling is controlled by the coupling weight coefficient α;

[0012] S3, according to the spatial tracking error e of each axis joint q , establish the balanced scaling matrix β of each axis system under different characteristics b , used to adjust the control commands of each axis, correct the collaborative coupling error E, and obtain the collaborative coupling error E1 that includes the characteristics of each axis;

[0013] S4. Design a three-loop cascade control system for the robot’s electromechanical system to track the robot’s set target torque, velocity, and position;

[0014] S5. Design a synchronization controller τ based on the RBF neural network dynamic model to establish a robot multi-axis collaborative system and achieve high-precision end-path control of multi-axis synchronization performance.

[0015] Furthermore, in step S1, the specific process of obtaining the Cartesian space path error δ includes the following steps:

[0016] S11, acquiring tool position data from a CAD or CAM system, wherein the data is the tool position and orientation defined in a workpiece coordinate system;

[0017] S12, the robot motion path is expressed in parameterized form through motion planning and trajectory interpolation, and the interpolated posture of the robot at each moment is obtained as the desired posture p d , and obtain the actual pose p at each moment through the measuring instrument a The deviation between the two is the tracking error e of the robot in Cartesian space p ;

[0018] S13, actual pose p a Distance to desired pose p d The closest point error is used as the Cartesian space path error δ of the robot. The spatial curve profile is fitted with an osculating circle. In each sampling period, the osculating circle of the curve at the desired point is drawn. The Cartesian space path error δ of the robot consists of two vectors, δ = δ a +δ b , δ a is the perpendicular vector between the actual point and the tangent vector, that is, the distance between the actual point and the tangent line of the desired contour, δ bIt is the direction vector between the perpendicular point and the not plane, and the combination is:

[0019]

[0020] Among them, e p =[e x e y e z e α e β e γ ] T Represents the tracking error in Cartesian space, for the parameterized curve tangent vector, t is the tangent vector, n is the normal vector, and ρ is the curvature radius.

[0021] Furthermore, step S1 specifically includes:

[0022] S14. Let M = SJ. The goal is to make the Cartesian space path error δ zero, so δ = Me q =0 6×1 , thus constructing Among them, c i (i=1...6), i.e. c1c2c3c4c5c6 represents the correlation coefficient; e q (q=1...6), that is, e1e2e3e4e5e6 represents the spatial tracking error of each axis joint;

[0023] S15, when the spatial tracking error of each axis joint is e q The convergence of the robot's Cartesian space path error δ can be achieved when c1e1=c2e2=c3e3=c4e4=c5e5=c6e6 is satisfied.

[0024] Furthermore, in step S2, the specific process includes the following steps:

[0025] S21. In a closed-loop system, taking two adjacent axes as objects, a multi-axis synchronization error ε is constructed, which is composed of the tracking error of the single axis itself and the tracking errors of other axes. Then ε=[ε1ε2ε3ε4ε5ε6] T for:

[0026]

[0027] S22, finally converted into a T matrix to represent the synchronization transformation matrix, which is used to establish the spatial tracking error e of each axis joint q The relationship between it and the multi-axis synchronization error ε:

[0028] ε=Te q

[0029] S23, construct the spatial tracking error e of each axis joint qThe cooperative coupling error E combined with the multi-axis synchronization error ε is used to design an adaptive law to adjust the coupling coefficient to control the degree of coupling:

[0030] E=e q +αε=e q +αTe q

[0031] Where α represents the coupling coefficient. Increasing the weight will enhance the influence of synchronization, while decreasing the weight will reduce synchronization. It is used to balance the synchronization behavior of the system. When the multi-axis synchronization error ε is large, the weight is adaptively increased, and when the multi-axis synchronization error ε is small, the weight is adaptively decreased.

[0032] Furthermore, in step S3, the specific process includes the following steps:

[0033] S31, drive the robot joint space motion, after inverse kinematics to obtain the theoretical joint position q in the joint space d , for each axis, the actual joint position q on the connecting rod side is obtained through the grating secondary encoder under uniform speed operation a After that, the deviation between the two is the spatial tracking error of each axis joint e q , e q =q d -q a ;

[0034] S32, the spatial tracking error e of each axis joint q Normalize and get the offline normalized error vector, which is used as the balance scaling matrix β b =[β1 β2 β3 β4 β5 β6] T

[0035] in,

[0036]

[0037] Among them, e ti (i=1…6) represents the offline collected tracking error of each axis under uniform speed operation;

[0038] S33, by balancing the scaling matrix β b Correct the coordinated coupling error E to obtain the coordinated coupling error E1 including the characteristics of each axis:

[0039] Furthermore, in step S4, the robot electromechanical system includes a command layer, an upper control layer and a lower drive layer. The command layer includes joint space motion, Cartesian space motion and NC code parsing. The upper control layer includes motion control, feedforward control and feedback control. The motion control includes data registers, trajectory planners, forward and inverse kinematics. The robot's motion control mainly generates the robot's forward and inverse kinematics solution and generates the target position and speed of each axis. The feedforward control uses the RBF neural network to predict the robot's dynamic model parameters and calculate the feedforward torque. The feedback control is the coupling error feedback under multi-axis collaborative control. The lower-level control is a motor servo drive control module including PID control and three-loop cascade control, which controls the robot's electromechanical system in torque mode. Each link has independent PID control and is responsible for tracking the set target torque, speed and position.

[0040] Furthermore, in step S4, the three-loop cascade control specifically includes a torque loop, a speed loop and a position loop. The speed loop and the torque loop use feedback from the motor encoder system, and the position loop uses position feedback from the joint side grating encoder system. The torque loop, as the inner loop, responds the fastest and can quickly respond to instantaneous changes in the motor, while the position loop, as the outer loop, responds relatively slowly, and a filter is designed to eliminate data noise to ensure the stability and accuracy of the motion.

[0041] Furthermore, in step S5, the process includes the following steps:

[0042] S51, the robot upper control layer generates a compensation torque, which is directly superimposed on the set value of the torque loop, and combines the RBF neural network torque calculation method to design the torque feedforward control law τ of the robot f for:

[0043]

[0044] in, and They represent the predicted values ​​of the robot inertia term, Coriolis force, centripetal force term and gravity term by the neural network, respectively. d represents the desired joint positions of the robot, and denote the first and second order differentials of the desired joint position, respectively;

[0045] S52, based on the cooperative coupling error E1 including the characteristics of each axis, a torque feedback control τ is generated through a PI controller b , τ b =K P E1+K I ∫E1dt;

[0046] S53, design torque feedforward control τ based on RBF neural network dynamic modelf Online torque feedback control τ with cooperative coupling error E1 containing each axis characteristics b , and its synchronization controller is τ:

[0047] τ=τ f +K P E1+K I ∫E1dt+τ r

[0048] Among them, K P and K I is a positive proportional gain and integral gain matrix, τ r =K r sgn(e q ) represents the robust term, K r is the tuning coefficient matrix.

[0049] By means of the above technical solution, the present invention provides a high-precision terminal path control method for improving the multi-axis synchronization performance of a robot, which has at least the following beneficial effects:

[0050] (1) By establishing a correlation matrix between joint space tracking error and multi-axis synchronization error, the end-path error control problem is transformed into a multi-axis collaborative control problem. Combined with the dynamic weight control mechanism of position error and synchronization error in the coupled position error, the adaptive optimization of the motion relationship of each axis of the robot is achieved. Compared with the traditional single-axis independent control method, which requires solving the path error before controlling, this method significantly improves the processing accuracy in high-dynamic, multi-axis linkage scenarios;

[0051] (2) By constructing a coordinated error vector that includes a balanced scaling factor and normalized error correction, combined with offline normalized parameters of each axis motor's characteristics (such as inertia and response delay), the synchronization detuning problem caused by mechanical characteristic differences in traditional control is solved;

[0052] (3) The RBF neural network dynamics model predictive feedforward control is combined with three-loop cascade servo control. The upper control layer generates dynamic torque compensation instructions through inverse dynamics solution, and the lower layer uses position-speed-torque three-loop PID to achieve high-precision tracking. It has the advantages of both fast response and steady-state accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0054] Figure 1 is the relationship between the Cartesian space path error and the Cartesian space pose tracking error;

[0055] Figure 2 It is a three-loop cascade control diagram of the robot’s electromechanical system;

[0056] Figure 3 This is the composition diagram of the robot's electromechanical system architecture;

[0057] Figure 4 This is the design diagram of the synchronous controller. DETAILED DESCRIPTION

[0058] To make the above-mentioned objectives, features, and advantages of the present invention more clearly understood, the present invention is further described below in detail with reference to the accompanying drawings and specific embodiments. This will enable a full understanding of how this application uses technical means to solve technical problems and achieve technical effects, and to implement the invention accordingly.

[0059] Those skilled in the art will appreciate that all or part of the steps in the above-mentioned embodiment methods can be accomplished by instructing the relevant hardware through a program. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0060] Please refer to Figures 1-4 , shows a specific implementation of this embodiment. The technical problem to be solved by this embodiment is to address the shortcomings of the above-mentioned existing technologies. In robot processing applications, the shape accuracy of the product corresponds to the path accuracy of the robot movement. Therefore, a coupling error control method is provided with the purpose of path error control and multi-axis synchronous control as a means. First, the Cartesian space posture tracking error e is established. p The relationship between the Cartesian space path error δ and the robot Jacobian matrix is ​​used to establish the spatial tracking error e of each axis joint. q and the Cartesian space path error δ, then establish the spatial tracking error e of each axis joint q The collaborative coupling error E, composed of the multi-axis synchronization error ε, is used to regulate the coupling factor and adjust the weight between the two. Furthermore, a balance scaling factor is designed based on the independent characteristics of the motors. Based on the three-loop cascade control of the motors, a control law based on RBF feedforward and coupling error feedback is designed in the torque loop. This allows for correction of the robot path and improves the coordination of multi-axis motion by providing guidance on the robot's path error. Therefore, by transforming the robot path error problem into a motion synchronization problem, the present invention achieves coordinated control of multiple axes, thereby improving contour accuracy.

[0061] Please refer to Figure 1This embodiment proposes a high-precision end-path control method for improving the multi-axis synchronization performance of a robot, the method comprising the following steps:

[0062] S1. The path accuracy of the robot's motion determines the shape accuracy of the product to be processed. The Cartesian space posture tracking error e p The transformation matrix S between the Cartesian space path error δ and the robot Jacobian matrix J is used to establish the Cartesian space path error δ and the spatial tracking error e of each axis joint. q The correlation matrix δ=SJe q , through the correlation coefficient c i Achieve robot Cartesian space path error convergence; such as Figure 1 The figure shows the relationship between the desired position, actual pose and path error pose, and the Cartesian space pose tracking error e is obtained. p The error δ with the Cartesian space path;

[0063] As a preferred implementation of step S1, in step S1, the specific process of obtaining the Cartesian space path error δ includes the following steps:

[0064] S11, acquiring tool position data from a CAD or CAM system, wherein the data is the tool position and orientation defined in a workpiece coordinate system;

[0065] S12, the robot motion path is expressed in parameterized form through motion planning and trajectory interpolation, and the interpolated posture of the robot at each moment is obtained as the desired posture p d , and obtain the actual pose p at each moment through the measuring instrument a The deviation between the two is the tracking error e of the robot in Cartesian space p ;

[0066] S13, actual pose p a Distance to desired pose p d The closest point error is used as the Cartesian space path error δ of the robot. The spatial curve profile is fitted with an osculating circle. In each sampling period, the osculating circle of the curve at the desired point is drawn. The Cartesian space path error δ of the robot consists of two vectors, δ = δ a +δ b , δ a is the perpendicular vector between the actual point and the tangent vector, that is, the distance between the actual point and the tangent line of the desired contour, δ b It is the direction vector between the perpendicular point and the not plane, and the combination is:

[0067]

[0068] Among them, e p =[ex e y e z e α e β e γ ] T Represents the tracking error in Cartesian space. For the parameterized curve tangent vector, t is the tangent vector, n is the normal vector, and ρ is the curvature radius, which can be solved by the following formula:

[0069]

[0070]

[0071] Where l represents the arc length of the curve;

[0072] More specifically, step S1 further includes:

[0073] The robot's Cartesian space pose tracking error e p Converted into the spatial tracking error e of each axis joint q , thereby establishing the Cartesian space path error δ and the spatial tracking error e of each axis joint q The correlation matrix δ=SJe q ;

[0074] The spatial tracking error e of each axis joint can be obtained through the joint grating secondary encoder feedback system. q =[e1e2e3e4e5e6] T Therefore, the robot Cartesian space path error δ and the axis joint space tracking error e q Establish the following relationship:

[0075]

[0076] Where J is the robot Jacobian matrix and S is the Cartesian space pose tracking error e p The transformation matrix between the Cartesian space path error δ, t x , t y and t z are the projections of the tangent vector t in the x, y and z directions under the base coordinates, and we have:

[0077]

[0078] Among them, e t =e x t x +e y t y +e z t z , and are the first and second derivatives of the axis angle with respect to arc length, m x 、m y 、m z 、e t 、m α 、m β 、m γ Each represents a symbolic function;

[0079] Let M = SJ, the goal is to make the Cartesian space path error δ zero, so δ = Me q =0 6×1 ,

[0080]

[0081] From c1e1=c2e2=c3e3=c4e4=c5e5=c6e6=k, we can get e i =k / c i , put it into the above formula, and eliminate k, and we can get

[0082]

[0083] Solve the system of equations by Cramer's rule to construct Among them, c i (i=1…6), that is, c1c2c3c4c5c6 represents the correlation coefficient. When the tracking error of each axis joint is e q The convergence of the robot's Cartesian space path error δ can be achieved when c1e1=c2e2=c3e3=c4e4=c5e5=c6e6 is satisfied.

[0084] S2, the relationship between the multi-axis is regulated by the multi-axis synchronization error ε of the robot, and the joint space tracking error e is constructed. q The cooperative coupling error E is a linear combination of the multi-axis synchronization error ε, and the degree of coupling is controlled by the coupling weight coefficient α;

[0085] As a preferred implementation of step S2, in step S2, the specific process includes the following steps:

[0086] S21. In a closed-loop system, taking two adjacent axes as objects, a multi-axis synchronization error ε is constructed, which is composed of the tracking error of the single axis itself and the tracking errors of other axes. Then ε=[ε1ε2ε3ε4ε5ε6] T for:

[0087]

[0088] S22, finally converted into a T matrix to represent the synchronization transformation matrix, the synchronization transformation matrix T is used to establish the spatial tracking error e of each axis joint. q The relationship between it and the multi-axis synchronization error ε:

[0089] ε=Te q

[0090] S23, construct the spatial tracking error e of each axis joint q The cooperative coupling error E combined with the multi-axis synchronization error ε is used to design an adaptive law to adjust the coupling coefficient α to control the degree of coupling:

[0091] E=e q +αε=e q +αTe q

[0092] Where α represents the coupling coefficient. Increasing the weight will enhance the influence of synchronization, while decreasing the weight will reduce synchronization. It is used to balance the synchronization behavior of the system. When the multi-axis synchronization error ε is large, the weight is adaptively increased, and when the multi-axis synchronization error ε is small, the weight is adaptively decreased.

[0093] In this embodiment, for a multi-axis robot system, simply maintaining the errors of each axis consistent cannot guarantee path accuracy. Instead, a certain synergistic relationship between the errors of each axis is required to effectively improve the robot's path accuracy. The collaborative control method provides an effective solution for maintaining the interrelationships between multiple axes. This method transforms the problem of eliminating path errors at the robot's end point into a problem of synchronous control between the motion axes. The synchronization error vector is designed, and by regulating the synchronization errors of each axis, a certain relationship between the motions of the robot's axes is maintained.

[0094] S3, taking into account the inconsistency of the motion characteristics of each motor axis, according to the spatial tracking error e of each axis joint q , establish the balanced scaling matrix β of each axis system under different characteristics b , used to adjust the control commands of each axis so that the responses of each axis to the control commands are as consistent as possible, thereby correcting the collaborative coupling error E and obtaining the collaborative coupling error E1 that includes the characteristics of each axis;

[0095] As a preferred implementation of step S3, in step S3, the specific process includes the following steps:

[0096] S31, Attachment Figure 4 Medium x d Represents the desired pose in Cartesian space, drives the robot to move in joint space, and obtains the theoretical joint position q in joint space after inverse kinematics d , for each axis, the actual joint position q on the connecting rod side is obtained through the grating secondary encoder under uniform speed operation a After that, the deviation between the two is the spatial tracking error of each axis joint e q , e q =q d -q a ;

[0097] S32, the spatial tracking error e of each axis joint q Normalize and get the offline normalized error vector, which is used as the balance scaling matrix β b =[β1β2β3β4β5β6] T

[0098] in,

[0099]

[0100] Among them, e ti (i=1…6) represents the offline collected tracking error of each axis under uniform speed operation;

[0101] S33, normalized balanced scaling matrix β acquired offline b Correct the coordinated coupling error E to obtain the coordinated coupling error E1 including the characteristics of each axis:

[0102] S4. Design a three-loop cascade control system for the robot’s electromechanical system to track the robot’s set target torque, velocity, and position;

[0103] As a preferred embodiment of step S4, in step S4, Figure 3 As shown, the robot electromechanical system includes a command layer, an upper control layer and a lower drive layer. The command layer includes joint space motion, Cartesian space motion and NC code parsing. The upper control layer includes motion control, feedforward control and feedback control. Motion control includes data registers, trajectory planners, forward and inverse kinematics. The robot's motion control mainly generates the robot's forward and inverse kinematics solution and generates the target position and speed of each axis. Feedforward control uses RBF neural network to predict the robot's dynamic model parameters and calculate the feedforward torque. Feedback control is coupling error feedback under multi-axis collaborative control. The lower-level control is a motor servo drive control module including PID control and three-loop cascade control. It controls the robot's electromechanical system in torque mode. Each link has independent PID control and is responsible for tracking the set target torque, speed and position.

[0104] More specifically, in step S4, a three-loop cascade control of the robot's electromechanical system is designed, a robot controller is designed based on a PLC platform, and the six-axis robot is decomposed into six separate motors. The actual robot system is the individual control of the motors through motion control operations.

[0105] like Figure 2As shown in the figure, the three-loop cascade control system specifically includes a torque loop, a velocity loop, and a position loop. The velocity and torque loops utilize feedback from the motor encoder system, while the position loop utilizes position feedback from the joint-side optical encoder system to ensure motor response speed and joint position accuracy, eliminating issues such as gear backlash and hysteresis. The position loop ensures that the robot joints precisely rotate to the target position. Through real-time feedback, the position loop promptly adjusts the motor's movement to achieve high-precision positioning. The velocity loop regulates the motor's speed, responding quickly to commands from the position loop to ensure smooth and stable speed changes during motion. The torque loop controls the motor's output torque, ensuring stable output under load changes and preventing overload or loss of control. The three loops have varying response speeds. The torque loop, as the inner loop, responds the fastest, quickly responding to transient motor changes. The position loop, as the outer loop, responds relatively slowly. Filters are designed to eliminate data noise, gear backlash, and hysteresis to ensure stable and accurate motion.

[0106] S5, such as Figure 4 The controller shown in the figure combines the RBF neural network dynamics feedforward with the coupled collaborative error feedback torque and robust terms to design a synchronization controller τ based on the RBF neural network dynamics model, thereby establishing a robot multi-axis collaborative system and achieving high-precision end-path control with multi-axis synchronization performance.

[0107] As a preferred implementation of step S5, in step S5, the process includes the following steps:

[0108] S51, the robot upper control layer generates a compensation torque, which is directly superimposed on the set value of the torque loop, and combines the RBF neural network torque calculation method to design the torque feedforward control law τ of the robot f for:

[0109]

[0110] in, and They represent the predicted values ​​of the robot inertia term, Coriolis force, centripetal force term and gravity term by the neural network, respectively. d represents the desired joint positions of the robot, and denote the first and second order differentials of the desired joint position, respectively;

[0111] S52, based on the cooperative coupling error E1 including the characteristics of each axis, a torque feedback control τ is generated through a PI controller b , τ b =K P E1+K I ∫E1dt;

[0112] S53, design torque feedforward control τ based on RBF neural network dynamic model f Online torque feedback control τ with cooperative coupling error E1 containing each axis characteristics b , and its synchronization controller is τ:

[0113] τ=τ f +K P E1+K I ∫E1dt+τ r

[0114] Among them, K P and K I is a positive proportional gain and integral gain matrix, τ r =K r sgn(e q ) represents the robust term, K r is the tuning coefficient matrix.

[0115] In this embodiment, the motion control of the robot mainly generates the forward and inverse kinematics solutions of the robot to generate the target position and speed of each axis. Combining the inverse dynamics solution with the torque command required by the motor, the control torque required to correct the contour error through the synchronization error is sent to the robot to form the upper structure of the control system. The RBF neural network is used to predict the parameters of the robot dynamic model and calculate the feedforward torque. Based on the online measurement of the grating encoder, the feedback torque based on the coupling error is designed to form the torque feedforward control of the RBF neural network dynamic model and the online feedback controller of the coupling coordination error. The feedforward and feedback torques, as well as the designed robust terms, are merged and sent to the lower-level control. The motor servo control system composed of three-loop cascade control is responsible for tracking the set target value.

[0116] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.

[0117] The logic and / or steps represented in the flowchart or otherwise described herein may be considered, for example, as an ordered list of executable instructions for implementing logical functions, and may be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device).

[0118] The above embodiments provide a detailed introduction to the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A high-precision end-path control method for improving the multi-axis synchronization performance of a robot, characterized in that: The following steps are involved: S1, tracking error e through Cartesian space pose p The transformation matrix S between the Cartesian space path error δ and the robot Jacobian matrix J is used to establish the Cartesian space path error δ and the spatial tracking error e of each axis joint. q The correlation matrix δ=SJe q , through the correlation coefficient c i Achieve robot Cartesian space path error convergence; S2, the relationship between the multi-axis is regulated by the multi-axis synchronization error ε of the robot, and the joint space tracking error e is constructed. q The cooperative coupling error E is a linear combination of the multi-axis synchronization error ε, and the degree of coupling is controlled by the coupling weight coefficient α; S3, according to the spatial tracking error e of each axis joint q , establish the balanced scaling matrix β of each axis system under different characteristics b , used to adjust the control commands of each axis, correct the collaborative coupling error E, and obtain the collaborative coupling error E1 that includes the characteristics of each axis; S4. Design a three-loop cascade control system for the robot’s electromechanical system to track the robot’s target torque, velocity, and position. S5. Design a synchronization controller τ based on the RBF neural network dynamic model to establish a robot multi-axis collaborative system and achieve high-precision end-path control of multi-axis synchronization performance.

2. A high-precision end-path control method for improving multi-axis synchronization performance of a robot according to claim 1, characterized in that: In step S1, the specific process of obtaining the Cartesian space path error δ includes the following steps: S11, acquiring tool position data from a CAD or CAM system, wherein the data is the tool position and orientation defined in a workpiece coordinate system; S12, the robot motion path is expressed in parameterized form through motion planning and trajectory interpolation, and the interpolated posture of the robot at each moment is obtained as the desired posture p d , and obtain the actual pose p at each moment through the measuring instrument a The deviation between the two is the tracking error e of the robot in Cartesian space p ; S13, actual pose p a Distance to desired pose p d The closest point error is used as the Cartesian space path error δ of the robot. The spatial curve profile is fitted with an osculating circle. In each sampling period, the osculating circle of the curve at the desired point is drawn. The Cartesian space path error δ of the robot consists of two vectors, δ = δ a +δ b , δ a is the perpendicular vector between the actual point and the tangent vector, that is, the distance between the actual point and the tangent line of the desired contour, δ b It is the direction vector between the perpendicular point and the not plane, and the combination is: Among them, e p =[e x e y e z e α e β e γ ] T Represents the tracking error in Cartesian space, for the parameterized curve tangent vector, t is the tangent vector, n is the normal vector, and ρ is the curvature radius.

3. The high-precision terminal path control method for improving multi-axis synchronization performance of a robot according to claim 1, characterized in that: Step S1 specifically also includes: S14. Let M = SJ. The goal is to make the Cartesian space path error δ zero, so δ = Me q =0 6×1 , thus constructing Among them, c i (i=1…6), that is, c1 c2 c3 c4 c5 c6 represent the correlation coefficient; e q (q=1…6), that is, e1 e2 e3 e4 e5 e6 represent the spatial tracking errors of each axis joint; S15, when the spatial tracking error of each axis joint is e q The convergence of the robot's Cartesian space path error δ can be achieved when c1e1=c2e2=c3e3=c4e4=c5e5=c6e6 is satisfied.

4. The high-precision terminal path control method for improving multi-axis synchronization performance of a robot according to claim 3, characterized in that: In step S2, the specific process includes the following steps: S21. In a closed-loop system, taking two adjacent axes as objects, a multi-axis synchronization error ε is constructed, which is composed of the tracking error of the single axis itself and the tracking errors of other axes. Then ε=[ε1 ε2 ε3 ε4 ε5 ε6] T for: S22, finally converted into a T matrix to represent the synchronization transformation matrix, which is used to establish the spatial tracking error e of each axis joint q The relationship between it and the multi-axis synchronization error ε: ε=Te q S23, construct the spatial tracking error e of each axis joint q The cooperative coupling error E combined with the multi-axis synchronization error ε is used to design an adaptive law to adjust the coupling coefficient to control the degree of coupling: And=and q +αε=e q +αTe q Where α represents the coupling coefficient. Increasing the weight will enhance the influence of synchronization, while decreasing the weight will reduce synchronization. It is used to balance the synchronization behavior of the system. When the multi-axis synchronization error ε is large, the weight is adaptively increased, and when the multi-axis synchronization error ε is small, the weight is adaptively decreased.

5. The high-precision terminal path control method for improving multi-axis synchronization performance of a robot according to claim 1, characterized in that: In step S3, the specific process includes the following steps: S31, drive the robot joint space motion, after inverse kinematics to obtain the theoretical joint position q in the joint space d , for each axis, the actual joint position q on the connecting rod side is obtained through the grating secondary encoder under uniform speed operation a After that, the deviation between the two is the spatial tracking error of each axis joint e q , e q =q d -q a ; S32, the spatial tracking error e of each axis joint q Normalize and get the offline normalized error vector, which is used as the balance scaling matrix β b =[β1 β2 β3 β4 β5 β6] T in, Among them, e ti (i=1…6) represents the offline collected tracking error of each axis under uniform speed operation; S33, by balancing the scaling matrix β b Correct the coordinated coupling error E to obtain the coordinated coupling error E1 including the characteristics of each axis:

6. The high-precision terminal path control method for improving multi-axis synchronization performance of a robot according to claim 1, characterized in that: In step S4, the robot electromechanical system includes a command layer, an upper control layer and a lower drive layer. The command layer includes joint space motion, Cartesian space motion and NC code parsing. The upper control layer includes motion control, feedforward control and feedback control. The motion control includes data registers, trajectory planners, forward and inverse kinematics. The robot's motion control mainly generates the robot's forward and inverse kinematics solution and generates the target position and speed of each axis. The feedforward control uses the RBF neural network to predict the robot's dynamic model parameters and calculate the feedforward torque. The feedback control is the coupling error feedback under multi-axis collaborative control. The lower-level control is a motor servo drive control module including PID control and three-loop cascade control. It controls the robot's electromechanical system in torque mode. Each link has independent PID control and is responsible for tracking the set target torque, speed and position.

7. The high-precision terminal path control method for improving multi-axis synchronization performance of a robot according to claim 1, characterized in that: In step S4, the three-loop cascade control specifically includes a torque loop, a speed loop, and a position loop. The speed loop and the torque loop use feedback from the motor encoder system, and the position loop uses position feedback from the joint-side grating encoder system. The torque loop, as the inner loop, responds the fastest and can quickly respond to instantaneous changes in the motor, while the position loop, as the outer loop, responds relatively slowly. A filter is designed to eliminate data noise to ensure the stability and accuracy of the motion.

8. The high-precision terminal path control method for improving multi-axis synchronization performance of a robot according to claim 1, characterized in that: In step S5, the process includes the following steps: S51, the robot upper control layer generates a compensation torque, which is directly superimposed on the set value of the torque loop, and combines the RBF neural network torque calculation method to design the torque feedforward control law τ of the robot f for: in, and They represent the predicted values ​​of the robot inertia term, Coriolis force, centripetal force term and gravity term by the neural network, respectively. d represents the desired joint positions of the robot, and denote the first and second order differentials of the desired joint position, respectively; S52, based on the cooperative coupling error E1 including the characteristics of each axis, a torque feedback control τ is generated through a PI controller b , τ b =K P E1+K I ∫E1dt; S53, design torque feedforward control τ based on RBF neural network dynamic model f Online torque feedback control τ with cooperative coupling error E1 containing each axis characteristics b , and its synchronization controller is τ: τ=τ f +K P E1+K I ∫E1dt+τ r Among them, K P and K I is a positive proportional gain and integral gain matrix, τ r =K r sgn(e q ) represents the robust term, K r is the tuning coefficient matrix.

Images