Industrial robot joint calibration and semi-closed loop compensation control method

Abstract

The application discloses a kind of industrial robot joint calibration and semi-closed loop compensation control method, specifically relates to industrial robot precision control technical field, by establishing the full-process technical scheme of two-stage encoder selection, installation, calibration and semi-closed loop control based on absolute linear grating ruler, solve the problem of low absolute positioning accuracy of in-service industrial robot.The radial roundness runout measurement and control before installation effectively suppresses the detection error introduced by installation eccentricity;Through the joint calibration process of zero point and angle increment, the mapping relationship between grating reading and actual joint rotation angle is accurately established, which improves the calibration accuracy compared with traditional single calibration method.At the same time, by combining the discrete PID control law with the hierarchical coarse and fine correction strategy, the mechanical vibration is effectively suppressed while ensuring the fast response of joint error correction, which significantly improves the steady-state control accuracy of joint rotation angle.

Description

Technical Field

[0001] This invention belongs to the field of precision control technology for industrial robots, specifically relating to a joint calibration and semi-closed-loop compensation control method for industrial robots. Background Technology

[0002] Industrial robots are increasingly used in high-end fields such as aerospace manufacturing and precision machining, placing higher demands on their absolute positioning and trajectory tracking accuracy. For six-joint serial industrial robots, the error sources of end-effector pose can be divided into geometric errors and non-geometric errors. The former mainly includes kinematic model constant parameters and joint angle zero-position deviation, while the latter encompasses factors such as reducer gear backlash, joint and link flexibility deformation, and thermal effects. In the robot joint transmission chain, the servo motor drives the link to rotate via the reducer. The rotation angle measured by the first measurement system (motor encoder) is converted into the joint angle after transmission ratio conversion. However, nonlinear errors such as gear backlash in the reducer, link self-weight, and flexibility deformation caused by end-effector load lead to a deviation between this indirectly calculated value and the actual joint angle. Traditional semi-closed-loop control relying on a single motor encoder cannot effectively compensate for this deviation.

[0003] To address the aforementioned issues, existing technologies propose adding a second encoder at the joint to directly measure the actual joint rotation angle. However, in the retrofitting of in-service industrial robots, if a circular grating is used as the second encoder, the joint linkage needs to be disassembled, and the grating axis must be aligned with the joint rotation axis. This installation process is complex and can easily compromise the original transmission accuracy. While attaching a linear grating ruler to the joint's arc surface is simpler, the center of the grating ruler's contact surface is difficult to align with the joint's rotation axis, resulting in installation eccentricity errors. Furthermore, uneven adhesive thickness means the contact surface is not a standard arc surface, leading to roundness errors. Therefore, the linear grating ruler must be specially calibrated to establish a precise mapping between the grating reading and the joint rotation angle. In addition, a single feedback correction algorithm struggles to balance response speed and steady-state accuracy when dealing with complex multi-axis movements of the robot. Summary of the Invention

[0004] The purpose of this invention is to provide a method for joint calibration and semi-closed-loop compensation control of industrial robots to solve the above-mentioned problems.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for joint calibration and semi-closed-loop compensation control of an industrial robot, the specific steps of which are as follows: S1, for each target joint of the industrial robot, an absolute linear grating ruler is attached to the precision-machined outer surface of the joint shaft. S2, calibrate the linear grating ruler after it is attached and installed, determine the zero-position reading of the grating in the HOME position of the robot, and the correspondence between the grating reading increment and the actual joint rotation angle increment, so as to establish an accurate mapping model between the grating signal and the actual joint rotation angle, and write the calibration results into the control system variables; S3 uses the motor's built-in encoder as the first measurement system to provide feedback for the current loop and speed loop, and uses the absolute grating ruler calibrated in S20 as the second measurement system to provide feedback for the actual joint rotation angle of the position loop, thus forming a semi-closed-loop feedback control loop for the joint. S4. Based on the sampling period of the control system, the continuous PID control model is converted into a discrete PID control model by using the discretization method of replacing the derivative with the difference and replacing the integral with the summation, so as to adapt to the periodic feedback processing of the digital control system. S5. In the offline stage, Chebyshev polynomials are used to establish a spatial error estimation model to achieve first-level error prediction and joint mapping. During the online phase, the corrected joint command is input into the controller, and the actual rotation angle fed back by the grating is compared with the commanded rotation angle in real time. The deviation is processed by the PD controller to generate the joint correction amount.

[0006] Preferably, step S1 further includes measuring the radial runout value of the outer surface of each shaft with a dial indicator before installation, requiring the runout value to be controlled within 0.05mm to ensure the fitting accuracy and concentricity of the grating ruler.

[0007] Preferably, S2 specifically includes: (1) Zero position calibration: Control the robot joint to move to the zero position from both positive and negative directions at the same speed, record the grating readings in the positive and negative directions respectively, repeat the test multiple times and take the average value, and take the midpoint of the average value of the grating readings in the positive and negative directions as the grating zero position reference value of the joint. (2) Angle increment calibration: Rotate the target joint individually while keeping the other axes stationary. Record the current grating reading at fixed angle intervals within the grating angle calibration range. Simultaneously, use a laser tracker to measure the target ball position. Fit all target ball positions to the joint rotation circle. Obtain the actual joint rotation angle increment by measuring the angle between the adjacent target ball positions and the center of the circle. Combine this with the corresponding grating reading increment to calculate the grating reading corresponding to the unit angle. Unit angle grating reading = grating reading increment / actual rotation angle increment; generate the second measurement system pulse quantity for one rotation of the joint; the formula for calculating the second measurement system pulse quantity for one rotation of the joint is: Ji_ActPos( From a practical perspective ) = (TO_REAL(Ji_GratingPos( Current grating reading ))-Zero-position grating reading)*(unit angle grating reading).

[0008] Preferably, S3 specifically involves the relative encoder built into the motor providing current and speed feedback signals to the current controller and speed controller, and the absolute grating ruler installed at the joint directly feeding back the joint angle information to the position controller.

[0009] Preferably, the specific algorithm for converting the continuous PID control model into a discrete PID control model in step S4 is as follows: The position controller uses a PID control law, and the continuous PID control model is as follows: ; In the formula: For error signals, , , These are the proportional parameter, integral parameter, and derivative parameter; the values ​​of these three parameters are obtained through experimental tuning. Proportional parameter This determines the instantaneous response strength of the controller to the current error. Its output is proportional to the error signal. The initial value is set to 0.05, and then gradually increased until the response is rapid and without significant overshoot or oscillation. Integral parameter Used to eliminate static errors, its output is proportional to the accumulated error. However, the integral term reduces system stability and can easily cause overshoot, oscillation, or even "integral saturation." The initial value should be set to a minimum of ≤0.001, and gradually increased experimentally to observe whether oscillation occurs. Differential parameter It is used to predict the trend of error change, provides damping, and suppresses overshoot and oscillation. Its output is proportional to the rate of change of the error. First, set... ,Adjustment To the critical response, then gradually increase Observe whether the overshoot is suppressed until the response is stable and without oscillation; To accommodate the periodic sampling control of the CNC system, this continuous model needs to be discretized. The integral term is replaced by a summation, and the differential term is replaced by a first-order backward difference. ; The final discretized PID model is as follows: .

[0010] Preferably, S5 specifically involves: after the robot moves into position, the host computer acquires the joint grating feedback signal, calculates the actual joint rotation angle value through the grating calibration model; the difference between the theoretical joint value and the actual rotation angle value is processed by the PD controller to become the joint correction amount, and the correction amount is sent to the robot controller for execution via communication; The correction process is divided into coarse correction and fine correction: (1) In the coarse correction stage, the target angle is approached with a fixed step size; (2) When the deviation between the target angle and the actual angle of the joint is less than the set threshold, the system automatically switches to the fine correction stage and uses a PD controller for fine adjustment. After parameter tuning, the PD controller determines the proportional and derivative gain constants to ensure that a high level of control accuracy can be achieved in this stage.

[0011] The technical effects and advantages of this invention are as follows: By establishing a complete technical solution for the selection, installation, calibration, and semi-closed-loop control of a two-stage encoder based on an absolute linear grating ruler, the problem of low absolute positioning accuracy in in-service industrial robots is solved. Through radial runout measurement and control of the dial indicator before installation, the detection error introduced by installation eccentricity is effectively suppressed. Through a joint calibration process of zero point and angle increment, the mapping relationship between the grating reading and the actual joint angle is accurately established, improving calibration accuracy compared to traditional single calibration methods. Simultaneously, by combining a discretized PID control law with a hierarchical coarse-fine correction strategy, mechanical vibration is effectively suppressed while ensuring rapid response to joint error correction, significantly improving the steady-state control accuracy of the joint angle. This method is applicable to various configurations of six-joint serial industrial robots, and especially provides a non-disassembly, low-impact feasible solution for the precision upgrade and transformation of existing robots. By employing a layered and zoned control strategy, the system achieves "coarse correction for stability and speed, and fine correction for precision and accuracy." This allows for the use of a higher fine correction gain to obtain excellent steady-state accuracy without compromising system stability under large deviations. This mechanism is the core reason why the present invention can achieve joint control accuracy superior to existing technologies with high probability and stability, while effectively suppressing mechanical vibrations during the correction process. Attached Figure Description

[0012] Figure 1 This is a diagram of a robot joint transmission mechanism with a second encoder installed.

[0013] Figure 2 This is a block diagram of the joint semi-closed-loop feedback control model in this invention.

[0014] Figure 3 This is a block diagram of the joint closed-loop feedback control model for the hierarchical correction strategy in this invention.

[0015] Figure 4 This is a flowchart of the two-stage feedback accuracy compensation control strategy in this invention.

[0016] Figure 5 This is a comparison chart of the absolute error measured before and after the application of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] This invention provides, for example Figures 1-5 The diagram illustrates a joint calibration and semi-closed-loop compensation control method for an industrial robot, including S10 and the selection and installation accuracy control of the second encoder. For a six-jointed industrial robot, two types of absolute linear encoders were compared and analyzed. Circular encoders require disassembly and installation of the robot's joint links, and the grating axis must coincide with the joint's rotation axis, making installation difficult and compromising concentricity, significantly impacting the accuracy of the semi-closed-loop feedback. Linear encoders, on the other hand, only need to be installed on the machined outer surface of the joint axis, simplifying installation and ensuring accuracy, but requiring subsequent calibration to obtain angle values. Considering all factors, an absolute linear encoder was selected as the second encoder, installed on each robot axis using a surface-mount method. Before installation, the radial runout of each axis's outer surface was measured using a dial indicator, requiring the runout to be controlled within 0.05mm to ensure grating installation accuracy.

[0019] S20, Joint calibration of zero position and angle increment of the second encoder: The linear encoder is calibrated using a laser tracker, ensuring that the radial runout is within the required range, and that the encoder's mounting position corresponds to the surface of each joint axis of the robot. The calibration process involves the following steps: (1) Zero position calibration: Control the robot joint to move to the zero position from both positive and negative directions at the same speed, record the grating readings in the positive and negative directions respectively, repeat the test multiple times to calculate the average value, and take the midpoint of the average value of the grating readings in the positive and negative directions as the grating zero position reading of the joint.

[0020] (2) Angle Increment Calibration: Rotate the target joint individually while keeping the other axes stationary. Determine the calibration stroke within the effective stroke of the grating ruler and divide the measurement points at fixed angle intervals (e.g., 10°). With the first measurement system activated and the second measurement system deactivated, move the robot joint point by point within the calibration stroke, pausing at each angle interval, and measure the target ball position using a laser tracker. Fit all target ball positions to the joint rotation circle in the measurement software, calculate the angle between adjacent measurement points by connecting the target ball position to the center of the circle, obtain the actual joint rotation angle increment, and record the corresponding grating reading increment. Calculate the grating reading corresponding to a unit angle to obtain the pulse quantity (pls / r) of the second measurement system for one rotation of the joint. Write this pulse quantity into the CNC system variable to complete the calibration.

[0021] S30. Construct a three-loop semi-closed-loop feedback control model: like Figure 3 As shown, a three-loop feedback control model for the joint motor, comprising a current loop, a speed loop, and a position loop, is established. The current loop primarily regulates the motor torque. The motor's built-in relative encoder serves as the first encoder, providing feedback signals to the current controller and the speed controller. An absolute grating ruler calibrated via S20 serves as the second encoder, directly feeding back the actual joint rotation angle information to the position controller. This semi-closed-loop control method employs online real-time compensation, significantly improving joint rotation angle accuracy and robot relative trajectory accuracy.

[0022] S40, Discretized PID algorithm implementation for position controller: The position controller uses a PID control law, and the continuous PID control model is as follows: ; In the formula: For error signals, , , These are the proportional parameter, integral parameter, and derivative parameter, respectively.

[0023] To accommodate the periodic sampling control of the CNC system, this continuous model needs to be discretized. The integral term is replaced by a summation, and the differential term is replaced by a first-order backward difference. ; The final discretized PID model is as follows: ; S50, Joint closed-loop feedback fine-tuning execution based on hierarchical correction strategy; like Figure 4As shown, the specific process of joint closed-loop feedback correction is as follows: After the robot moves to its position, the host computer acquires the joint grating feedback signal, calculates the actual joint angle value through the grating calibration model, and uses the difference between the theoretical joint command value and the actual angle value as the error input. To avoid excessive joint correction within a single RSI signal processing cycle, which could lead to severe robot vibration, the correction process is divided into two stages: coarse correction and fine correction. In the coarse correction stage, the correction amount approaches the target angle with a fixed step size (0.002° / sampling period T); when the difference between the target joint angle and the actual angle is less than 0.007°, it automatically switches to the fine correction stage. In the fine correction stage, a PD controller is used, and the proportional and differential gain constants are determined after parameter tuning. The joint correction amount output by the PD controller is sent to the robot controller via communication, and the robot controller completes the joint error correction accordingly. Changes in joint position cause the grating feedback signal to be updated in real time, and the host computer continuously acquires the actual joint angle value, forming a closed-loop iteration until the joint angle error converges to the preset accuracy range, at which point one correction cycle ends. In summary, the core innovation of this invention lies in constructing a more efficient, less intrusive, two-layer decoupled control architecture, and embedding a hierarchical nonlinear correction strategy in the upper layer. The coarse-fine correction strategy partitions the error space and sets distinct control laws for different partitions, creating a nonlinear switching control system. This resolves the fundamental contradiction that a single PD controller cannot simultaneously achieve speed and stability.

[0024] like Figure 5 As shown, a certain type of six-joint industrial robot (the main body is an existing mass-produced model) was used as the test object. The end-effector absolute positioning error was measured under the original control (motor encoder only) and under the semi-closed-loop compensation scheme of the present invention. Experimental data shows that the average end-effector absolute error of the robot under the original control is about ±6.8mm. After adopting the scheme of the present invention, the average error is compressed to ±3.2mm. Compared with the traditional scheme, the present invention improves the absolute positioning accuracy by about 52.9%, and the error distribution is more concentrated without sudden spikes. This is due to the fact that the linear grating scale process proposed in the present invention effectively eliminates installation eccentricity and roundness errors. At the same time, the layered correction strategy suppresses high-frequency vibration while ensuring fast response, making the steady-state accuracy significantly better than the conventional single feedback compensation method. The comparison results fully verify the practicality and superiority of the present invention in the accuracy upgrade of in-service robots.

[0025] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for joint calibration and semi-closed-loop compensation control of an industrial robot, characterized in that: The specific steps are as follows: S1, for each target joint of the industrial robot, an absolute linear grating ruler is attached to the precision-machined outer surface of the joint shaft. S2, calibrate the linear grating ruler after it is attached and installed, determine the zero-position reading of the grating in the HOME position of the robot, and the correspondence between the grating reading increment and the actual joint rotation angle increment, so as to establish an accurate mapping model between the grating signal and the actual joint rotation angle, and write the calibration results into the control system variables; S2 specifically includes: (1) Zero position calibration: Control the robot joint to move to the zero position from both positive and negative directions at the same speed, record the grating readings in the positive and negative directions respectively, repeat the test multiple times and take the average value, and take the midpoint of the average value of the grating readings in the positive and negative directions as the grating zero position reference value of the joint. (2) Angle increment calibration: Rotate the target joint separately while keeping the other axes stationary. Record the current grating reading at each fixed angle interval within the grating angle calibration range. At the same time, use a laser tracker to measure the target ball position and fit all the target ball positions to the joint rotation circle. Obtain the actual joint rotation angle increment by measuring the angle between the adjacent target ball positions and the center of the circle. Combine the corresponding grating reading increment to calculate the grating reading corresponding to the unit angle and generate the second measurement system pulse quantity for one rotation of the joint. S3 uses the motor's built-in encoder as the first measurement system to provide feedback for the current loop and speed loop, and uses the absolute grating ruler calibrated in S2 as the second measurement system to provide feedback for the actual joint rotation angle of the position loop, thus forming a semi-closed-loop feedback control loop for the joint. S4. Based on the sampling period of the control system, the continuous PID control model is converted into a discrete PID control model by using the discretization method of replacing the derivative with the difference and replacing the integral with the summation, so as to adapt to the periodic feedback processing of the digital control system. S5. In the offline stage, Chebyshev polynomials are used to establish a spatial error estimation model to achieve first-level error prediction and joint mapping. During the online phase, the corrected joint command is input into the controller, and the actual rotation angle fed back by the grating is compared with the commanded rotation angle in real time. The deviation is processed by the PD controller to generate the joint correction amount. S5 specifically refers to: after the robot moves into position, the host computer obtains the joint grating feedback signal, and calculates the actual joint rotation angle value through the grating calibration model; the difference between the theoretical joint value and the actual rotation angle value is processed by the PD controller to become the joint correction amount, and the correction amount is sent to the robot controller for execution through communication. The correction process is divided into coarse correction and fine correction: (1) In the coarse correction stage, the target angle is approached with a fixed step size; (2) When the deviation between the target angle of the joint and the actual angle is less than the set threshold, the system automatically switches to the fine correction stage and uses the PD controller for fine adjustment. After parameter tuning, the PD controller determines the proportional and differential gain constants to ensure that the control accuracy can be achieved in this stage.

2. The method for joint calibration and semi-closed-loop compensation control of an industrial robot according to claim 1, characterized in that: S1 also includes measuring the radial runout value of the outer surface of each shaft with a dial indicator before installation, requiring the runout value to be controlled within 0.05mm to ensure the fitting accuracy and concentricity of the grating ruler.

3. The method for joint calibration and semi-closed-loop compensation control of an industrial robot according to claim 1, characterized in that: Specifically, S3 involves the relative encoder built into the motor providing current and speed feedback signals to the current controller and speed controller, and the absolute grating ruler installed at the joint directly feeding back the joint angle information to the position controller.

4. The method for joint calibration and semi-closed-loop compensation control of an industrial robot according to claim 1, characterized in that: The specific algorithm for converting the continuous PID control model into a discrete PID control model in S4 is as follows: The position controller uses a PID control law, and the continuous PID control model is as follows: ； In the formula: For error signals, , , These are proportional parameters, integral parameters, and differential parameters, respectively. To adapt to the periodic sampling control of the CNC system, the continuous model needs to be discretized; the integral term is replaced by a summation, and the differential term is replaced by a first-order backward difference. ； The final discretized PID model is as follows: ； In the formula At the nth sampling period, the joint correction value, i.e., the angle value, output by the controller is automatically calculated by the control system without manual setting; where The joint angle error at the nth sampling period is obtained by subtracting the actual joint angle measured by the grating in real time by the control system from the theoretical joint angle. In the formula This is the sampling period, a fixed period for the robot's RSI interface. It cannot be modified arbitrarily and must be set according to the period of the real-time control interface.

Images