A method for predicting and compensating position errors of industrial robots throughout their life cycle

Abstract

The present invention proposes a method for predicting and compensating the position error of an industrial robot in the whole life cycle, including: generating n sample points in the robot's workspace as the robot's theoretical position P; measuring the actual position error E of the sample points by a laser tracker, and establishing a robot position error prediction model; inputting the theoretical position of the point to be compensated into the established error prediction model, the error prediction model is used to predict the robot position error of the point, and then predict the error εl caused by the rotation error of the 1 axis, and then predict the base flexibility error Δf caused by the base deformation; superimposing the predicted position errors εl and Δf on the theoretical position of the point to be measured to obtain the compensated target position P'; inputting the compensation position command of the compensated target position P' into the controller to complete the error compensation of the robot's processing trajectory. The present invention can continuously adjust the position error samples to fine-tune the accuracy of the position error prediction model.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial robots, and in particular to a method for predicting and compensating position errors of industrial robots throughout their life cycle. Background Art

[0002] In order to improve the processing accuracy of industrial robots, there are many researches on position compensation methods. There are model-based and non-model-based compensation methods, which are expected to improve the motion accuracy of robots. The most common DH parameter kinematic model, due to the singularity problem of robots, many people have made many optimizations based on this model, such as MDH model, CPC model, POE model, etc.

[0003] However, the factors that affect the robot's motion error are not limited to this. The connection of the robot's components, variable loads, and thermal deformation will all affect the motion accuracy of the robot's end. For such a "black box" system, some people have begun to study the mapping relationship between the robot's position error and its theoretical posture and joint angle. For example, the robot's estimated position error can be superimposed on the expected position instruction through spatial interpolation methods such as neural networks, fuzzy interpolation, and bilinear interpolation to achieve the robot's position accuracy compensation.

[0004] The prior art proposes a robot theoretical and actual position error mapping model based on a deep belief network to achieve position error estimation. However, it has the following disadvantages: as the robot is used for a longer period of time, its performance will degrade and the robot's position accuracy will deteriorate significantly. Since this mapping model cannot be changed once it is established, as the robot's working hours increase, under the influence of multiple factors such as wear and collision, the accuracy of the robot model will continue to decline, and the predicted position error will become increasingly inaccurate. Therefore, this model-based and non-model-based position compensation method is less effective. Therefore, this established position accuracy optimization model is not suitable for the entire life cycle of the robot, and requires continuous iterative search and optimization of the error prediction model.

[0005] In addition, the current position error prediction model and position error compensation method give little consideration to the error of the robot base. For robots whose bases need to be moved, such as composite robots, 7-axis robots mounted on slide rails, and other robots that require the robot base to move, the robot's coordinate system needs to be rebuilt after the base position changes, and a prediction model is established for the data after each movement. For robots that need to move, their position error includes the flexibility error of the base, so it is also important to compensate for the flexibility error of the robot's base to improve the robot's position accuracy.

[0006] At present, error compensation is a key technology for high-precision manufacturing robots. A large number of universities, research institutions, and companies have conducted extensive and in-depth research on this. The offline calibration method currently best meets the needs of engineering applications. Among them, the compensation method based on the kinematic model has the highest computational complexity. Since the robot control cabinet is generally not open, it is very difficult to modify the kinematic parameters in the robot's control cabinet with this method, resulting in poor performance of this precision compensation method. In addition, spatial interpolation is also a feasible method, but because it requires a large amount of measurement data, this offline calibration method is not very common. At present, most non-model-based error compensation methods require measuring multiple measurement points in the robot workspace. Because a large number of measurement points are required, the error prediction model established by this method is complex and inefficient. Summary of the invention

[0007] The object of the present invention is to solve at least one of the technical drawbacks.

[0008] To this end, the purpose of the present invention is to propose a method for predicting and compensating position errors of industrial robots throughout their life cycle.

[0009] In order to achieve the above object, an embodiment of the present invention provides a method for predicting and compensating position errors of an industrial robot throughout its life cycle, comprising the following steps:

[0010] Step S1, generating n sample points in the robot's workspace as the robot's theoretical position P;

[0011] Step S2, measuring the actual position error E of the sample point by a laser tracker, and establishing a robot position error prediction model;

[0012] Step S3: The points to be compensated The theoretical position of the point is input into the established error prediction model, and the error prediction model is used to predict the point The robot position error Then the error εl caused by the rotation error of the 1-axis is predicted, and then the base flexibility error Δf caused by the base deformation is predicted;

[0013] Step S4: The predicted position error εl and Δf are superimposed on the theoretical position of the point to be measured The target position P′ obtained by compensation is

[0014]

[0015] Step S5, inputting a compensation position command of the compensated target position P′ into the controller to complete the error compensation of the machining trajectory of the robot.

[0016] Further, the theoretical position P of the robot in step S1 and the joint torque Φ of the robot at the theoretical position are used as inputs of the error training model, and the actual position error E measured by the laser tracker in step S2 is used as output of the error training model, thereby establishing an error prediction model;

[0017] The theoretical position P of the robot includes: the theoretical position p of n sample points i =(x i ,y i , z i ) T ; The joint torque Φ of the robot at the theoretical position includes: the joint torque of n sample points The actual position error E measured by the laser tracker includes: the actual position error e of n sample points i =(e xi , e yi , e zi ) T ;The above i∈[o,n].

[0018] Further, in step S3, when measuring the sample point in step S2, the robot's axis 1 is rotated by a preset angle based on the target position P so that the laser tracker can measure normally; When the end position error ε1 caused by the rotation error of axis 1 is predicted,

[0019] The error ε1 caused by the rotation error of the predicted 1 axis includes:

[0020] ε1=F(θ 1′ +Δθ 1′ , θ2+Δθ2, θ3+Δθ3, θ4+Δθ4, θ5+Δθ5, θ6+Δθ6)

[0021] Among them, F is the positive kinematic function of the robot end obtained by the six joint angles, θ 1′ is the joint angle after axis 1 rotates a certain angle, Δθ 1′ is the joint angle error after axis 1 rotates a certain angle; θ i The theoretical position that needs to be compensated The corresponding joint angle, Δθ i is the corresponding joint angle error; where i=2, 3, 4, 5.

[0022] Further, in step S3, the terminal position error Δf caused by the predicted base deformation is x , Δf y , Δf z ) T ,include:

[0023] The end position error of the robot caused by base deformation in different postures is expressed as:

[0024]

[0025] Where Δf x is the end position error in the x direction, Δf y is the end position error in the y direction, Δf z is the end position error in the z direction; M and F are the external moment and external force on the base, respectively. i and M i are the components of F and M in the i direction, i = x, y, z; and For flexible base in F i and M i The flexibility coefficient in direction j under action, j = x, y, z.

[0026] Furthermore, the elastic coefficient parameters of the base in the j direction are identified using the robot's original position and then the rotation angle is compensated.

[0027] The full life cycle industrial robot position error prediction and compensation method of the embodiment of the present invention first performs position error compensation of the ISVM position error prediction model, then performs base flexible position error compensation, and finally performs joint motion error compensation. The ISVM position error prediction model can continuously add new position error samples and continuously fine-tune the accuracy of the model itself, which is suitable for the full working life cycle of the robot.

[0028] The method for predicting and compensating position errors of industrial robots throughout their life cycle according to an embodiment of the present invention has the following beneficial effects:

[0029] (1) The robot position error prediction model of the present invention can be applied to the entire life cycle of the robot, and the position error samples are continuously adjusted to fine-tune the accuracy of the position error prediction model.

[0030] (2) The robot position error compensation method of the present invention can compensate for the error caused by the flexibility of the robot base, and is suitable for mobile robots or large-load robots.

[0031] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0033] Figure 1is a flow chart of a method for predicting and compensating position errors of an industrial robot throughout its life cycle according to an embodiment of the present invention;

[0034] Figure 2 Schematic diagram of a method for predicting and compensating position errors of an industrial robot throughout its life cycle according to an embodiment of the present invention;

[0035] Figure 3a and Figure 3b is a schematic diagram of the robot end trajectory before position error compensation according to an embodiment of the present invention;

[0036] Figure 4a and Figure 4b A schematic diagram of a robot end trajectory after position error compensation using an ISVM position error prediction model according to an embodiment of the present invention;

[0037] Figure 5a and Figure 5b A schematic diagram of a motion trajectory of a robot end after compensating for a base flexibility error according to an embodiment of the present invention;

[0038] Figure 6a and Figure 6b A schematic diagram of the motion trajectory of the robot end after compensating for the 1-axis backlash error according to an embodiment of the present invention;

[0039] Figures 7a to 7c A comparison diagram of position errors in three directions, X, Y, and Z, of the robot end before and after compensation using the ISVM error prediction model according to an embodiment of the present invention;

[0040] Figures 8a to 8c A comparison diagram of position errors in three directions, X, Y, and Z, of the robot end before and after base flexibility compensation according to an embodiment of the present invention;

[0041] Figure 9a and Figure 9b 1 is a comparison diagram of position errors in the X and Y directions of the robot end before and after 1-axis motion error compensation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0042] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0043] The present invention provides a method for predicting and compensating the position error of an industrial robot throughout its life cycle based on an ISVM algorithm. The method establishes a position error prediction model based on the ISVM algorithm, and the error prediction model can be applied to the entire life cycle of the robot. The position error prediction model uses the actual position error of the robot as a sample, and continuously adds new sample data to fine-tune the accuracy of the position error prediction model. The present invention also incorporates the flexibility error of the robot base into the position error compensation. For large load conditions, the industrial robot itself and the complex gravity cannot be ignored. The movement of the center of mass caused by the rotation of the 1-axis will cause the base to deform flexibly. There is a correlation between the movement of the 1-axis and the flexible deformation of the base. Through the mapping relationship, this method can also compensate for this part of the position error.

[0044] In the following, in conjunction with the accompanying drawings, the position error prediction and compensation of the motion of one axis and the base flexibility at the end are taken as an example to illustrate the position error prediction and compensation method of the industrial robot throughout its life cycle of the present invention.

[0045] like Figure 1 As shown, the full life cycle industrial robot position error prediction and compensation method of the embodiment of the present invention includes the following steps:

[0046] Step S1, generate n sample points in the robot's workspace as the robot's theoretical position P.

[0047] Step S2, using a laser tracker to measure the actual position error E of the sample point and establish a robot position error prediction model.

[0048] In step S2, the theoretical position P of the robot in step S1 and the joint torque Φ of the robot at the theoretical position are used as inputs of the error training model, and the actual position error E measured by the laser tracker in step S2 is used as output of the error training model, thereby establishing an error prediction model;

[0049] The theoretical position P of the robot includes: the theoretical position p of n sample points i =(x i ,y i , z i ) T ; The joint torque Φ of the robot at the theoretical position includes: the joint torque of n sample points The actual position error E measured by the laser tracker includes: the actual position error e of n sample points i =(e xi , e yi , e zi ) T ;The above i∈[0,n].

[0050] Step S3: The points to be compensated The theoretical position of the error prediction model is input into the established error prediction model, which is used to predict the point The robot position error Then the error εl caused by the rotation error of axis 1 is predicted, and then the base flexibility error Δf caused by the base deformation is predicted.

[0051] The implementation principles of the ISVM algorithm and position error prediction mode are as follows:

[0052] For the sample set {(x,y)x∈R m ,y i = ±1, i = 1, 2, ... n}, the optimal problem when using SVM classification is

[0053]

[0054] Where ε is the slack variable and C is the regularization parameter. With the introduction of the Lagrange coefficient a, when dealing with nonlinear problems, the problem is expressed in logarithmic form using the Lagrange multiplier method, as shown below.

[0055]

[0056] Where Q i，j =y i y i K(x i ,x j ), and K is the kernel function. The Karush-Kuhn-Tucker (KKT) condition is a sufficient condition optimization problem to solve the above problem, which can be expressed as

[0057]

[0058] The KKT condition divides the training set D into three categories:

[0059] 1) Edge support vector set S, G i >0;

[0060] 2) Error support vector set E, G i <0;

[0061] 3) Remaining vector set R, G i >0.

[0062] As new samples are introduced during the incremental learning process, the margin vector coefficients change simultaneously to satisfy the KKT conditions of all previously trained samples. For a new sample m, it is considered a candidate support vector, and the condition can be expressed as:

[0063]

[0064] For the edge vector set S = {s1, s2...sl}, G i = 0, so that the change of coefficients {a, b} satisfies

[0065]

[0066] where Ω is a symmetric but non-positive definite Jacobian matrix

[0067]

[0068] By defining the sensitivity of the coefficients,

[0069]

[0070] Where ρ = Ω -1 , and β j = 0. For all j except S, we get an equation:

[0071]

[0072] Therefore, a changing condition is obtained according to the following formula

[0073]

[0074] Add a candidate vector m to the set of edge vectors S, and ρ is expanded to .

[0075]

[0076] Conversely, to remove vector k from S, we reduce ρ to

[0077]

[0078] Therefore, when a new sample m is added to the training dataset D: l+1 =D l ∪{m}, the solution update of parameters {a, b} and the candidate parameter x m ,y m , the current solution is related to the inverse matrix ρ of the Jacobian matrix.

[0079] For industrial robots in certain working scenarios, it is inconvenient to move the robot when it is installed at a specific workstation, and it is blocked by obstacles and the laser tracker cannot be used to measure the position error. Therefore, when measuring the sample point in step S2, it is necessary to rotate the robot's axis 1 by a certain angle based on the target position P so that the laser tracker can measure normally. When , it is necessary to predict the end position error ε1 caused by the rotation error of axis 1.

[0080] Predict the error ε1 caused by the rotation error of axis 1, including:

[0081] ε1=F(θ 1′ +Δθ 1′ , θ2+Δ1θ2, θ3+Δθ3, θ4+Δθ4, θ5+Δθ5, θ6+Δθ6)

[0082] Where F is the positive kinematic function of the robot end obtained by the six joint angles, θ 1′ is the joint angle after axis 1 rotates a certain angle, Δθ 1′ is the joint angle error after axis 1 rotates a certain angle; θ i The theoretical position that needs to be compensated The corresponding joint angle, Δθ i is the corresponding joint angle error; where i=2, 3, 4, 5.

[0083] It should be noted that "rotating by a certain angle" does not refer to a specific angle value, as long as the rotation angle of one axis can enable the laser tracker to complete the measurement normally.

[0084] Specifically, the end position error caused by the 1-axis motion error is as follows:

[0085] For a general six-degree-of-freedom industrial robot, the end position error is a function of the six joint angles. The robot end position error caused by the robot joint angle can be expressed as:

[0086] ε=P(θ1+Δθ1, θ2+Δθ2, θ3+Δθ3, θ4+Δθ4, θ5+Δθ5, θ6+Δθ6) (12)

[0087] Where P is the positive kinematic function of the robot end obtained by the six joint angles, θ i are the corresponding six joint angles, Δθ i is the corresponding joint angle error.

[0088] The end position error of the robot after rotating for several angles can be expressed as

[0089] ε1=P(θ 1′ +Δθ 1′ , θ2+Δθ2, θ3+Δθ3, θ4+Δθ4, θ5+Δθ5, θ6+Δθ6) (13)

[0090] In the formula, θ 1′ is the joint angle of the 1-axis after a certain rotation angle, Δθ 1′ is the corresponding joint angle error.

[0091] In step S3, the terminal position error Δf caused by the base deformation is predicted to be (Δf x, Δf y , Δf z ) T ,include:

[0092] The end position error of the robot caused by base deformation in different postures is expressed as:

[0093]

[0094] Where Δf x is the end position error in the x direction, Δf y is the end position error in the y direction, Δf z is the end position error in the z direction; M and F are the external moment and external force on the base, respectively. i and M i are the components of F and M in the i direction, i = x, y, z; and For flexible base in F i and M i The flexibility coefficient in direction j under action, j = x, y, z.

[0095] Use the robot's original position to identify the elastic coefficient parameters of the base in the j direction and then compensate for the rotation angle.

[0096] Specifically, the end position error caused by base flexibility is as follows:

[0097] The error before and after the robot rotates is mainly caused by the gravity of the robot at different positions. M and F are the external force and torque received by the base, respectively, expressed as:

[0098]

[0099] In the formula, m i is the mass of each link, r is the coordinate of the center of mass in the corresponding coordinate system, and p is the adjacent transformed coordinate system.

[0100] In the robot's base coordinate system, the change in the tool tip position error before and after the robot rotates is:

[0101]

[0102] Among them, (Δfx, Δfy, Δfz) are the errors before and after the rotation, θ1 is the angle of the standard robot trajectory axis 1, θ2 is the angle after the axis 1 is rotated by several angles, and m is the corresponding measurement point. The end error caused by the base deformation in different robot postures can be expressed as.

[0103]

[0104] Where Fi and M i are the components of F and M in the i direction, and λF ij and λM ij is the flexibility of the flexible base in direction j. The total mass of the robot before and after rotation remains unchanged, the error caused by gravity remains unchanged, λF ij It can be considered as 0. Finally, the elastic coefficient parameters of the base in the j direction are identified using the robot's original position and then the rotation angle is compensated.

[0105] Step S4: The predicted position error εl and Δf are superimposed on the theoretical position of the point to be measured The target position P′ obtained by compensation is

[0106]

[0107] Step S5, inputting a compensation position command of the compensated target position P′ into the controller to complete the error compensation of the machining trajectory of the robot.

[0108] The present invention uses KUKA KR160 as the test object of the scheme. The present invention will let KR160 move along a rectangular contour of 800×800 mm, and use a laser tracker to measure the robot end motion position error without position error compensation and with position error compensation.

[0109] Figure 3a and Figure 3b Schematic diagram of the robot end trajectory before position error compensation according to an embodiment of the present invention. Figure 4a and Figure 4b Schematic diagram of the robot end trajectory after position error compensation using the ISVM position error prediction model according to an embodiment of the present invention. Figure 5a and Figure 5b Schematic diagram of the motion trajectory of the robot end after compensating for the base flexibility error according to an embodiment of the present invention. Figure 6a and Figure 6b Schematic diagram of the motion trajectory of the robot end after compensating for the 1-axis backlash error according to an embodiment of the present invention.

[0110] Figures 7a to 7c The figure is a comparison diagram of the position errors in the three directions of X, Y, and Z of the robot end before and after compensation using the ISVM error prediction model according to an embodiment of the present invention. It can be seen from the figure that after the introduction of ISVM compensation, the position accuracy of the robot in the three directions is improved, and the position accuracy in the Z direction is significantly improved.

[0111] Figures 8a to 8cThe figure is a comparison diagram of the position errors in the three directions of X, Y, and Z of the robot end before and after the base flexibility compensation according to an embodiment of the present invention. It can be seen from the figure that after the ISVM position error prediction model is compensated, the base flexibility error compensation is adopted, and the overall position accuracy of the robot is further improved, and the position errors in the three directions can be converged within 0.5 mm.

[0112] Figure 9a and Figure 9b This is a comparison diagram of the position errors in the X and Y directions of the robot end before and after the 1-axis motion error compensation according to an embodiment of the present invention. It can be seen from the figure that after the ISVM position error prediction model compensation and the base flexibility error compensation are performed, the motion error compensation of the 1-axis is further introduced, and the motion position accuracy of the end of the machine can be further improved. The position accuracy in the X and Y directions converges to within 0.2mm.

[0113] It can be more intuitively seen from Table 1 that the end position accuracy of the robot is significantly improved after adopting the present invention.

[0114] Table 1

[0115]

[0116] According to the full life cycle industrial robot position error prediction and compensation method of the embodiment of the present invention, a position error prediction model is established based on the ISVM algorithm, and the error prediction model can be applied to the full life cycle of the robot. The position error prediction model will use the actual position error of the robot as a sample, and continuously add new sample data to fine-tune the accuracy of the position error prediction model. The present invention also incorporates the flexibility error of the robot base into the position error compensation. For large load conditions, the industrial robot itself and the complex gravity cannot be ignored. The movement of the center of mass caused by the rotation of the 1-axis will cause the base to deform flexibly. There is a correlation between the movement of the 1-axis and the flexible deformation of the base. Through the mapping relationship, this method can also compensate for this part of the position error.

[0117] The full life cycle industrial robot position error prediction and compensation method of the embodiment of the present invention first performs position error compensation of the ISVM position error prediction model, then performs base flexible position error compensation, and finally performs joint motion error compensation. The ISVM position error prediction model can continuously add new position error samples and continuously fine-tune the accuracy of the model itself, which is suitable for the full working life cycle of the robot.

[0118] The method for predicting and compensating position errors of industrial robots throughout their life cycle according to an embodiment of the present invention has the following beneficial effects:

[0119] (1) The robot position error prediction model of the present invention can be applied to the entire life cycle of the robot, and the position error samples are continuously adjusted to fine-tune the accuracy of the position error prediction model.

[0120] (2) The robot position error compensation method of the present invention can compensate for the error caused by the flexibility of the robot base, and is suitable for mobile robots or large-load robots.

[0121] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means 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 invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0122] Although the embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and cannot be understood as limiting the present invention. Those skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention without departing from the principles and purpose of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for predicting and compensating position errors of industrial robots throughout their life cycle, characterized in that: The steps include: Step S1, generating n sample points in the robot's workspace as the robot's theoretical position P; Step S2, measuring the actual position error E of the sample point by a laser tracker, and establishing a robot position error prediction model; Step S3: The points to be compensated The theoretical position of the point is input into the established error prediction model, and the error prediction model is used to predict the point The robot position error Then the error εl caused by the rotation error of the 1-axis is predicted, and then the base flexibility error Δf caused by the base deformation is predicted; Step S4: The predicted position error The error εl caused by the rotation error of the 1-axis and the base flexibility error Δf are superimposed on the theoretical position of the point to be measured. The target position P′ obtained by compensation is Step S5, inputting a compensation position command of the compensated target position P′ into the controller to complete the error compensation of the machining trajectory of the robot.

2. The method for predicting and compensating position errors of industrial robots throughout their life cycle as claimed in claim 1, characterized in that: In step S2, the theoretical position P of the robot in step S1 and the joint torque Φ of the robot at the theoretical position are used as inputs of the error training model, and the actual position error E measured by the laser tracker in step S2 is used as output of the error training model, thereby establishing an error prediction model; The theoretical position P of the robot includes: the theoretical position p of n sample points i =(x i ,y i , z i ) T ; The joint torque Φ of the robot at the theoretical position includes: the joint torque of n sample points The actual position error E measured by the laser tracker includes: the actual position error e of n sample points i =(e xi , e yi , e zi ) T ;The above i∈[0,n].

3. The method for predicting and compensating position errors of industrial robots throughout their life cycle as claimed in claim 1, characterized in that: In step S3, when measuring the sample point in step S2, the robot's axis 1 is rotated by a preset angle based on the target position P so that the laser tracker can measure normally; When the end position error ε1 caused by the rotation error of axis 1 is predicted, The error ε1 caused by the rotation error of the predicted 1 axis includes: ε1=F(θ 1′ +Δθ 1′ ,θ2+Δθ2,θ3+Δθ3,θ4+Δθ4,θ5+Δθ5,θ6+Δθ6) Among them, F is the positive kinematic function of the robot end obtained by the six joint angles, θ 1′ is the joint angle after axis 1 rotates a certain angle, Δθ 1′ is the joint angle error after axis 1 rotates a certain angle; θ i The theoretical position that needs to be compensated The corresponding joint angle, Δθ i is the corresponding joint angle error; where i=2, 3, 4, 5, 6.

4. The method for predicting and compensating position errors of industrial robots throughout their life cycle as claimed in claim 1, characterized in that: In step S3, the terminal position error Δf caused by the predicted base deformation is x , Δf y , Δf z ) T ,include: The end position error of the robot caused by base deformation in different postures is expressed as: Where Δf x is the end position error in the x direction, Δf y is the end position error in the y direction, Δf z is the end position error in the z direction; M and F are the external moment and external force on the base, respectively. i and M i are the components of F and M in the i direction, i = x, y, z; and For flexible base in F i and M i The flexibility coefficient in direction j under action, j = x, y, z.

5. The method for predicting and compensating position errors of industrial robots throughout their life cycle as claimed in claim 4, characterized in that: Use the robot's original position to identify the elastic coefficient parameters of the base in the j direction and then compensate for the rotation angle.

Images