CN113500585B - Robot measurement pose evaluation method and evaluation device for kinematics calibration - Google Patents

Abstract

The application discloses a robot measurement pose evaluation method and an evaluation device for kinematic calibration, wherein the method is used for evaluating a set of given measurement poses and task space in which the robot needs to ensure precision, the estimated square value of the residual root mean square of poses in the task space subjected to the kinematic calibration based on the set of measurement poses is expected to serve as a standard for evaluating the set of measurement poses, and corresponding weighted evaluation indexes are given by considering different noise intensities of measurement components and different weights of pose precision. The problem that the conventional measuring pose evaluation method focuses on geometric error observability and cannot directly reflect the kinematic calibration effect is solved, the measuring pose evaluation and the kinematic calibration can be more accurately and intuitively connected, and the amplitude of the residual error after the kinematic calibration in the robot task space is reflected.

Description

Robot measurement pose evaluation method and evaluation device for kinematics calibration

technical field

The present application relates to the technical field of robot kinematics calibration, in particular to a robot measurement pose evaluation method and evaluation device for kinematics calibration.

Background technique

Due to the geometric error of the robot due to factors such as manufacturing and assembly, the positioning accuracy of the robot will be reduced, which will limit the industrial application of the robot. Therefore, it is necessary to calibrate the kinematics of the robot before leaving the factory. Usually, the kinematic calibration method is to establish a geometric error model, measure the poses of multiple groups of robot end effectors, use the theoretical and actual pose deviations to identify geometric errors, and then correct the robot kinematics model to improve the end positioning of the robot. Attitude accuracy.

For kinematics calibration, in order to ensure the positioning accuracy after calibration in the overall task space, and at the same time take into account the measurement cost caused by the high number of measurement poses, it is necessary to optimize the measurement pose. The premise of optimizing the measurement pose is to perform quantitative index evaluation on a certain set of measurement poses.

At present, the research on evaluation methods mainly focuses on the analysis of the error identification matrix, which is the analysis of the observability of geometric errors, including the parameter variance minimization index Condition number reciprocal index/> Terminal pose uncertainty minimization index O ₃ =σ _L , noise method index/> And the optimal index of A/>

In the prior art mentioned above, the current evaluation method conducts a comprehensive analysis of the error identification equation in all aspects of statistics, but its limitation is that it only analyzes the observability of the geometric error. The pose effect of the actuator is not consistent, so the observation of the geometric error and the residual error after kinematic calibration are not strictly consistent. Note that the purpose of kinematics calibration is to improve the positioning accuracy of the overall task space, and the observability of geometric errors is an indirect reflection. Therefore, an index that directly reflects the positioning accuracy of the overall task space after kinematics calibration can more intuitively reflect the measurement position. The influence of posture evaluation on kinematic calibration.

Contents of the invention

This application aims to solve one of the technical problems in the related art at least to a certain extent.

For this reason, one purpose of this application is to propose a robot measurement pose evaluation method for kinematics calibration, which solves the problem that the current measurement pose evaluation method focuses on the observability of geometric errors and cannot directly reflect the effect of kinematics calibration .

Another object of the present application is to propose a robot measurement pose evaluation device for kinematics calibration.

In order to achieve the above purpose, an embodiment of the present application proposes a robot measurement pose evaluation method for kinematic calibration, including:

Establish the geometric error model of the robot;

Determining and discretizing the task space of the robot, determining the measurement noise intensity weight and pose accuracy weight of the robot, and for each group of poses in the discretized task space of the robot, according to the position The attitude accuracy weight determines the corresponding error transfer matrix set and preprocesses it;

Determine the error identification equation according to the pose to be evaluated and the geometric error model, and determine the root mean square square of the kinematically calibrated pose residual in the task space of the robot after discretization according to the measurement noise intensity weight value Value expected mean value, using the root mean square value expected mean value of the pose residual to evaluate the calibration effect of the robot.

In order to achieve the above purpose, another embodiment of the present application proposes a robot measurement pose evaluation device for kinematic calibration, including:

A modeling module is used to establish a geometric error model of the robot;

The calculation module is used to determine and discretize the task space of the robot, determine the measurement noise intensity weight and the pose accuracy weight of the robot, and calculate each group of poses in the discretized task space of the robot , determining a set of corresponding error transfer matrices according to the pose accuracy weights and preprocessing them;

An evaluation module, configured to determine an error identification equation according to the pose to be evaluated and the geometric error model, and determine a discretized kinematically calibrated pose residual in the task space of the robot according to the measured noise intensity weight The expected mean value of the root mean square value of , and the expected mean value of the root mean square value of the pose residual is used to evaluate the calibration effect of the robot.

The robot measurement pose evaluation method and evaluation device for kinematics calibration of the embodiment of the present application proposes a measurement pose evaluation method that considers the pose residual in the task space after kinematic calibration based on the group of measurement poses , for a given set of measured poses and a task space where the robot needs to ensure accuracy, the estimated square value of the root mean square of the pose residual error in the task space after kinematic calibration based on the set of measured poses is expected As a standard for evaluating the group of measured poses, and considering different noise intensities of the measured components and different weights of pose accuracy, the corresponding weighted evaluation indicators are given. The proposed index is different from the previous evaluation index that characterizes the observability of kinematic parameter errors, but is used to characterize the amplitude of the residual error after kinematic calibration in the robot task space, and is used in the field of robot kinematic calibration. The method optimizes the selected measurement poses to improve the accuracy of the robot or reduce the number of measurement poses required for kinematics calibration. As a result, the problem that the current measurement pose evaluation method focuses on the observability of geometric errors and cannot directly reflect the effect of kinematics calibration is solved, so that the measurement pose evaluation and kinematics calibration can be linked more accurately and intuitively, reflecting the robot's task space. The magnitude of the residual after kinematic calibration.

Additional aspects and advantages of the application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the application.

Description of drawings

The above and/or additional aspects and advantages of the present application will become apparent and easy to understand from the following description of the embodiments in conjunction with the accompanying drawings, wherein:

Fig. 1 is a flow chart of a robot measurement pose evaluation method for kinematics calibration according to an embodiment of the present application;

Fig. 2 is a typical hybrid robot configuration according to an embodiment of the present application;

Figure 3 is a schematic diagram of a typical task space discretization;

Fig. 4 is a schematic structural diagram of a robot measurement pose evaluation device for kinematics calibration according to an embodiment of the present application.

Reference signs: 1-first branch; 2-second branch; 3-third branch; 4-lower fixed platform; 5-C-shaped member; 6-A-shaped member; 7-moving platform; 8-upper fixed platform.

Detailed ways

Embodiments of the present application are described in detail below, examples of which are shown in the drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the figures are exemplary, and are intended to explain the present application, and should not be construed as limiting the present application.

The robot measurement pose evaluation method and evaluation device for kinematics calibration proposed according to the embodiments of the present application will be described below with reference to the accompanying drawings.

Firstly, a robot measurement pose evaluation method for kinematics calibration proposed according to an embodiment of the present application will be described with reference to the accompanying drawings.

Fig. 1 is a flow chart of a robot measurement pose evaluation method for kinematics calibration according to an embodiment of the present application.

As shown in Figure 1, the robot measurement pose evaluation method for kinematics calibration includes the following steps:

Step S1, establishing a geometric error model of the robot.

Optionally, in one embodiment of the present application, the geometric error model of the robot is established as:

Among them, δb _E is the position of the robot end effector, ω _E is the attitude error of the robot end effector, Represents a total of n uncorrelated geometric errors, and M is the corresponding error transfer matrix, which represents the influence of the geometric error in ∈ on the position and attitude error of the robot end effector, which is a function of the displacement vector q of the robot drive shaft.

Step S2, determine and discretize the robot's task space, determine the robot's measurement noise intensity weight and pose accuracy weight, for each group of poses in the discretized robot task space, determine according to the pose accuracy weight Correspondingly the set of error transfer matrices is preprocessed.

Determine the basic information of the robot and perform preprocessing. First, determine and discretize the task space of the robot. The task space of the robot refers to the space formed by the end pose of the robot to achieve the task requirements, and usually needs to ensure the positioning accuracy. The task space is a space in which the robot's terminal pose is continuously distributed. Since there are infinite elements in the space, it needs to be discretized for analysis. Discretization refers to the representation of the overall space by using the limited elements in the space. Uniform sampling is achieved. The discretized _task space can be expressed as a set Π={s ₁ , s ₂ , .

Secondly, determine the measurement noise intensity weight of the robot, and measure the end pose error at the measurement pose of the robot. The measurement accuracy will be affected by the measurement noise, and the measurement noise It is assumed that an independent normal distribution with a mean of 0 is satisfied, but the variance of the normal distribution is inconsistent due to the different component strengths, /> The variance matrix of is normalized to a symmetric positive definite matrix W ^-1 , and the normalization method can be used to scale its specific elements to 1 or other methods, which can be used to characterize the measurement noise intensity weight of the robot, and through the measurement instrument and measurement scheme determined a priori.

Thirdly, determine the weight of the robot’s pose accuracy: the robot’s position and attitude are expressed in different units, and the actual robot’s requirements for position and attitude accuracy are also different. The pose accuracy weight can be expressed as a diagonal matrix C, especially , when the required ratio of position and attitude accuracy is r(rad ⁻¹ ), C can be determined as C=diag(1, 1, 1, r, r, r).

Finally, data preprocessing: For each group of poses in the discretized task space Π, determine the corresponding set of error transfer matrices and write it as Π _A ={A ₁ , A ₂ ,...,A _m }, where A _i =M(s _i ), preconditioning matrix The robot needs to be calculated only once for a specific structural parameter, task space and geometric error model.

Step S3, determine the error identification equation according to the pose to be evaluated and the geometric error model, and determine the expected mean value of the root mean square value of the kinematically calibrated pose residual in the discretized task space of the robot according to the weight of the measurement noise intensity , using the expected mean value of the root mean square value of the pose residual to evaluate the calibration effect of the robot.

Optionally, in one embodiment of the present application, S3 further includes:

According to the pose to be evaluated and the geometric error model, the error identification equation is determined as:

δ=M _a ∈

in, M _i is the error transfer matrix, and δ is the corresponding pose error measurement;

Based on the robot's measurement noise intensity weights, an estimate of the geometric error is determined according to a weighted least squares method Among them, W is the inverse matrix of the measurement noise variance matrix;

Determine the estimated and actual values of the geometric error, and normalize the pose residual after kinematic calibration in the task space of the discretized robot where, /> is the estimated value of the geometric error, ∈ ^* is the actual value of the geometric error, C is the matrix corresponding to the pose accuracy weight, A _i is the error transfer matrix corresponding to the pose s _i ;

Determine the root mean square of the kinematically calibrated pose residual in the task space of the discretized robot based on the estimated value and the pose residual where m is the total number of poses in the discretized task space;

Determining the expected mean value of root mean square value based on statistical knowledge Among them, A is the preprocessing matrix.

Specifically, to evaluate any measurement pose, first, based on the measurement pose combination Ω to be evaluated and the geometric error model, the error identification equation can be determined as δ=M _a ∈, where As determined by stacking of the error transfer matrix M _i =M(q _i ), δ is the corresponding pose error measurement.

Based on the robot's measurement noise intensity weights, an estimate of ∈ can be determined according to weighted least squares The widely used least squares method ∈=MaTMa-1MaTδ can be considered as a special case of equalizing the components of the noise intensity.

Estimated value for ∈ and the actual value ∈ ^* , the normalized pose residual after kinematic calibration at the task space _si can be expressed as /> Among them, the pose accuracy weight matrix C of the robot converts the pose error into the same unit to achieve normalization, and directly considers As a normalized error, it is a special case of the weight matrix C as an identity matrix.

Based on the above estimates, the root mean square of the kinematically calibrated pose residual in the discretized task space Π is expressed as Through the analysis of statistical knowledge, the expected mean value of its square value can be determined Taking η as the pose evaluation value proposed by this method, the larger the value is, the larger the pose residual after kinematic calibration is, and the worse the kinematic calibration effect is.

It should be noted that for the pose evaluation index In , due to the nature of the matrix, η can also be expressed as /> and other equivalent forms.

Figure 2 shows a typical configuration of a hybrid robot. The five-degree-of-freedom hybrid robot includes a three-degree-of-freedom parallel mechanism and a two-degree-of-freedom serial mechanism connected in series with the parallel mechanism. The three-degree-of-freedom parallel mechanism includes an upper fixed platform 8 , a lower fixed platform 4 , a parallel linkage platform 7 and three branch components 1 , 2 , 3 . Among the three branch assemblies, the first branch assembly 1 and the second branch assembly 2 with the same structure are on the same plane and pass through the upper fixed platform 8, and are connected with the upper fixed platform 8 by a rotary hinge. The third branch assembly 3 passes through the fixed platform 4 and is connected with the fixed platform 4 with a rotary hinge. The front ends of the first branch assembly 1 and the second branch assembly 2 are connected to the parallel linkage platform 7 through rotating hinges, and the front ends of the third branch assembly 3 are fixedly connected to the parallel linkage platform 7 . The two-degree-of-freedom attitude series mechanism includes a C-shaped member 5 and an A-shaped member 6 . The C-shaped member 5 is connected with the parallel linkage platform 7 with a rotary hinge. The first end of the A-shaped component 6 is provided with a matching hole connected to the knife handle, and the plane where the hole is located is used as the terminal moving platform of the robot, and the second end is connected with the C-shaped component through a rotary hinge. The C-shaped member 5, the A-shaped member 6 and the three branch assemblies 1, 2, 3 are used as five drive shafts of the robot. The proposed flow chart of a robot measurement pose evaluation method for kinematics calibration is applied to the hybrid robot, and the specific method steps are as follows:

1) By analyzing the configuration of the robot, the geometric error model of the robot can be established:

where δb _E and ω _E represent the position and attitude errors of the robot end effector respectively, Represents a total of 38 uncorrelated geometric errors, which can be expressed as:

M is the corresponding error transfer matrix, which represents the influence of the geometric error in ∈ on the position and attitude error of the robot end effector, and is the robot drive axis displacement vector q=[l ₁ , l ₂ , l ₃ , θ _C , θ _A ] A function of ^T , where l ₁ , l ₂ and l ₃ are the lengths of the three branches, respectively, and θ _C and θ _A are the rotation angles of the C-type and A-type members relative to the initial pose.

2) Determine the basic information and preprocessing of the robot, mainly including:

2-1) Determine and discretize the task space of the robot: Take the 2-dimensional space shown in Figure 3 as an example, where the rectangular range represents the task space of the robot, then the set of diamond-shaped points uniformly distributed in the rectangular range in Figure 3 can be as a discretized task space for the robot, where ellipses represent diamond-shaped points in between. For the specific implementation of the five-degree-of-freedom hybrid robot, the discretization method in the 5-dimensional space shown in Figure 3 can be used. The discretized task space can be expressed as a set of robot drive axis displacement vectors Π={s ₁ , s ₂ , ..., _sm }.

2-2) Determine the weight of the robot's measurement noise intensity: the pose measurement noise can be determined a priori through the measuring instruments and measurement schemes in the kinematic calibration of the five-degree-of-freedom hybrid robot The variance matrix of is a symmetric positive definite matrix P, and the As a normalized variance matrix, where P(1, 1) is the value in row 1, column 1 of matrix P.

2-3) Determining the pose accuracy weight of the robot: the five-DOF hybrid robot has different requirements for position and attitude accuracy, and its pose accuracy weight can be expressed as a diagonal matrix C. In particular, the position and attitude accuracy When the required ratio is r(rad ⁻¹ ), C can be determined as C=diag(1, 1, 1, r, r, r), and r can be adjusted according to the requirements of the robot.

2-4) Data preprocessing: For each group of poses in Π, determine the corresponding set of error transfer matrices as Π _A ={A ₁ , A ₂ ,...,A _m }, where A _i =M (s _i ), the preconditioning matrix

3) For any set of measured poses to be evaluated, it is expressed as a set of displacement vectors of the robot's drive shaft Ω={q ₁ , q ₂ , . . . , q _n }.

4) Determine the evaluation value of this group of measurement poses: based on the measurement pose combination Ω to be evaluated and the geometric error model, the error identification equation can be determined as: δ=M _a ∈, where is determined by stacking the error transfer matrix M _i =M(q _i ), and δ is the corresponding measurement value of the pose error. Through the analysis of statistical knowledge, it can be determined that the mean square root square expected mean /> η is the evaluation value of this group of poses calculated according to this method.

According to the robot measurement pose evaluation method for kinematic calibration proposed in the embodiment of the present application, a measurement pose evaluation method considering the pose residual in the task space after kinematic calibration based on the group of measurement poses is proposed, For a given set of measurement poses and a task space where the robot needs to ensure accuracy, the estimated square value of the root mean square of the pose residual error in the task space after kinematic calibration based on the set of measurement poses is taken as Evaluate the standard of this group of measurement poses, and give corresponding weighted evaluation indexes considering different noise strengths of measurement components and different weights of pose accuracy. The proposed index is different from the previous evaluation index that characterizes the observability of kinematic parameter errors, but is used to characterize the amplitude of the residual error after kinematic calibration in the robot task space, and is used in the field of robot kinematic calibration. The method optimizes the selected measurement poses to improve the accuracy of the robot or reduce the number of measurement poses required for kinematics calibration. As a result, the problem that the current measurement pose evaluation method focuses on the observability of geometric errors and cannot directly reflect the effect of kinematics calibration is solved, so that the measurement pose evaluation and kinematics calibration can be linked more accurately and intuitively, reflecting the robot's task space. The magnitude of the residual after kinematic calibration.

Next, a robot measurement pose evaluation device for kinematics calibration proposed according to an embodiment of the present application will be described with reference to the accompanying drawings.

Fig. 4 is a schematic structural diagram of a robot measurement pose evaluation device for kinematics calibration according to an embodiment of the present application.

As shown in FIG. 4 , the robot measurement pose evaluation device for kinematic calibration includes: a modeling module 100 , a calculation module 200 and an evaluation module 300 .

The modeling module 100 is used to establish a geometric error model of the robot.

The calculation module 200 is used to determine and discretize the task space of the robot, determine the measurement noise intensity weight and the pose accuracy weight of the robot, and for each group of poses in the discretized robot task space, according to the pose accuracy The weights determine the set of corresponding error transfer matrices and preprocess them.

The evaluation module 300 is used to determine the error identification equation according to the pose to be evaluated and the geometric error model, and determine the root mean square square of the kinematically calibrated pose residual in the task space of the discretized robot according to the measurement noise intensity weight The expected mean value of the value is used, and the expected mean value of the root mean square value of the pose residual is used to evaluate the calibration effect of the robot.

Optionally, in one embodiment of the present application, the geometric error model is:

Among them, δb _E is the position of the robot end effector, ω _E is the attitude error of the robot end effector, Represents a total of n uncorrelated geometric errors, and M is the corresponding error transfer matrix, which represents the influence of the geometric error in ∈ on the position and attitude error of the robot end effector, which is a function of the displacement vector q of the robot drive shaft.

Optionally, in one embodiment of the present application, for each group of poses in the task space of the discretized robot, determining a corresponding set of error transfer matrices and preprocessing it includes:

For each group of poses in the discretized task space of the robot, determine the corresponding error transfer matrix set as: Π _A ={A ₁ , A ₂ ,...,A _m }, where A _i =M (s _i ), the preconditioning matrix C is the matrix corresponding to the pose accuracy weights, s _i is the displacement vector of the ith drive axis in the discretized task space, and m is the total number of poses in the discretized task space.

Optionally, in one embodiment of the present application, the evaluation module 300 is specifically configured to determine the error identification equation according to the pose to be evaluated and the geometric error model as:

δ=M _a ∈

in, M _i is the error transfer matrix, and δ is the corresponding pose error measurement;

Based on the robot's measurement noise intensity weights, an estimate of the geometric error is determined according to a weighted least squares method Among them, W is the inverse matrix of the measurement noise variance matrix;

Determine the estimated and actual values of the geometric error, and normalize the pose residual after kinematic calibration in the task space of the discretized robot where, /> is the estimated value of the geometric error, ∈ ^* is the actual value of the geometric error, C is the matrix corresponding to the pose accuracy weight, A _i is the error transfer matrix corresponding to the pose s _i ;

Determine the root mean square of the kinematically calibrated pose residual in the task space of the discretized robot based on the estimated value and the pose residual where m is the total number of poses in the discretized task space;

Determining the expected mean value of root mean square value based on statistical knowledge Among them, A is the preprocessing matrix.

Optionally, in one embodiment of the present application, the expected mean value of the root mean square value of the pose residual is used to evaluate the calibration effect of the robot, including:

The larger the expected mean value of the root mean square value of the pose residual, the worse the calibration effect of the robot.

It should be noted that the foregoing explanations of the method embodiment are also applicable to the device of this embodiment, and details are not repeated here.

According to the robot measurement pose evaluation device for kinematic calibration proposed in the embodiment of the present application, a measurement pose evaluation method considering the pose residual in the task space after kinematic calibration based on the group of measurement poses is proposed, For a given set of measurement poses and a task space where the robot needs to ensure accuracy, the estimated square value of the root mean square of the pose residual error in the task space after kinematic calibration based on the set of measurement poses is taken as Evaluate the standard of this group of measurement poses, and give corresponding weighted evaluation indexes considering different noise strengths of measurement components and different weights of pose accuracy. The proposed index is different from the previous evaluation index that characterizes the observability of kinematic parameter errors, but is used to characterize the amplitude of the residual error after kinematic calibration in the robot task space, and is used in the field of robot kinematic calibration. The method optimizes the selected measurement poses to improve the accuracy of the robot or reduce the number of measurement poses required for kinematics calibration. As a result, the problem that the current measurement pose evaluation method focuses on the observability of geometric errors and cannot directly reflect the effect of kinematics calibration is solved, so that the measurement pose evaluation and kinematics calibration can be linked more accurately and intuitively, reflecting the robot's task space. The magnitude of the residual after kinematic calibration.

In addition, the terms "first" and "second" are used for descriptive purposes only, and cannot be interpreted as indicating or implying relative importance or implicitly specifying the quantity of indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include at least one of these features. In the description of the present application, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

In the description of this specification, descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific examples", or "some examples" mean that specific features described in connection with the embodiment or example , structure, material or characteristic is included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the described specific features, structures, materials or characteristics may be combined in any suitable manner in any one or more embodiments or examples. In addition, those skilled in the art can combine and combine different embodiments or examples and features of different embodiments or examples described in this specification without conflicting with each other.

Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limitations on the present application, and those skilled in the art can make the above-mentioned The embodiments are subject to changes, modifications, substitutions and variations.

Claims (6)

1. A robot measurement pose evaluation method for kinematics calibration, is characterized in that, comprises the following steps: Establish the geometric error model of the robot; Determining and discretizing the task space of the robot, determining the measurement noise intensity weight and pose accuracy weight of the robot, and for each group of poses in the discretized task space of the robot, according to the position The attitude accuracy weight determines the corresponding error transfer matrix set and preprocesses it; Determine the error identification equation according to the pose to be evaluated and the geometric error model, and determine the root mean square square of the kinematically calibrated pose residual in the task space of the robot after discretization according to the measurement noise intensity weight value Value expected mean value, utilize the root mean square value expected mean value of described pose residual to evaluate the calibration effect of described robot; Wherein, for each group of poses in the discretized task space of the robot, determining a corresponding set of error transfer matrices and preprocessing it includes: for each group of poses in the discretized task space of the robot For a set of poses, determine the set of corresponding error transfer matrices as: , where /> , the preprocessing matrix , C is the matrix corresponding to the pose accuracy weight, /> is the i-th driving axis displacement vector in the discretized task space, m is the total number of poses in the discretized task space; Determine the error identification equation according to the pose to be evaluated and the geometric error model, and determine the root mean square square of the kinematically calibrated pose residual in the task space of the robot after discretization according to the measurement noise intensity weight value The expected mean of values, including: The error identification equation determined according to the pose to be evaluated and the geometric error model is: in, , /> is the error transfer matrix, /> is the measured value of the corresponding pose error; Based on the measured noise intensity weights of the robot, an estimate of the geometric error is determined according to a weighted least squares method ; where /> is the inverse matrix of the measurement noise variance matrix; Determining the estimated value and the actual value of the geometric error, and normalizing the pose residual after kinematic calibration in the task space of the robot after discretization ; where /> is the geometric error estimate, /> is the actual value of the geometric error, /> is the matrix corresponding to the pose precision weight, /> for pose /> The corresponding error transfer matrix; Determine the root mean square of the kinematically calibrated pose residual in the task space of the robot after discretization according to the estimated value and the pose residual , where m is the total number of poses in the discretized task space; Determine the expected mean value of the root mean square value according to statistical knowledge , where A is the preprocessing matrix. 2. The method according to claim 1, wherein the geometric error model is: in, is the position of the robot end effector, /> is the attitude error of the robot end effector, /> On behalf of shared /> uncorrelated geometric errors, /> is the corresponding error transfer matrix, denoting /> The influence of the geometric error in on the position and attitude error of the robot end effector is the displacement vector of the robot drive shaft /> The function. 3. The method according to claim 1, characterized in that, utilizing the root mean square value expected mean value of the pose residual to evaluate the calibration effect of the robot, comprising: The greater the expected mean value of the root mean square value of the pose residual, the worse the calibration effect of the robot. 4. A robot measurement pose evaluation device for kinematics calibration, characterized in that it comprises: A modeling module is used to establish a geometric error model of the robot; The calculation module is used to determine and discretize the task space of the robot, determine the measurement noise intensity weight and the pose accuracy weight of the robot, and calculate each group of poses in the discretized task space of the robot , determining a set of corresponding error transfer matrices according to the pose accuracy weights and preprocessing them; An evaluation module, configured to determine an error identification equation according to the pose to be evaluated and the geometric error model, and determine a discretized kinematically calibrated pose residual in the task space of the robot according to the measured noise intensity weight The root mean square value expected mean value of the root mean square value, utilizes the root mean square value expected mean value of the pose residual to evaluate the calibration effect of the robot; For each group of poses in the discretized task space of the robot, determining a set of corresponding error transfer matrices and preprocessing it includes: For each group of poses in the discretized task space of the robot, the set of corresponding error transfer matrices is determined as: , where /> , the preprocessing matrix /> , C is the matrix corresponding to the pose accuracy weight, /> is the i-th driving axis displacement vector in the discretized task space, m is the total number of poses in the discretized task space; The evaluation module is specifically used to determine the error identification equation according to the pose to be evaluated and the geometric error model as: in, , /> is the error transfer matrix, /> is the measured value of the corresponding pose error; Based on the measured noise intensity weights of the robot, an estimate of the geometric error is determined according to a weighted least squares method ; where /> is the inverse matrix of the measurement noise variance matrix; Determining the estimated value and the actual value of the geometric error, and normalizing the pose residual after kinematic calibration in the task space of the robot after discretization ; where /> is the geometric error estimate, /> is the actual value of the geometric error, /> is the matrix corresponding to the pose precision weight, /> for pose /> The corresponding error transfer matrix; Determine the root mean square of the kinematically calibrated pose residual in the task space of the robot after discretization according to the estimated value and the pose residual , where m is the total number of poses in the discretized task space; Determine the expected mean value of the root mean square value according to statistical knowledge , where A is the preprocessing matrix. 5. The device according to claim 4, wherein the geometric error model is: in, is the position of the robot end effector, /> is the attitude error of the robot end effector, /> On behalf of shared /> uncorrelated geometric errors, /> is the corresponding error transfer matrix, denoting /> The influence of the geometric error in on the position and attitude error of the robot end effector is the displacement vector of the robot drive shaft /> The function. 6. The device according to claim 4, characterized in that, utilizing the root mean square value expected mean value of the pose residual to evaluate the calibration effect of the robot, comprising: The greater the expected mean value of the root mean square value of the pose residual, the worse the calibration effect of the robot.