CN106737855A - A kind of robot precision compensation method of comprehensive position and attitude error model and rigidity compensation - Google Patents

Abstract

The invention discloses robot precision's compensation method of a kind of comprehensive position and attitude error model and rigidity compensation, comprise the following steps：Step 1, set up robot motion model according to the structural parameters of robot；Step 2, set up robot inaccuracy model；Step 3, in robot working space, random given object pose point, when robot end moves to specified point, record joint angles now；Step 4, the actual coordinate Pa using the given object pose point of position measuring instrument measurement；Step 5, error parameter is recognized using least square method；Step 6, in robot end's imposed load, measure its deflection, return to step 3 compensates the structural failure after recognizing again to motion model again afterwards, so as to eliminate end position and attitude error caused by the deformation that load causes, the deformation data that load causes is stored in database simultaneously, for the accuracy compensation in later stage.The present invention can significantly improve the absolute fix precision of robot, simply, efficiently.

Description

A kind of robot precision compensation method of comprehensive position and attitude error model and rigidity compensation

Technical field

The invention belongs to Robot calibration technical field, the machine of particularly a kind of comprehensive position and attitude error model and rigidity compensation Device people's precision compensation method.

Background technology

With《The outline of made in China 2025》Promulgation, industrial robot will China start again development upsurge.Essence Spend includes repetitive positioning accuracy and absolute fix precision as the important indicator for weighing industrial robot performance.Present industrial machine Device people's repetitive positioning accuracy is high, and absolute fix precision is low, is unfavorable for that off-line programing and high accuracy are processed.And robot motion The effective means that demarcation is raising robot localization precision is learned, is mainly included modeling, is measured, recognizes, compensating four-stage.

Traditional method carries out Robot calibration, is only directed to kinematics model parameter and causes error to compensate, and does not have There is the deflection for considering that load causes, do not carry out comprehensive compensation.And traditional rigidity compensation aspect is, it is necessary to carry out substantial amounts of formula Derive, solve each joint amount of deflection of robot and armed lever amount of deflection, set up rigidity model.Process is complicated, less efficient.

The content of the invention

Technical problem solved by the invention is the robot for providing a kind of comprehensive position and attitude error model and rigidity compensation Precision compensation method.

The technical solution for realizing the object of the invention is：A kind of robot of comprehensive position and attitude error model and rigidity compensation Precision compensation method, comprises the following steps：

Step 1, set up robot motion model according to the structural parameters of robot；Specially：

Step 1-1, the homogeneous transform matrix between robot adjacent segment is set up according to DH methods, the matrix is：

In above formula, a_iIt is length of connecting rod, α_iIt is joint torsional angle, d_iIt is connecting rod offset distance, θ_iIt is joint rotation angle, x is connecting rod coordinate It is X-axis coordinate, z is link rod coordinate system Z axis coordinate；

Step 1-2, the homogeneous transform matrix to being set up in step 1-1 introduce the rotation Rot (y, β) around y-axis, eliminate middle Between connecting rod because axle it is parallel or almost parallel when produce it is unusual, adjacent segment movement relation is modified to：

In above formula, β_iThe anglec of rotation of robot the i-th bar coordinate system around y-axis is represented, y is link rod coordinate system y-axis coordinate, c It is cos, s is sin；

Step 1-3, one additional parameter z of increase_n, description instrument coordinate system builds along the translation of tail end connecting rod coordinate system z-axis Found tool coordinates system is relative to the transformation matrix of tail end connecting rod coordinate system：

Step 1-4, for N articulated robots, according to above-mentioned steps obtain robot tool coordinate system and base coordinate system it Between kinematic relation be：

T=⁰A₁×¹A₂ײA₃׳A₄×⁴A₅×⁵A₆×…×^n-1A_n

The value of n is 1...N, and N >=1 kinematic relation is robot motion model.

Step 2, the robot motion model set up according to step 1 set up robot inaccuracy model；Specially：

Step 2-1, intermediate connecting rod actual converted matrix is obtained by differential transform principle with nominal transition matrix error dT_i For：

And

Ti is intermediate connecting rod transition matrix；

Step 2-2, tool coordinates system is obtained by differential transform principle relative to tail end connecting rod Conversion Matrix of Coordinate error：

Tn is tool coordinates system and tail end connecting rod Conversion Matrix of Coordinate；

Step 2-3, for n articulated robots, its final error model be intermediate connecting rod error add tool coordinates system Relative to the error that tail end connecting rod is converted, following formula is obtained after differential of demanding perfection：

Wherein p is terminal position coordinate；

Write：

Δ p=J_δΔX

Wherein Δ X=(Δ θ₁...Δθ_n, Δ α₁...Δα_n,Δa₁...Δa_n,Δd₁...Δd_n,Δβ₁...Δβ_n,Δ z_n)^TRepresent each bar linkage structure error J of robot_δIt is identification Jacobian matrix.

Step 3, in robot working space, random given object pose point, its nominal coordinate is Pn, when robot end When end moves to specified point, record joint angles now；

Step 4, the actual coordinate Pa using the given object pose point of position measuring instrument measurement；

Step 5, error parameter is recognized using least square method, the structural failure that will be picked out is compensated gives motion mould Type name parameter, is verified whether to meet requirement, and next step, otherwise return to step 3 are performed if meeting and requiring；

It is specifically to add structural parameters to pick out that the structural failure that will be picked out is compensated to motion model name parameter Error parameter；

Verify whether that meeting requirement specifically refers to：Data before measurement actual coordinate, with compensation are compared, if meet The margin of tolerance ± 0.2mm, if in the margin of tolerance, meeting and requiring, is otherwise unsatisfactory for requiring.

Step 6, in robot end's imposed load, measure its deflection, afterwards return to step 3 will again recognize after knot Structure error is compensated to motion model again, so as to eliminate end position and attitude error caused by the deformation that load causes, while by load The deformation data for causing is stored in database, for the accuracy compensation in later stage.In robot end's imposed load, its deformation is measured Amount is specially：

Step 6-1, in the range of rated load, to robot end apply different quality load, using stress-strain gage Each part distortion amount of robot measurement, and using the actual coordinate of apparatus measures now robot arrival specified point；

Step 6-2, measurement result is processed, obtain each part distortion amount of robot with the curve of load change, obtained Take curve of robot end's site error with load change.

Compared with prior art, its remarkable advantage is the present invention：(1) kinematics model that the present invention is used is in MDH methods On the basis of increased tool coordinates system, constitute 6 parameter models so that motion model is more complete, can preferably describe machine People's model.(2) method of the present invention uses least square method to robot architecture's error identification, can be with softwares such as matlab Quickly obtain structural failure identification result.(3) after the method for the present invention is compensated according to position and attitude error model, for machine The deflection that people's end load causes has carried out second compensation, without the need for rigidity model is set up to robot again, eliminates Complicated algorithm development process.(4) method of the present invention obtains change curve of the robot body deflection to load so that machine Device people transfers offline result when similar operating mode is run into, and improves robot and works online efficiency.

The present invention is described in further detail below in conjunction with the accompanying drawings.

Brief description of the drawings

Fig. 1 is robot precision's compensation method flow of a kind of comprehensive position and attitude error model of the invention and rigidity compensation Figure.

Fig. 2 is that each armed lever end caused by robot end's imposed load deforms, wherein figure (a) bends for armed lever stress Deformation pattern, figure (b) is armed lever stress stretcher strain figure, and figure (c) is armed lever stress torsional deflection figure.

Specific embodiment

With reference to Fig. 1, Fig. 2, the robot precision compensation side of a kind of comprehensive position and attitude error model of the invention and rigidity compensation Method, comprises the following steps：

Step 1, set up robot motion model according to the structural parameters of robot；Specially：

Step 1-1, the homogeneous transform matrix between robot adjacent segment is set up according to DH methods, the matrix is：

In above formula, a_iIt is length of connecting rod, α_iIt is joint torsional angle, d_iIt is connecting rod offset distance, θ_iIt is joint rotation angle, X is connecting rod coordinate It is X-axis, Z is link rod coordinate system Z axis.

Step 1-2, the homogeneous transform matrix to being set up in step 1-1 introduce the rotation Rot (y, β) around y-axis, eliminate middle Between connecting rod because axle it is parallel or almost parallel when produce it is unusual, adjacent segment movement relation is modified to：

In above formula, β_iThe anglec of rotation of robot the i-th bar coordinate system around y-axis is represented, y is link rod coordinate system y-axis, and c is Cos, s are sin；

Step 1-3, one additional parameter z of increase_n, description instrument coordinate system builds along the translation of tail end connecting rod coordinate system z-axis Found tool coordinates system is relative to the transformation matrix of tail end connecting rod coordinate system：

Step 1-4, for N articulated robots, according to above-mentioned steps obtain robot tool coordinate system and base coordinate system it Between kinematic relation be：

T=⁰A₁×¹A₂ײA₃׳A₄×⁴A₅×⁵A₆×…×^n-1A_n

The value of n is 1...N, and N >=1, the kinematic relation is robot motion model.

Step 2, the robot motion model set up according to step 1 set up robot inaccuracy model；Specially：

Step 2-1, intermediate connecting rod actual converted matrix is obtained by differential transform principle with nominal transition matrix error dT_i For：

And

Ti is intermediate connecting rod transition matrix.

Step 2-2, tool coordinates system is obtained by differential transform principle relative to tail end connecting rod Conversion Matrix of Coordinate error：

Tn is tool coordinates system and tail end connecting rod Conversion Matrix of Coordinate.

Step 2-3, for n articulated robots, its final error model be intermediate connecting rod error add tool coordinates system Relative to the error that tail end connecting rod is converted, following formula is obtained after differential of demanding perfection：

Wherein p is terminal position coordinate.

Write：

Δ p=J_δΔX

Wherein Δ X=(Δ θ₁...Δθ_n, Δ α₁...Δα_n,Δa₁...Δa_n,Δd₁...Δd_n,Δβ₁...Δβ_n,Δ z_n)^TRepresent each bar linkage structure error J of robot_δIt is identification Jacobian matrix.

Step 3, in robot working space, random given object pose point, its nominal coordinate is Pn, when robot end When end moves to specified point, record joint angles now；

Step 4, the actual coordinate Pa using the given object pose point of position measuring instrument measurement；

Step 5, error parameter is recognized using least square method, the structural failure that will be picked out is compensated gives motion mould Type name parameter, is verified whether to meet requirement, and next step, otherwise return to step 3 are performed if meeting and requiring；

It is specifically to add structural parameters to pick out that the structural failure that will be picked out is compensated to motion model name parameter Error parameter；

Verify whether that meeting requirement specifically refers to：Data before measurement actual coordinate, with compensation are compared, if meet The margin of tolerance ± 0.2mm, if in the margin of tolerance, meeting and requiring, is otherwise unsatisfactory for requiring.

Shown in step 6, reference picture 2, the deformation of armed lever can be caused in robot end's imposed load, cause robot end Position error.In robot end's imposed load, its deflection is measured, the structure after return to step 3 will be recognized again afterwards is missed Difference is compensated to motion model again, so that end position and attitude error caused by the deformation that load causes is eliminated, while load is caused Deformation data be stored in database, for the accuracy compensation in later stage.Specially：

Step 6-1, in the range of rated load, to robot end apply different quality load, using stress-strain gage Each part distortion amount of robot measurement, and using the actual coordinate of apparatus measures now robot arrival specified point；

Step 6-2, measurement result is processed, obtain each part distortion amount of robot with the curve of load change, obtained Take curve of robot end's site error with load change.

In sum, the robot precision compensation side of a kind of comprehensive position and attitude error model disclosed by the invention and rigidity compensation Method, given object pose records nominal coordinate and joint angles at random in robot working space, measures set point reality Pose, sets up robot inaccuracy model, and error is recognized by least square method, and the error compensation that will be picked out is to fortune Movable model name parameter, then in robot end's imposed load, measures deflection, carries out second compensation.Realize that robot is exhausted Compensation to positioning precision.The present invention can significantly improve the absolute fix precision of robot, simply, efficiently.

Claims (5)

1. robot precision's compensation method of a kind of comprehensive position and attitude error model and rigidity compensation, it is characterised in that including following Step：

Step 1, set up robot motion model according to the structural parameters of robot；

Step 2, the robot motion model set up according to step 1 set up robot inaccuracy model；

Step 3, in robot working space, random given object pose point, its nominal coordinate is Pn, when robot end moves When moving specified point, record joint angles now；

Step 4, the actual coordinate Pa using the given object pose point of position measuring instrument measurement；

Step 5, error parameter is recognized using least square method, the structural failure that will be picked out is compensated gives motion model name Adopted parameter, is verified whether to meet requirement, and next step, otherwise return to step 3 are performed if meeting and requiring；

Step 6, in robot end's imposed load, measure its deflection, afterwards return to step 3 will again recognize after structure miss Difference is compensated to motion model again, so that end position and attitude error caused by the deformation that load causes is eliminated, while load is caused Deformation data be stored in database, for the accuracy compensation in later stage.

2. robot precision's compensation method of comprehensive position and attitude error model and rigidity compensation, its feature according to claim 1 It is that the structural parameters in step 1 according to robot set up robot motion model, specially：

Step 1-1, the homogeneous transform matrix between robot adjacent segment is set up according to DH methods, the matrix is：

$\begin{array}{c}{A_{i-1}}_i=Rot(z,{\mathrm{\θ}}_i)\×Trans(a_i,0,0)\×Trans(0,0,d_i)\×Rot(x,{\mathrm{\α}}_i)\\ =\left[\begin{array}{cccc}{\mathrm{cos\θ}}_i& -{\mathrm{sin\θ}}_i{\mathrm{cos\α}}_i& {\mathrm{sin\θ}}_i{\mathrm{sin\α}}_i& a_i{\mathrm{cos\θ}}_i\\ {\mathrm{sin\θ}}_i& \cos a_i{\mathrm{cos\θ}}_i& -{\mathrm{cos\θ}}_i{\mathrm{sin\α}}_i& a_i{\mathrm{sin\θ}}_i\\ 0& {\mathrm{sin\α}}_i& {\mathrm{cos\α}}_i& d_i\\ 0& 0& 0& 1\end{array}\right]\end{array}$

In above formula, a_iIt is length of connecting rod, α_iIt is joint torsional angle, d_iIt is connecting rod offset distance, θ_iIt is joint rotation angle, x is link rod coordinate system X-axis Coordinate, z is link rod coordinate system Z axis coordinate；

Step 1-2, the homogeneous transform matrix to being set up in step 1-1 introduce the rotation Rot (y, β) around y-axis, eliminate intermediate connecting rod Between because axle it is parallel or almost parallel when produce it is unusual, adjacent segment movement relation is modified to：

$\begin{array}{c}{A_{i-1}}_i=Rot(x_i,{\mathrm{\α}}_i)\×Trans(a_i,0,0)\×Rot(z_i,{\mathrm{\θ}}_i)\×Trans(0,0,d_i)\×Rot(y,{\mathrm{\β}}_i)\\ =\left[\begin{array}{cccc}{\mathrm{c\β}}_i{\mathrm{c\θ}}_i& -{\mathrm{s\θ}}_i& {\mathrm{s\β}}_i{\mathrm{c\θ}}_i& a_i\\ {\mathrm{c\α}}_i{\mathrm{c\β}}_i{\mathrm{s\θ}}_i+{\mathrm{s\α}}_i{\mathrm{s\β}}_i& {\mathrm{c\α}}_i{\mathrm{c\θ}}_i& {\mathrm{c\α}}_i{\mathrm{s\β}}_i{\mathrm{s\θ}}_i-{\mathrm{c\β}}_i{\mathrm{s\α}}_i& -d_i{\mathrm{s\α}}_i\\ {\mathrm{c\β}}_i{\mathrm{s\α}}_i{\mathrm{s\β}}_i-{\mathrm{c\α}}_i{\mathrm{s\β}}_i& {\mathrm{s\α}}_i{\mathrm{c\θ}}_i& {\mathrm{c\α}}_i{\mathrm{c\β}}_i+{\mathrm{s\α}}_i{\mathrm{s\β}}_i{\mathrm{s\θ}}_i& d_i{\mathrm{c\α}}_i\\ 0& 0& 0& 1\end{array}\right]\end{array}$

In above formula, β_iThe anglec of rotation of robot the i-th bar coordinate system around y-axis is represented, y is link rod coordinate system y-axis coordinate, and c is cos, S is sin；

Step 1-3, one additional parameter z of increase_n, description instrument coordinate system sets up work along the translation of tail end connecting rod coordinate system z-axis Have coordinate system is relative to the transformation matrix of tail end connecting rod coordinate system：

$\begin{array}{c}{A}_n=Rot(x,{\mathrm{\α}}_n)\×Trans(a_n,0,0)\×Rot(z,{\mathrm{\θ}}_7)\×Trans(0,0,d_n)\×Rot(y,{\mathrm{\β}}_n)\×Trans(0,0,z_n)\\ =\left[\begin{array}{cccc}{\mathrm{c\β}}_n{\mathrm{c\θ}}_n& -{\mathrm{s\θ}}_n& {\mathrm{s\β}}_n{\mathrm{c\θ}}_n& a_n+{\mathrm{s\β}}_n{\mathrm{c\θ}}_n\\ {\mathrm{c\α}}_n{\mathrm{c\α}}_n{\mathrm{s\θ}}_n+{\mathrm{s\α}}_n{\mathrm{s\β}}_n& {\mathrm{cvc\θ}}_n& {\mathrm{c\α}}_n{\mathrm{s\β}}_7{\mathrm{s\θ}}_n-{\mathrm{c\β}}_n{\mathrm{s\α}}_n& -z_n*({\mathrm{c\β}}_n{\mathrm{s\α}}_n-{\mathrm{c\α}}_n{\mathrm{s\β}}_n{\mathrm{s\θ}}_n-d_n{\mathrm{s\α}}_n)\\ {\mathrm{c\β}}_n{\mathrm{s\α}}_n{\mathrm{s\β}}_n-{\mathrm{c\α}}_n{\mathrm{s\β}}_n& {\mathrm{s\α}}_n{\mathrm{c\θ}}_n& {\mathrm{c\α}}_n{\mathrm{c\β}}_n+{\mathrm{s\α}}_n{\mathrm{s\β}}_n{\mathrm{s\θ}}_n& z_n*({\mathrm{c\β}}_n{\mathrm{c\α}}_n+{\mathrm{s\α}}_n{\mathrm{s\β}}_n{\mathrm{s\θ}}_n+d_n{\mathrm{c\α}}_n)\\ 0& 0& 0& 1\end{array}\right]\end{array}$

Step 1-4, for N articulated robots, obtained between robot tool coordinate system and base coordinate system according to above-mentioned steps Kinematic relation is：

T=⁰A₁×¹A₂ײA₃׳A₄×⁴A₅×⁵A₆×…×^n-1A_n

The value of n is 1...N, and N >=1 kinematic relation is robot motion model.

3. robot precision's compensation method of comprehensive position and attitude error model according to claim 1 and rigidity compensation, it is special Levy and be, robot inaccuracy model is set up in step 2 and is specially：

Step 2-1, intermediate connecting rod actual converted matrix is obtained by differential transform principle with nominal transition matrix error dT_iFor：

${\mathrm{dT}}_i=\frac{\∂T_i}{\∂{\mathrm{\θ}}_i}{\mathrm{\Δ\θ}}_i+\frac{\∂T_i}{\∂{\mathrm{\α}}_i}{\mathrm{\Δ\α}}_i+\frac{\∂T_i}{\∂d_i}{\mathrm{\Δd}}_i+\frac{\∂T_i}{\∂a_i}{\mathrm{\Δa}}_i+\frac{\∂T_i}{\∂{\mathrm{\β}}_i}{\mathrm{\Δ\β}}_i$

And

Ti is intermediate connecting rod transition matrix；

Step 2-2, tool coordinates system is obtained by differential transform principle relative to tail end connecting rod Conversion Matrix of Coordinate error：

${\mathrm{dT}}_n=\frac{\∂T_n}{\∂{\mathrm{\θ}}_n}{\mathrm{\Δ\θ}}_n+\frac{\∂T_n}{\∂{\mathrm{\α}}_n}{\mathrm{\Δ\α}}_n+\frac{\∂T_n}{\∂d_n}{\mathrm{\Δd}}_n+\frac{\∂T_n}{\∂a_n}{\mathrm{\Δa}}_n+\frac{\∂T_n}{\∂{\mathrm{\β}}_n}{\mathrm{\Δ\β}}_n+\frac{\∂T_n}{\∂z_n}{\mathrm{\Δz}}_n$

Tn is tool coordinates system and tail end connecting rod Conversion Matrix of Coordinate；

Step 2-3, for n articulated robots, its final error model be intermediate connecting rod error plus tool coordinates system it is relative In the error of tail end connecting rod conversion, following formula is obtained after differential of demanding perfection：

$\mathrm{\Δ}p=\underset{i=1}{\overset{n}{\Σ}}\frac{\∂p^N}{\∂{\mathrm{\θ}}_i}{\mathrm{\Δ\θ}}_i+\underset{i=1}{\overset{n}{\Σ}}\frac{\∂p^N}{\∂{\mathrm{\α}}_i}{\mathrm{\Δ\α}}_i+\underset{i=1}{\overset{n}{\Σ}}\frac{\∂p^N}{\∂a_i}{\mathrm{\Δa}}_i+\underset{i=1}{\overset{n}{\Σ}}\frac{\∂p^N}{\∂d_i}{\mathrm{\Δd}}_i+\underset{i=1}{\overset{n}{\Σ}}\frac{\∂p^N}{\∂\mathrm{\β}i}{\mathrm{\Δ\β}}_i+\frac{\∂p^N}{\∂z_n}{\mathrm{\Δz}}_n$

Wherein p is terminal position coordinate；

Write：

$\begin{array}{c}\mathrm{\Δ}p=J_{\mathrm{\δ}}\mathrm{\Δ}X\\ J_{\mathrm{\δ}}=\left[\begin{array}{cccccccccccccccc}\frac{\∂p_x}{\∂a_1}& \mathrm{...}& \frac{\∂p_x}{\∂a_n}& \frac{\∂p_x}{\∂d_1}& \mathrm{...}& \frac{\∂p_x}{\∂d_n}& \frac{\∂p_x}{\∂{\mathrm{\α}}_1}& \mathrm{...}& \frac{\∂p_x}{\∂{\mathrm{\α}}_n}& \frac{\∂p_x}{\∂{\mathrm{\θ}}_1}& \mathrm{...}& \frac{\∂p_x}{\∂{\mathrm{\θ}}_n}& \frac{\∂p_x}{\∂{\mathrm{\β}}_1}& \mathrm{...}& \frac{\∂p_x}{\∂{\mathrm{\β}}_n}& \frac{\∂p_x}{\∂z_n}\\ \frac{\∂p_y}{\∂a_1}& \mathrm{...}& \frac{\∂p_y}{\∂a_n}& \frac{\∂p_y}{\∂d_1}& \mathrm{...}& \frac{\∂p_y}{\∂d_n}& \frac{\∂p_y}{\∂{\mathrm{\α}}_1}& \mathrm{...}& \frac{\∂p_y}{\∂{\mathrm{\α}}_n}& \frac{\∂p_y}{\∂{\mathrm{\θ}}_1}& \mathrm{...}& \frac{\∂p_y}{\∂{\mathrm{\θ}}_n}& \frac{\∂p_y}{\∂{\mathrm{\β}}_1}& \mathrm{...}& \frac{\∂p_y}{\∂{\mathrm{\β}}_n}& \frac{\∂p_y}{\∂z_n}\\ \frac{\∂p_z}{\∂a_1}& \mathrm{...}& \frac{\∂p_z}{\∂a_n}& \frac{\∂p_z}{\∂d_1}& \mathrm{...}& \frac{\∂p_z}{\∂d_n}& \frac{\∂p_z}{\∂{\mathrm{\α}}_1}& \mathrm{...}& \frac{\∂p_z}{\∂{\mathrm{\α}}_n}& \frac{\∂p_z}{\∂{\mathrm{\θ}}_1}& \mathrm{...}& \frac{\∂p_z}{\∂{\mathrm{\θ}}_n}& \frac{\∂p_z}{\∂{\mathrm{\β}}_1}& \mathrm{...}& \frac{\∂p_z}{\∂{\mathrm{\β}}_n}& \frac{\∂p_z}{\∂z_n}\end{array}\right]\end{array}$

Wherein Δ X=(Δ θ₁...Δθ_n, Δ α₁...Δα_n,Δa₁...Δa_n,Δd₁...Δd_n,Δβ₁...Δβ_n,Δz_n)^T Represent each bar linkage structure error J of robot_δIt is identification Jacobian matrix.

4. robot precision's compensation method of comprehensive position and attitude error model according to claim 1 and rigidity compensation, it is special Levy and be, it is specifically to add structural parameters to distinguish that the structural failure that will be picked out in step 5 is compensated to motion model name parameter Know the error parameter for；

Verify whether that meeting requirement specifically refers to：Data before measurement actual coordinate, with compensation are compared, if meet tolerance Scope ± 0.2mm, if in the margin of tolerance, meeting and requiring, is otherwise unsatisfactory for requiring.

5. robot precision's compensation method of comprehensive position and attitude error model according to claim 1 and rigidity compensation, it is special Levy and be, in robot end's imposed load in step 6, measure its deflection and be specially：

Step 6-1, in the range of rated load, to robot end apply different quality load, using stress-strain gage measure Each part distortion amount of robot, and using the actual coordinate of apparatus measures now robot arrival specified point；

Step 6-2, measurement result is processed, obtain curve of each part distortion amount of robot with load change, obtain machine Device people's terminal position error with load change curve.