CN105538312A - Robot hand grabbing strategic planning method based on environment attracting domain - Google Patents

Abstract

The invention provides a robot hand grabbing strategic planning method based on the environment attracting domain. The robot hand grabbing strategic planning method comprises the steps that model parameters of a robot hand and model parameters of workpieces to be grabbed are extracted according to the type of the robot hand and the structure of the workpieces to be grabbed; the state variables and the output variables of a constraint function with the environment attracting domain are selected based on the model parameters of the robot hand and the model parameters of the workpieces to be grabbed; the states value of the state variables are obtained based on the value range of the state variables, and the output values of the corresponding output variables are obtained according to the state values of the state variables; the environment attracting domain of the constraint function is determined based on the state value of the state variables and the output value of the output variables; the robot hand grabbing strategy is planned based on the environment attracting domain of the constraint function. The application range of the method is wide, and the method can be used for various robot hands with different structures and types and objects with different sizes to be grabbed. By means of the method, the grabbing strategy of the robot hand can be planned quickly and accurately. Sensors are not depended on, the grabbing stability is good, and the cost is low.

Description

Mechanical hand based on environment domain of attraction captures strategic planning method

Technical field

The present invention relates to robotics, particularly relate to a kind of mechanical hand based on environment domain of attraction and capture strategic planning method.

Background technology

Along with the fast development of modern manufacturing industry, industrial machine Man's Demands is promoted gradually.Especially, grasping manipulation is an important step of industrial robot application, realizes capturing for complete operation task particularly important fast and reliablely.Usually, robot grasping manipulation is teaching Grasp Modes, and which carries out parameter adjustment realization by artificial experience or by great many of experiments usually, and its generalization ability is poor, and efficiency is lower.In addition, even if introduce the location grasping manipulation task that associated picture heat transfer agent can complete certain precision, but the relative deficient in stability of crawl strategy of this kind of method, and heavy dependence sensor accuracy, thus increase holistic cost.

Summary of the invention

(1) technical problem that will solve

In order to solve prior art problem, the invention provides a kind of mechanical hand based on environment domain of attraction and capturing strategic planning method.

(2) technical scheme

The invention provides a kind of mechanical hand based on environment domain of attraction and capture strategic planning method, comprising: steps A: according to type and the structure treating grabbing workpiece of mechanical hand, extract the model parameter of mechanical hand and treat the model parameter of grabbing workpiece; Step B: based on mechanical hand and treat that the model parameter of grabbing workpiece chooses state variable and the output variable of the constraint function that there is environment domain of attraction; Step C: the span based on state variable obtains the state value of state variable, is obtained the output valve of corresponding output variable by the state value of state variable; Step D: based on the environment domain of attraction of the state value of state variable and the output valve determination constraint function of output variable; And step e: the environment domain of attraction planning mechanical hand based on constraint function captures strategy.

Preferably, described step B specifically comprises: sub-step B1: the environment domain of attraction of definition constraint function; Sub-step B2: based on mechanical hand and treat that the model parameter of grabbing workpiece chooses the state variable of constraint function, wherein, the element of this state variable X comprises: the finger spacing l of mechanical hand, the projection convex polygon central point abscissa x of grabbing workpiece at horizontal plane is treated _cwith projection convex polygon anglec of rotation θ; Sub-step B3: based on mechanical hand and treat that the model parameter of grabbing workpiece chooses the output variable of constraint function, wherein, this output variable is that first group of finger of mechanical hand and second group of finger two groups of all touching when grabbing workpiece point spacing d.

Preferably, described step C specifically comprises: sub-step C1: the span of set condition variable, and obtains the value vector of each variable in state variable according to span, using the state value of the element in value vector as state variable; Sub-step C2: the state value based on state variable obtains corresponding output variable output valve.

Preferably, described sub-step C1 specifically comprises: choose the central point abscissa x referring to spacing l, projection convex polygon _cinterval with the projection convex polygon anglec of rotation is the span of θ, each state variable of decile, obtains the value vector of above-mentioned three state variables and is designated as L=[l ₁l ₂l _ml _m], X=[x _c1x _c2x _cnx _cN] and ψ=[θ ₁θ ₂θ _pθ _p], the element in value vector L, X and ψ is the state value of state variable, any one group of (l wherein _m, x _cn, θ _p) one group of state value of equal corresponding states variablees.

Preferably, described sub-step C2 specifically comprises: son is C2a step by step: for arbitrary group of state value of state variable, utilizes iterative method to calculate two groups of finger spacing corresponding to this group state value; Son is C2b step by step: judge that two groups after the iteration whether satisfied first group of finger of finger spacing and second group of finger all touch and treat grabbing workpiece, if meet, then it can be used as this group state value (l _m, x _cn, θ _p) corresponding output variable output valve d _i, and perform son C2c step by step, otherwise, return son C2a continuation iterative computation step by step; Son is C2c step by step: repeat son C2a and C2b step by step, obtains all M × N × P group state value (l _m, x _cn, θ _p) corresponding output variable output valve d=[d ₁d ₂d _id _i], wherein I=M × N × P.

Preferably, described son step by step C2a specifically comprise: to arbitrary group of state value (l _m, x _cn, θ _p), setting step-length h and maximum step number K, then to kth+1 step, calculate two groups of finger space D as follows:

D(k+1)＝D(k)-ηh

Wherein, η is descent coefficient, and 0 < η < 1, initial value D (0) is greater than the two groups of finger spacing first group of finger and second group pointed and all touch when grabbing workpiece.

Preferably, described step D specifically comprises: sub-step D1: choose a variable in state variable as fixed variable, for any one state value of this fixed variable, make the functional image between the state value of its dependent variable of the one group of state value jointly forming state vector with this state value and corresponding output variable output valve; Sub-step D2: the environment domain of attraction finding out functional image, and the environment domain of attraction minimum point obtaining this functional image; Sub-step D3: for each state value of this fixed variable, repeat sub-step D1 and D2, obtains the environment domain of attraction of constraint function.

Preferably, this sub-step D1 specifically comprises: choose state variable (l, x _c, θ) in finger spacing l be fixed variable, for any one state value l of this fixed variable _m, make N × P group state value (l _m, x _cn, θ _p) and and this N × P group state value (l _m, x _cn, θ _p) corresponding output variable output valve d _ibetween functional image, wherein n ∈ { 1, N}, p ∈ { 1, P}, i ∈ { 1, N × P}; This sub-step D2 specifically comprises: based on the definition of above-mentioned constraint function environment domain of attraction, find out the environment domain of attraction in functional image, the minimum point (X of record environment domain of attraction _m, ψ _m, d _m), if there is multiple environment domain of attraction, then choose the environment domain of attraction minimum point of minimum environment domain of attraction minimum point as this functional image; This sub-step D3 specifically comprises: to each state value l referring to spacing l _m, { 1, M} performs sub-step D1 and D2, finds out the minimum point (X of the environment domain of attraction of its correspondence m ∈ _m, ψ _m, d _m), { 1, M} selects min{d to m ∈ _mand be designated as d ^*, and d ^*corresponding environment domain of attraction is designated as Ω ₀, by Ω ₀as the environment domain of attraction of constraint function, d ^*the state value of corresponding state variable is designated as (l ^*, x _c ^*, θ ^*).

Preferably, described step e specifically comprises: sub-step E1: using the state value of state variable corresponding for the environment domain of attraction of constraint function as crawl dbjective state, and region corresponding to the environment domain of attraction of constraint function is as initially capturing feasible zone; Sub-step E2: any one state in initial crawl feasible zone of choosing, as original state point, utilizes Newton method to plan state transition path; Sub-step E3: using each state in state transition path as the mechanical hand captured in space and the state treating grabbing workpiece, thus the mechanical hand obtained based on environment domain of attraction captures strategy.

Preferably, this sub-step E1 specifically comprises: by the environment domain of attraction Ω of constraint function ₀the state value X of corresponding state variable ^*=(l ^*, x _c ^*, θ ^*) as capturing dbjective state, the environment domain of attraction Ω of constraint function ₀corresponding region captures feasible zone Ω as initial ^*; This sub-step E2 specifically comprises: choose initial crawl feasible zone Ω ^*in any one state X ⁰=(l ⁰, x _c ⁰, θ ⁰) as original state point, utilize Newton method to plan state transition path, the following formulae discovery of state transition path:

$X^{n+1}=X^n-{\&lsqb;Hf\left(X^n\right)\&rsqb;}^{-1}\&dtri;f\left(X^n\right)$

Wherein, H is hessian matrix.

(3) beneficial effect

As can be seen from technique scheme, the present invention has following beneficial effect:

(1) this method is applied widely, all can use for the mechanical hand of various different types of structure and the crawl object of different size;

(2) the crawl strategy of mechanical hand can be cooked up quickly and accurately;

(3) do not rely on sensor, grasp stability is good, and cost is low.

Accompanying drawing explanation

Fig. 1 is the flow chart of the crawl of the mechanical hand based on the environment domain of attraction strategic planning method of the embodiment of the present invention;

Fig. 2 is four finger projections in the horizontal plane and the schematic diagram of coordinate system of robot arm;

Fig. 3 treats the perspective view of grabbing workpiece under coordinate system of robot arm.

Detailed description of the invention

For making the object, technical solutions and advantages of the present invention clearly understand, below in conjunction with specific embodiment, and with reference to accompanying drawing, the present invention is described in more detail.It should be noted that, in accompanying drawing or description describe, similar or identical part all uses identical figure number.The implementation not illustrating in accompanying drawing or describe is form known to a person of ordinary skill in the art in art.In addition, although herein can providing package containing the demonstration of the parameter of particular value, should be appreciated that, parameter without the need to definitely equaling corresponding value, but can be similar to corresponding value in acceptable error margin or design constraint.The direction term mentioned in embodiment, such as " on ", D score, "front", "rear", "left", "right" etc., be only the direction with reference to accompanying drawing.Therefore, the direction term of use is used to illustrate and is not used for limiting the scope of the invention.

Present invention employs a kind of mechanical hand based on environment domain of attraction and capture strategic planning method, it can for mechanical hand and the parameter capturing object, by building constraint function, and the crawl strategy of mechanical hand is cooked up based on environment domain of attraction, this method is applied widely, all can adapt to for the mechanical hand of various different types of structure and the crawl object of different size, the crawl strategy of mechanical hand can be cooked up quickly and accurately, do not rely on sensor, grasp stability is good, and cost is low.

The mechanical hand based on environment domain of attraction of the embodiment of the present invention captures strategic planning method, and see Fig. 1, it comprises:

Steps A: according to type and the structure treating grabbing workpiece of mechanical hand, extract the model parameter of mechanical hand and treat the model parameter of grabbing workpiece.

Steps A specifically comprises:

Sub-step A1: set up coordinate system of robot arm in the projection of horizontal plane based on mechanical hand.

Preferably, this sub-step A1 specifically comprises: see Fig. 2, Industrial robot arm is referred to for four, its four finger projections are in the horizontal plane respectively the first finger subpoint 21, first finger subpoint 22, 3rd finger subpoint 23 and the 4th finger subpoint 24, wherein, two finger composition first group fingers of the first finger subpoint 21 and the first finger subpoint 22 correspondence, the finger of the 3rd finger subpoint 23 and the 4th finger subpoint 24 correspondence forms second group of finger, two subpoints 21, mid point between 22 and second group of two subpoint 23 pointed, distance between mid point between 24 is two groups of finger space D, with the mid point O of two groups of finger space D for the origin of coordinates, with the straight line at two groups of finger space D places for X-axis, set up right-handed coordinate system XOY, using this right-handed coordinate system XOY as coordinate system of robot arm.

Sub-step A2: based on coordinate system of robot arm, extracts the model parameter of mechanical hand.

Preferably, this sub-step A2 specifically comprises: between first, second finger subpoint of first group of finger first refer to that spacing points with second group first, second point second between subpoint and refer to that spacing is equal, be l.

Line between first, second finger subpoint of first group of finger and the counterclockwise angle of Y-axis are the first angle α, line between 3rd, the 4th finger subpoint of second group of finger and the counterclockwise angle of Y-axis are the second angle β, by (l, D, α, β) as the mechanical hand model parameter extracted.

Sub-step A3: based on coordinate system of robot arm, extracts the model parameter treating grabbing workpiece.

Preferably, this sub-step A3 specifically comprises: treat that grabbing workpiece is irregular prismatic object workpiece, it is a convex polygon in the projection of horizontal plane, as shown in Figure 3, in coordinate system of robot arm, the summit of this convex polygon is respectively the first subpoint 31, second subpoint 32, the 3rd subpoint 33, the 4th subpoint 34 and the 5th subpoint 35, and its coordinate is respectively (x ₁, y ₁), (x ₂, y ₂), (x ₃, y ₃), (x ₄, y ₄), (x ₅, y ₅);

Central point C (the x of convex polygon _c, y _c) coordinate x _c, y _c:

$x_c=\frac{1}{5}\underset{i=1}{\overset{5}{\&Sigma;}}{x}_i,y_c=\frac{1}{5}\underset{i=1}{\overset{5}{\&Sigma;}}{y}_i$

The convex polygon anglec of rotation is θ, by (x _c, y _c, θ) and treat grabbing workpiece model parameter as what extract.

Wherein, above-mentioned mechanical hand and treat that the model parameter of grabbing workpiece can be inputted by user or revise on initial value.

Step B: based on mechanical hand and treat that the model parameter of grabbing workpiece chooses state variable and the output variable of the constraint function that there is environment domain of attraction.

Step B specifically comprises:

Sub-step B1: the environment domain of attraction of definition constraint function.

Preferably, this sub-step B1 specifically comprises: the environment domain of attraction of constraint function refers to, select suitable state variable, make to there is a state set Ω in constraint function, this state set Ω meets the irrelevant input U of existence one and state, makes nonlinear system free position from this state set Ω converge to a less state set Ω ₀.

Specifically, the environment domain of attraction of constraint function is defined as: for a nonlinear system wherein, X is the state of nonlinear system, and u is the input of nonlinear system, and a given function g (X), if exist system mode X ₀with real number ε, for meeting || X-X ₀|| the state X of < ε, has:

(1)g(X)＞g(X ₀)，X≠X ₀；

(2)g(X)＝g(X ₀)，X＝X ₀；

(3) g (X) has continuous offset derivative about X;

(4)dg(X)/dt＜0；

Then nonlinear system is at X ₀epsilon neighborhood in stable, this epsilon neighborhood is called environment domain of attraction, X ₀be the stable state of this environment domain of attraction, wherein, g (X) is generally energy function.

Sub-step B2: based on mechanical hand and treat that the model parameter of grabbing workpiece chooses the state variable of constraint function.

Preferably, this sub-step B2 specifically comprises: choose the central point abscissa x that first, second refers to spacing l, convex polygon _cwith the convex polygon anglec of rotation be θ as state variable X, i.e. state variable X=(l, x _c, θ).

Sub-step B3: based on mechanical hand and treat that the model parameter of grabbing workpiece chooses the output variable of constraint function.

Preferably, this sub-step B3 specifically comprises: first group of finger and second group of finger are all touched two groups of finger spacing d when the grabbing workpiece as output variable, and namely output variable d is that mechanical hand two groups of grabbing when grabbing workpiece point spacing.

Thus the model parameter of mechanical hand is as the state variable of constraint function, and the environment domain of attraction of constraint function is the higher-dimension domain of attraction that dimension is greater than 3, the stable seized condition of the corresponding mechanical hand of minimum point of domain of attraction.

Step C: the span based on state variable obtains the state value of state variable, is obtained the output valve of corresponding output variable by the state value of state variable.

Step C specifically comprises:

Sub-step C1: the span of set condition variable, and the value vector obtaining each variable in state variable according to span, using the state value of the element in value vector as state variable.

Preferably, this sub-step C1 specifically comprises: choose the central point abscissa x that first, second refers to spacing l, convex polygon _cwith the interval that the convex polygon anglec of rotation is the span of θ, each variable of decile, the value vector obtaining three variablees is designated as L, X and ψ, is respectively L=[l ₁l ₂l _ml _m], X=[x _c1x _c2x _cnx _cN], ψ=[θ ₁θ ₂θ _pθ _p], the element in value vector L, X and ψ is the state value of state variable, any one group of (l wherein _m, x _cn, θ _p) one group of state value of equal corresponding states variablees.Wherein, first, second refers to the central point abscissa x of spacing l, convex polygon _ccan set according to embody rule situation with the span that the convex polygon anglec of rotation is θ, namely according to the mechanical hand in practical application scene and treat grabbing workpiece to set.

Sub-step C2: the state value based on state variable obtains corresponding output variable output valve.

Sub-step C2 specifically comprises:

Son is C2a step by step: for arbitrary group of state value of state variable, utilizes iterative method to calculate two groups of finger spacing corresponding to this group state value.

Preferably, son step by step C2a specifically comprise: to arbitrary group of state value (l _m, x _cn, θ _p), setting step-length h and maximum step number K, then to kth+1 step, calculate two groups of finger space D as follows:

D(k+1)＝D(k)-ηh

Wherein η is descent coefficient, 0 < η < 1, initial value D (0) and step-length h, maximum step number K can according to practical application scene settings, and initial value D (0) should be greater than the two groups of finger spacing first group of finger and second group pointed and all touch when grabbing workpiece.

Son is C2b step by step: judge that two groups after the iteration whether satisfied first group of finger of finger spacing and second group of finger all touch and treat grabbing workpiece, if meet, point spacing, namely as this group state value (l for two groups that then it can be used as mechanical hand to grab when grabbing workpiece _m, x _cn, θ _p) corresponding output variable output valve d _i, and perform son C2c step by step, otherwise, return son C2a continuation iterative computation step by step.

Son is C2c step by step: repeat son C2a and C2b step by step, obtains all M × N × P group state value (l _m, x _cn, θ _p) corresponding output variable output valve d=[d ₁d ₂d _id _i], wherein I=M × N × P.

Step D: based on the environment domain of attraction of the state value of state variable and the output valve determination constraint function of output variable;

Step D specifically comprises:

Sub-step D1: choose a variable in state variable as fixed variable, for any one state value of this fixed variable, make the functional image between the state value of its dependent variable of the one group of state value jointly forming state vector with this state value and corresponding output variable output valve.

Preferably, this sub-step D1 specifically comprises: choose state variable (l, x _c, θ) in first, second refer to that spacing l is fixed variable, refer to any one state value l of spacing l for first, second _m, make N × P group state value (l _m, x _cn, θ _p) and and this N × P group state value (l _m, x _cn, θ _p) corresponding output variable output valve d _ibetween functional image, wherein n ∈ { 1, N}, p ∈ { 1, P}, i ∈ { 1, N × P}.

Sub-step D2: the environment domain of attraction finding out functional image, and the environment domain of attraction minimum point obtaining this functional image.

Preferably, this sub-step D2 specifically comprises: based on the definition of above-mentioned constraint function environment domain of attraction, find out the environment domain of attraction in functional image, the minimum point (X of record environment domain of attraction _m, ψ _m, d _m).If there is multiple environment domain of attraction, then choose the environment domain of attraction minimum point of minimum environment domain of attraction minimum point as this functional image.

Sub-step D3: for each state value of this fixed variable, repeat sub-step D1 and D2, obtains the environment domain of attraction of constraint function.

Preferably, this sub-step D3 specifically comprises: each state value L first, second being referred to spacing l _m, { 1, M} performs sub-step D1 and D2, finds out the minimum point (X of the environment domain of attraction of its correspondence m ∈ _m, ψ _m, d _m), { 1, M} selects min{d to m ∈ _mand be designated as d ^*, and d ^*corresponding environment domain of attraction is designated as Ω ₀, by Ω ₀as the environment domain of attraction of constraint function, d ^*the state value of corresponding state variable is designated as (l ^*, x _c ^*, θ ^*).

Step e: the environment domain of attraction planning mechanical hand based on constraint function captures strategy.

Step e specifically comprises:

Sub-step E1: by the environment domain of attraction Ω of constraint function ₀the state value X of corresponding state variable ^*=(l ^*, x _c ^*, θ ^*) as capturing dbjective state, the environment domain of attraction Ω of constraint function ₀corresponding region captures feasible zone Ω as initial ^*.

Sub-step E2: choose initial crawl feasible zone Ω ^*in any one state X ⁰=(l ⁰, x _c ⁰, θ ⁰) as original state point, utilize Newton method to plan state transition path, the following formulae discovery of state transition path:

$X^{n+1}=X^n-{\&lsqb;Hf\left(X^n\right)\&rsqb;}^{-1}\&dtri;f\left(X^n\right)$

Wherein H is hessian matrix.

Sub-step E3: using each state in state transition path as the mechanical hand captured in space and the state treating grabbing workpiece, thus the mechanical hand obtained based on environment domain of attraction captures strategy.

So far, by reference to the accompanying drawings the present embodiment has been described in detail.Describe according to above, those skilled in the art should capture strategic planning method to the mechanical hand based on environment domain of attraction of the present invention had and be clearly familiar with.

It should be noted that, in accompanying drawing or description text, the implementation not illustrating or describe, is form known to a person of ordinary skill in the art in art, is not described in detail.In addition, the above-mentioned definition to each element is not limited in the various modes mentioned in embodiment, and those of ordinary skill in the art can change simply it or replace, such as:

(1) mechanical hand model parameter and treat that the model parameter of grabbing workpiece can also adopt other parameters;

(2) additive method can also be adopted to realize state transition path;

(3) the direction term mentioned in embodiment, such as " on ", D score, "front", "rear", "left", "right" etc., be only the direction with reference to accompanying drawing, be not used for limiting the scope of the invention;

(4) above-described embodiment can based on design and the consideration of reliability, and being mixed with each other collocation uses or uses with other embodiment mix and match, and the technical characteristic namely in different embodiment can freely form more embodiment.

Above-described specific embodiment; object of the present invention, technical scheme and beneficial effect are further described; be understood that; the foregoing is only specific embodiments of the invention; be not limited to the present invention; within the spirit and principles in the present invention all, any amendment made, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (10)

1. the mechanical hand based on environment domain of attraction captures a strategic planning method, it is characterized in that, comprising:

Steps A: according to type and the structure treating grabbing workpiece of mechanical hand, extract the model parameter of mechanical hand and treat the model parameter of grabbing workpiece;

Step B: based on mechanical hand and treat that the model parameter of grabbing workpiece chooses state variable and the output variable of the constraint function that there is environment domain of attraction;

Step C: the span based on state variable obtains the state value of state variable, is obtained the output valve of corresponding output variable by the state value of state variable;

Step D: based on the environment domain of attraction of the state value of state variable and the output valve determination constraint function of output variable; And

Step e: the environment domain of attraction planning mechanical hand based on constraint function captures strategy.

2. mechanical hand as claimed in claim 1 captures strategic planning method, and it is characterized in that, described step B specifically comprises:

Sub-step B1: the environment domain of attraction of definition constraint function;

Sub-step B2: based on mechanical hand and treat that the model parameter of grabbing workpiece chooses the state variable of constraint function, wherein, the element of this state variable X comprises: the finger spacing l of mechanical hand, the projection convex polygon central point abscissa x of grabbing workpiece at horizontal plane is treated _cwith projection convex polygon anglec of rotation θ; And

Sub-step B3: based on mechanical hand and treat that the model parameter of grabbing workpiece chooses the output variable of constraint function, wherein, this output variable is that first group of finger of mechanical hand and second group of finger two groups of all touching when grabbing workpiece point spacing d.

3. mechanical hand as claimed in claim 2 captures strategic planning method, and it is characterized in that, described step C specifically comprises:

Sub-step C1: the span of set condition variable, and the value vector obtaining each variable in state variable according to span, using the state value of the element in value vector as state variable; And

Sub-step C2: the state value based on state variable obtains corresponding output variable output valve.

4. mechanical hand as claimed in claim 3 captures strategic planning method, and it is characterized in that, described sub-step C1 specifically comprises:

Choose the central point abscissa x referring to spacing l, projection convex polygon _cinterval with the projection convex polygon anglec of rotation is the span of θ, each state variable of decile, obtains the value vector of above-mentioned three state variables and is designated as L=[l ₁l ₂l _ml _m], X=[x _c1x _c2x _cnx _cN] and ψ=[θ ₁θ ₂θ _pθ _p], the element in value vector L, X and ψ is the state value of state variable, any one group of (l wherein _m, x _cn, θ _p) one group of state value of equal corresponding states variablees.

5. mechanical hand as claimed in claim 4 captures strategic planning method, and it is characterized in that, described sub-step C2 specifically comprises:

Son is C2a step by step: for arbitrary group of state value of state variable, utilizes iterative method to calculate two groups of finger spacing corresponding to this group state value;

Son is C2b step by step: judge that two groups after the iteration whether satisfied first group of finger of finger spacing and second group of finger all touch and treat grabbing workpiece, if meet, then it can be used as this group state value (l _m, x _cn, θ _p) corresponding output variable output valve d _i, and perform son C2c step by step, otherwise, return son C2a continuation iterative computation step by step; And

Son is C2c step by step: repeat son C2a and C2b step by step, obtains all M × N × P group state value (l _m, x _cn, θ _p) corresponding output variable output valve d=[d ₁d ₂d _id _i], wherein I=M × N × P.

6. mechanical hand as claimed in claim 5 captures strategic planning method, and it is characterized in that, described son step by step C2a specifically comprises:

To arbitrary group of state value (l _m, x _cn, θ _p), setting step-length h and maximum step number K, then to kth+1 step, calculate two groups of finger space D as follows:

D(k+1)＝D(k)-ηh

Wherein, η is descent coefficient, and 0 < η < 1, initial value D (0) is greater than the two groups of finger spacing first group of finger and second group pointed and all touch when grabbing workpiece.

7. mechanical hand as claimed in claim 6 captures strategic planning method, and it is characterized in that, described step D specifically comprises:

Sub-step D1: choose a variable in state variable as fixed variable, for any one state value of this fixed variable, make the functional image between the state value of its dependent variable of the one group of state value jointly forming state vector with this state value and corresponding output variable output valve;

Sub-step D2: the environment domain of attraction finding out functional image, and the environment domain of attraction minimum point obtaining this functional image; And

Sub-step D3: for each state value of this fixed variable, repeat sub-step D1 and D2, obtains the environment domain of attraction of constraint function.

8. mechanical hand as claimed in claim 7 captures strategic planning method, it is characterized in that,

This sub-step D1 specifically comprises: choose state variable (l, x _c, θ) in finger spacing l be fixed variable, for any one state value l of this fixed variable _m, make N × P group state value (l _m, x _cn, θ _p) and and this N × P group state value (l _m, x _cn, θ _p) corresponding output variable output valve d _ibetween functional image, wherein n ∈ { 1, N}, p ∈ { 1, P}, i ∈ { 1, N × P};

This sub-step D2 specifically comprises: based on the definition of above-mentioned constraint function environment domain of attraction, find out the environment domain of attraction in functional image, the minimum point (X of record environment domain of attraction _m, ψ _m, d _m), if there is multiple environment domain of attraction, then choose the environment domain of attraction minimum point of minimum environment domain of attraction minimum point as this functional image;

This sub-step D3 specifically comprises: to each state value l referring to spacing l _m, { 1, M} performs sub-step D1 and D2, finds out the minimum point (X of the environment domain of attraction of its correspondence m ∈ _m, ψ _m, d _m), { 1, M} selects min{d to m ∈ _mand be designated as d ^*, and d ^*corresponding environment domain of attraction is designated as Ω ₀, by Ω ₀as the environment domain of attraction of constraint function, d ^*the state value of corresponding state variable is designated as (l ^*, x _c ^*, θ ^*).

9. mechanical hand as claimed in claim 8 captures strategic planning method, and it is characterized in that, described step e specifically comprises:

Sub-step E1: using the state value of state variable corresponding for the environment domain of attraction of constraint function as crawl dbjective state, region corresponding to the environment domain of attraction of constraint function is as initially capturing feasible zone;

Sub-step E2: any one state in initial crawl feasible zone of choosing, as original state point, utilizes Newton method to plan state transition path; And

Sub-step E3: using each state in state transition path as the mechanical hand captured in space and the state treating grabbing workpiece, thus the mechanical hand obtained based on environment domain of attraction captures strategy.

10. mechanical hand as claimed in claim 9 captures strategic planning method, it is characterized in that,

This sub-step E1 specifically comprises: by the environment domain of attraction Ω of constraint function ₀the state value X of corresponding state variable ^*=(l ^*, x _c ^*, θ ^*) as capturing dbjective state, the environment domain of attraction Ω of constraint function ₀corresponding region captures feasible zone Ω as initial ^*;

This sub-step E2 specifically comprises: choose initial crawl feasible zone Ω ^*in any one state X ⁰=(l ⁰, x _c ⁰, θ ⁰) as original state point, utilize Newton method to plan state transition path, the following formulae discovery of state transition path:

$X^{n+1}=X^n-{\&lsqb;Hf\left(X^n\right)\&rsqb;}^{-1}\&dtri;f\left(X^n\right)$

Wherein, H is hessian matrix.