CN103706568A - System and method for machine vision-based robot sorting - Google Patents

Abstract

The invention discloses a system and a method for machine vision-based robot sorting. The system comprises a CCD digital camera, a camera lens, a light source, a six-axle joint robot main body, an electrical control cabinet and a vacuum sucker. The CCD digital camera is connected to an industrial computer by a switch. The six-axle joint robot main body is connected to the electrical control cabinet. The electrical control cabinet accesses the switch. The vacuum sucker is rigidly fixed to the tail end of the six-axle joint robot main body. The camera unit is used for making a photo of an object to be sorted, collecting data and transmitting the data to the industrial computer by the switch. The industrial computer is used for treating the acquired photo of the object to be sorted, carrying out accurate positioning and then transmitting a control signal to the electrical control cabinet by the switch. The electrical control cabinet is used for controlling the six-axle joint robot main body to carry out corresponding sorting processes according to the received control signal. The system and the method improve work efficiency, reduce operation workers and reduce a production cost in sorting.

Description

Robot sorting system and method based on machine vision

Technical field

The present invention relates to intelligent robot technology field, particularly relate to a kind of robot sorting system and method based on machine vision.

Background technology

Sorting operation is an important step on most of streamlined production lines.Robot sorting based on machine vision is compared with manual sorting's operation, not only efficient, accurate, and at aspects such as quality assurance, medical supports, has the irreplaceable advantage of manual work; Compare with traditional mechanical sorting operation, the robot sorting based on machine vision has wide accommodation, can convert at any time the advantage of manipulating object and conversion sorting operation.Robot sorting technology is the combination of Robotics and machine vision technique, Japan and more American-European developed countries, quite universal in production field applied robot sortings such as machinery, food, medicine, cosmetics, and the robot sorting system that China really puts into practice is at present almost also blank.According to the market demand situation of current China and correlation technique basis, research, development and application robot sorting technology have very important meaning.

Domestic also some people has been studied and has obtained certain achievement for the robot sorting system based on machine vision, the most architecture modes of the sorting system of development are for being used the existing industrial robot system of well-known producer at present, pass through serial ports, the logical information mode such as Ethernet, carry out data communication with NI Vision Builder for Automated Inspection, realize the integrated of sorting system, and such system architecture make NI Vision Builder for Automated Inspection and robot system relatively independent, very flexible, open low, dependence is high, inconvenient operation, portable poor, in carrying out these two system communication docking operations, difficulty is large simultaneously, R&D cycle is long.

Summary of the invention

The object of the present invention is to provide a kind of robot sorting system and method based on machine vision, in sorting operation, increase work efficiency, reduce operator, reduce production costs.

The technical solution that realizes the object of the invention is: a kind of robot sorting system based on machine vision, comprise control module, camera unit, robot cell and tool unit, wherein control module comprises industrial computer, switch, camera unit comprises CCD digital camera, camera lens and light source, robot cell comprises six axle articulated robot body and electrical control cabinets, and tool unit comprises vacuum cup;

CCD digital camera is connected with industrial computer by switch, six axle articulated robot bodies are connected with electrical control cabinet, electrical control cabinet access switch, vacuum cup is rigidly fixed in the end of six axle articulated robot bodies, and coaxial with the forearm of six axle articulated robot bodies;

Camera unit is sorted picture shooting, the data acquisition of article, and the data of the picture of shooting and collection are transferred to industrial computer by switch; Industrial computer carries out, behind image processing and accurate location, by switch, to electrical control cabinet, transmitting control signal to the item pictures that is sorted collecting; Electrical control cabinet is handled six axle articulated robot bodies according to the control signal of receiving and is carried out corresponding sorting operation.

A robot method for sorting based on machine vision, comprises the following steps:

Step 1, determines the target item being sorted, and will be sorted feature selecting or the input industrial computer of article;

Step 2, the angle of adjusting camera lens and light source makes the two aim at article to be sorted rest area, and CCD digital camera is sorted picture shooting, the data acquisition of article, and the data of the picture of shooting and collection are transferred to industrial computer by switch;

Step 3, the image processing module of industrial computer carries out image processing to being sorted item pictures, judges in image, whether there is target item: if do not had, return to step 2; If had, all target items are extracted, and enter next step;

Step 4, the robot control module of industrial computer carries out after hand and eye calibrating CCD digital camera and six axle articulated robot bodies, according to the transformational relation between image coordinate system, camera coordinates system and robot coordinate system, determine that each target item, at robot coordinate system's position and attitude information, is stored into the position of each target item obtaining and attitude information in file;

Step 5, target item rest area is sorted in the target location that definite target item sorting is stacked, and ask for each target item position and the contrary solution of attitude motion by robot control module, carry out the movement locus planning of six axle articulated robot bodies;

Step 6, by switch, to electrical control cabinet, transmit control signal, electrical control cabinet is handled six axle articulated robot bodies according to the control signal of receiving and is carried out corresponding sorting operation, and guiding six axle articulated robot bodies complete absorption, carrying and the stacking of all target items.

Compared with prior art, its remarkable advantage is in the present invention: (1) possesses that robot controls function and image processing function flexibility is good simultaneously, open high; (2) in sorting operation, increase work efficiency, reduce operator; (3) reduce production costs, and portable strong, wide adaptability.

Below in conjunction with accompanying drawing, the present invention is described in further detail.

Accompanying drawing explanation

Fig. 1 is the robot sorting system structure chart that the present invention is based on machine vision.

Fig. 2 is the robot method for sorting flow chart that the present invention is based on machine vision.

Fig. 3 is the conversion schematic diagram between image coordinate system of the present invention, camera coordinates system and robot coordinate system.

Fig. 4 carries out image processing process schematic diagram to being sorted item pictures in embodiment 1, wherein (a) is original image, (b) being the image after Threshold segmentation, is (c) image after connected region extraction, (d) is the image after feature extraction.

Label in Fig. 1: 1 is CCD digital camera, 2 is camera lens, and 3 is light source, and 4 is camera fixing support, 5 is vacuum cup, and 6 is article to be sorted rest areas, and 7 is sorting target item rest area, and 8 is six axle articulated robot bodies, 9 is electrical control cabinet, and 10 is switch, and 11 is industrial computer.

The specific embodiment

Below in conjunction with drawings and the specific embodiments, the present invention is described in further detail.

In conjunction with Fig. 1, the present invention is based on the robot sorting system of machine vision, comprise control module, camera unit, robot cell and tool unit, wherein control module comprises industrial computer 11, switch 10, camera unit comprises CCD digital camera 1, camera lens 2 and light source 3, robot cell comprises six axle articulated robot body 8 and electrical control cabinets 9, and tool unit comprises vacuum cup 5; CCD digital camera 1 is connected with industrial computer 11 by switch 10, six axle articulated robot bodies 8 are connected with electrical control cabinet 9, electrical control cabinet 9 access switch 10, vacuum cup 5 is rigidly fixed in the end of six axle articulated robot bodies 8, and coaxial with the forearm of six axle articulated robot bodies 8;

The CCD digital camera 1 of described camera unit, camera lens 2 and light source 3 are fixed on camera fixing support 4 in turn, and light source 3 is all adjustable with the coaxial setting of camera lens 2 and the two angle.Described tool unit also comprises air compressor machine, magnetic valve and vacuum generator, is sorted article draws for 5 pairs of vacuum cups.Described industrial computer 11 comprises image processing module and robot control module.

Camera unit is sorted picture shooting, the data acquisition of article, and the data of the picture of shooting and collection are transferred to industrial computer 11 by switch 10; The item pictures that is sorted that 11 pairs of industrial computers collect carries out, behind image processing and accurate location, by switch 10, to electrical control cabinet 9, transmitting control signal; Electrical control cabinet 9 is handled six axle articulated robot bodies 8 according to the control signal of receiving and is carried out corresponding sorting operation.

In conjunction with Fig. 2, the present invention is based on the robot method for sorting of machine vision, first utilize the CCD digital camera 1 of camera unit, the article to be sorted that camera lens 2 and light source 3 are treated segregating articles rest area 6 carry out IMAQ, via switch 10, be transferred to industrial computer 11, industrial computer 11 obtains the exact position of target item by image processing module, recycling robot control module asks for each target item position and attitude motion is learned contrary solution, carry out automatic orbit planning, thereby the vacuum cup 5 of six axle articulated robot bodies 8 of guided robot unit and electrical control cabinet 9 and tool unit completes the accurate absorption of all target items, carrying, stack action, make it be deposited in target item rest area 7, concrete steps are as follows:

Step 1, determines the target item being sorted, and will be sorted feature selecting or the input industrial computer 11 of article; Described will be sorted feature selecting or the input industrial computer 11 of article, be specially:

(1.1) if there is no required target item in database filelist, in interactive interface, target item feature is inputted and stored, form database file;

(1.2) if there is required target item in database filelist, directly selects and determine.

Step 2, the angle of adjusting camera lens 2 and light source 3 makes the two aim at article to be sorted rest area 6, CCD digital camera 1 is sorted picture shooting, the data acquisition of article, and the data of the picture of shooting and collection are transferred to industrial computer 11 by switch 10;

Step 3, the image processing module of industrial computer 11 carries out image processing to being sorted item pictures, judges in image, whether there is target item: if do not had, return to step 2; If had, all target items are extracted, and enter next step; The step that wherein image is processed is as follows:

(3.1) to being sorted item pictures, carry out gray processing; Gray value linear scale convergent-divergent formula is as follows:

f(g)＝min(max(ag+b+0.5,0),2 ^b-1)

In formula, a is contrast, and b is brightness, and g is gray value;

(3.2) image after gray processing is carried out to Threshold segmentation; To in image, in the gray value of appointment tonal range, all choose in output area S:

S＝{(r,c)∈R|g _min≤f _r,c≤g _max}

F _r,cfor coordinate in image is the gray value of (r, c) pixel; g _minfor the minimum gradation value for Threshold segmentation; g _max-be the maximum gradation value for Threshold segmentation, R is image-region;

(3.3) image after Threshold segmentation is carried out to connected region extraction;

(3.4) image after connected region extraction is carried out to feature extraction and matching; Provincial characteristics selects area M to be:

$M=\left|R\right|=\underset{(r,c)\∈R}{\mathrm{\Σ}}1$

M is the area in region, | R| is counting in region;

(3.5) according to the image after feature extraction and matching, judged whether target item, the elemental area value of given target item interval (m, n), institute's area that obtains is target item within this interval.

Step 4, the robot control module of industrial computer 11 carries out after hand and eye calibrating CCD digital camera 1 and six axle articulated robot bodies 8, according to the transformational relation between image coordinate system, camera coordinates system and robot coordinate system, determine that each target item, at robot coordinate system's position and attitude information, is stored into the position of each target item obtaining and attitude information in file; Detailed process is as follows:

(4.1) hand and eye calibrating

Known calibration plate is fixedly mounted on to the tool tip of six axle articulated robot bodies 8, robot motion makes scaling board move to CCD digital camera 1 within sweep of the eye, obtain image and image is processed, obtain scaling board at the pose of camera coordinates system, and record the now pose of robot tool coordinate system end; Mobile robot, repeats above step and obtains N group data for N time, makes scaling board cover the regional in the camera visual field;

(4.2) Coordinate Conversion

N group data are processed, wherein tool coordinates system and robot coordinate system's transformational relation ^baseh _toolknown, the transformational relation of scaling board coordinate system and camera coordinates system ^camh _calknown:

^camH _cal= ^camH _base ^baseH _tool ^toolH _cal

According to above formula, by the processing to N group data, obtain the transformational relation between camera coordinates and robot coordinate system ^camh _base, the transformational relation between scaling board coordinate system and coordinate system ^toolh _calfor intermediate variable, as shown in Figure 3.

Step 5, target item rest area 7 is sorted in the target location that definite target item sorting is stacked, and ask for each target item position and the contrary solution of attitude motion by robot control module, carry out the movement locus planning of six axle articulated robot bodies 8; The mode of described definite sorting target item rest area 7 comprises two kinds: utilize robot control module's crawl teaching target location and store, or off-line is manually inputted target location.Describedly ask for each target item position and attitude motion to learn contrary detailed process of separating as follows:

(5.1) in known six axle articulated robot bodies 8, each axle link parameters is as follows:

L ₆be that the first axle connecting rod axle center is that initial point is to the second distance of axle turning cylinder axle center in X-direction, l ₁be that the first axle connecting rod axle center is that initial point is to the second distance of axle turning cylinder axle center in Z-direction; l ₂be the length of the second axle connecting rod; l ₃be the 3rd axle connecting rod axle center to the four distances of axle connecting rod axle center in Z-direction; l ₄be the 3rd axle axle center to four shaft ends the distance in X-direction; l ₅be the length of the 5th axle connecting rod; l ₇be the first axle axle center to the three distances of axle connecting rod axle center in Y-axis, l in Gai robot ₇be 0;

Distal point pose is (x, y, z, A, B, C), and wherein (x, y, z) is the coordinate position under robot coordinate system, and (A, B, C) is the attitude angle under end robot coordinate system;

Solve the angle θ that in six axle articulated robot bodies 8, each axle rotates ₁, θ ₂, θ ₃, θ ₄, θ ₅, θ ₆, in following formula, c represents cos θ, s represents sin θ, the subscript of c, s represents corresponding angle, the double subscript of c, s represent corresponding angle and, i.e. c ₂₃represent cos (θ ₂+ θ ₃), s ₂₃represent sin (θ ₂+ θ ₃);

(5.2) solve θ ₁

According to robot end, put pose (x, y, z, A, B, C), can obtain the transition matrix T between robot coordinate system and tool coordinates system:

$T=\left[\begin{array}{cccc}{n}_x& o_x& a_x& x\\ n_y& o_y& a_y& y\\ n_z& o_z& a_{\mathrm{<figure-callout\; id="0"\; label="z\; z"\; filenames="HDA0000422563920000011.png"\; state="\{\{state\}\}">z</figure-callout>}}& z\\ 0& 0& 0& 1\end{array}\right]---\left(1\right)$

In formula,

$\left\{\begin{array}{c}{n}_x=c_B{c}_C\\ n_y=s_A{s}_B{c}_C-c_A{s}_C\\ n_z=c_A{s}_B{c}_C+s_A{s}_C\\ o_x=c_B{s}_C\\ o_y=s_A{s}_B{s}_C+c_A{c}_C\\ o_z=c_A{s}_B{s}_C-s_A{c}_C\\ a_x=-s_B\\ a_y=s_A{c}_B\\ a_z=c_A{c}_B\end{array}\right.---\left(2\right)$

$T_1^w=T_0^w*T_1^0=\left[\begin{array}{cccc}{c}_1& {-s}_1& 0& 0\\ s_1& c_1& 0& 0\\ 0& 0& 1& l_1\\ 0& 0& 0& 1\end{array}\right]---\left(3\right)$

In formula,

be the transformational relation between the first axis coordinate system and robot coordinate system, the like, w is robot coordinate system;

To formula (3) while premultiplication

:

${\left(T_1^w\right)}^{-1}*T=T*_3^2{}_2{}^{1}T*T_4^3*T_5^4*T_6^5*T_T^6---\left(4\right)$

$\left[\begin{array}{cccc}{c}_1& s_1& 0& 0\\ {-s}_1& c_1& 0& 0\\ 0& 0& 1& {-l}_1\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{cccc}{n}_x& o_x& a_x& x\\ n_y& o_y& a_y& y\\ n_z& o_z& a_{\mathrm{<figure-callout\; id="0"\; label="z\; z"\; filenames="HDA0000422563920000011.png"\; state="\{\{state\}\}">z</figure-callout>}}& z\\ 0& 0& 0& 1\end{array}\right]=T_T^1---\left(5\right)$

Order $T_T^1=\left[\begin{array}{cccc}{r}_{11}& r_{21}& r_{31}& q_{x1}\\ r_{12}& r_{22}& r_{32}& q_{y1}\\ r_{13}& r_{23}& r_{33}& q_{z1}\\ 0& 0& 0& 1\end{array}\right],$ In formula

r ₁₁＝c ₁n _x+s ₁n _y＝-c ₆(s ₅s ₂₃-c ₄c ₅c ₂₃)-s ₄c ₅s ₂₃ (6)

r ₂₁＝-s ₁n _x+c ₁n _y＝-c ₄c ₆-s ₄c ₅c ₆ (7)

r ₃₁＝n _z＝s ₄s ₆s ₂₃-c ₆(s ₅c ₂₃+c ₄c ₅s ₂₃) (8)

r ₁₂＝c ₁o _x+s ₁o _y＝s ₆(s ₅s ₂₃-c ₄c ₅c ₂₃)-s ₄c ₆c ₂₃ (9)

r ₂₂＝-s ₁o _x+c ₁o _y＝c ₅s ₄s ₆-c ₄c ₆ (10)

r ₃₂＝o _z＝s ₆(s ₅c ₂₃+c ₄c ₅s ₂₃)+s ₄c ₆s ₂₃ (11)

r ₁₃＝c ₁a _x+s ₁a _y＝-c ₅s ₂₃-c ₄s ₅c ₂₃ (12)

r ₂₃＝-s ₁a _x+c ₁a _y＝s ₄s ₅ (13)

r ₃₃＝c ₄s ₅s ₂₃-c ₅c ₂₃ (14)

q _x1＝l ₆+l ₂c ₂+l ₃c ₂₃-l ₄s ₂₃-l ₅(c ₅s ₂₃+c ₄s ₅c ₂₃) (15)

q _y1＝l ₅s ₄s ₅ (16)

q _z1＝-l ₂s ₂-l ₃s ₂₃-l ₄c ₂₃-l ₅(c ₅c ₂₃-c ₄s ₅s ₂₃) (17)

By formula (8) and formula (11), can be obtained:

${\mathrm{\&theta;}}_1=\arctan \left(\frac{y-a_y{l}_5}{x-a_x{l}_5}\right)---\left(18\right)$

(5.3) solve θ ₃

By θ ₁can obtain matrix

in every value, will after formula (10) and formula (12) square, be added:

${\mathrm{\&theta;}}_3=\arctan \left(\frac{l_3}{l_4}\right)-\arctan (k,\&PlusMinus;\sqrt{{l_3}^2+{l_4}^2-k^2})---\left(19\right)$

Wherein $k=\frac{{(q_{x1}-l_6-a_x{l}_5)}^2+{(q_{z1}-a_z{l}_5)}^2-{l_2}^2-{l_3}^2-{l_4}^2}{{2l}_2}---\left(20\right)$

(5.4) solve θ ₂

Formula (17) substitution formula (10) (12) simultaneous two formulas are obtained

${\mathrm{\&theta;}}_{23}={\mathrm{\&theta;}}_2+{\mathrm{\&theta;}}_3=\arctan \left(\frac{(q_{x1}-l_6-a_x{l}_5)(l_2{s}_3-l_4)-(q_{z1}-a_z{l}_5)(l_2{c}_3+l_3)}{(q_{x1}-l_6-a_x{l}_5)(l_2{c}_3+l_3)+(q_{z1}-a_z{l}_5)(l_2{s}_3-l_4)}\right)---\left(21\right)$

θ ₂＝θ ₂₃-θ ₃ (22)

(5.5) solve θ ₄

Simultaneous formula (7), (9):

c ₄s ₅＝a _zs ₂₃-a _xc ₂₃ (23)

Simultaneous formula (8), (10):

${\mathrm{\&theta;}}_4=\arctan \left(\frac{a_y}{a_z{s}_{23}-a_x{c}_{23}}\right)---\left(24\right)$

(5.6) solve θ ₅

Simultaneous formula (7) (9)

c ₅＝-a _xs ₂₃-a _zc ₂₃ (25)

Simultaneous formula (8), (19)

${\mathrm{\&theta;}}_5=\arctan \left(\frac{\frac{a_y}{\sin \left({\mathrm{\&theta;}}_4\right)}}{-(a_x{s}_{23}+a_z{c}_{23})}\right)---\left(26\right)$

If sin is (θ ₄)=0, simultaneous (7), (9) obtain

${\mathrm{\&theta;}}_5=\arctan \left(\frac{\frac{a_z{s}_{23}-a_x{c}_{23}}{\cos \left({\mathrm{\&theta;}}_4\right)}}{-(a_x{s}_{23}+a_z{c}_{23})}\right)---\left(27\right)$

Step 6, by switch 10, to electrical control cabinet 9, transmit control signal, electrical control cabinet 9 is handled six axle articulated robot bodies 8 according to the control signal of receiving and is carried out corresponding sorting operation, and guiding six axle articulated robot bodies 8 complete absorption, carrying and the stacking of all target items.

Embodiment 1

Some areas that are scattered at random in article to be sorted rest area 6 are about 16mm ², circular Cobastab tablet, utilizes this robot sorting system based on machine vision to complete the sorting operation of the tablet that is scattered, and its concrete steps are as follows:

1) the Cobastab tablet feature of being scattered is selected and inputted, determine to be Cobastab tablet by the target item of being drawn.

New destination article Cobastab tablet in interactive interface, and input according to characteristic sequence, shape selects circle, area to fill in 16mm ², allowable error fills in ± 5%, after determining and store.

2) with industrial CCD digital camera, the Cobastab tablet being scattered is gathered, obtain, after general image, by Ethernet, image data information being transferred to industrial computer.

3) utilize image processing program to carry out the processing such as Threshold segmentation, connected region extraction, feature extraction and matching to image in Fig. 4 (a), extract all satisfactory Cobastab tablets, and its number and state are shown in interactive interface.

1. image is carried out to Threshold segmentation, selected threshold is (86,255), as shown in Fig. 4 (b);

2. image is carried out to connected region extraction, adopt the method for 8 UNICOMs to extract common edge, as shown in Fig. 4 (c);

3. image is carried out to feature extraction and matching, elemental area interval (6000,12000), as shown in Fig. 4 (d);

4. judged whether target item: have

5. the number of display-object article and state in interactive interface: number is 6, state is for showing the position distribution in image of each Cobastab tablet.

4) camera and robot are carried out after hand and eye calibrating, by the calculating of image coordinate system, camera coordinates system and robot coordinate system's transformational relation, determining that each Cobastab tablet is in position and the attitude of robot world's coordinate system.

5) by interactive interface, determine the stacking target location of Cobastab tablet, utilize the teaching of robot control module's crawl to medicine bottle position and store;

6) each Cobastab tablet position and the attitude information that obtain are stored in file, and ask for each Cobastab tablet position and the contrary solution of attitude motion by robot control module, carry out automatic orbit planning, thereby guided robot completes accurate absorption, carrying, the stacking action of all vitamins B tablet.

According to the image after image is processed in Fig. 4 (d), can find out that 6 Cobastab tablets that are scattered are all detected, obtain each Cobastab tablet central point after Coordinate Conversion in the position of robot world's coordinate system simultaneously, and control is drawn all tablets, its error of drawing position and tablet physical location, within 0.2mm, meets application request.The robot sorting system of experimental result demonstration based on machine vision accurately drawn the Cobastab tablet being scattered rapidly, carries, is deposited in the medicine bottle of appointment.

Claims (10)

1. the robot sorting system based on machine vision, it is characterized in that, comprise control module, camera unit, robot cell and tool unit, wherein control module comprises industrial computer (11), switch (10), camera unit comprises CCD digital camera (1), camera lens (2) and light source (3), robot cell comprises six axle articulated robot bodies (8) and electrical control cabinet (9), and tool unit comprises vacuum cup (5);

CCD digital camera (1) is connected with industrial computer (11) by switch (10), six axle articulated robot bodies (8) are connected with electrical control cabinet (9), electrical control cabinet (9) access switch (10), vacuum cup (5) is rigidly fixed in the end of six axle articulated robot bodies (8), and coaxial with the forearm of six axle articulated robot bodies (8);

Camera unit is sorted picture shooting, the data acquisition of article, and the picture of taking and the data of collection are transferred to industrial computer (11) by switch (10); Industrial computer (11) carries out, behind image processing and accurate location, by switch (10), to electrical control cabinet (9), transmitting control signal to the item pictures that is sorted collecting; Electrical control cabinet (9) is handled six axle articulated robot bodies (8) according to the control signal of receiving and is carried out corresponding sorting operation.

2. the robot sorting system based on machine vision according to claim 1, it is characterized in that, the CCD digital camera (1) of described camera unit, camera lens (2) and light source (3) are fixed on camera fixing support (4) in turn, light source (3) and camera lens (2) coaxially arrange and the two angle all adjustable.

3. the robot sorting system based on machine vision according to claim 1, is characterized in that, described tool unit also comprises air compressor machine, magnetic valve and vacuum generator, for vacuum cup (5), to being sorted article, draws.

4. the robot sorting system based on machine vision according to claim 1, is characterized in that, described industrial computer (11) comprises image processing module and robot control module.

5. the robot method for sorting based on machine vision, is characterized in that, comprises the following steps:

Step 1, determines the target item being sorted, and will be sorted feature selecting or the input industrial computer (11) of article;

Step 2, the angle of adjusting camera lens (2) and light source (3) makes the two aim at article to be sorted rest area (6), CCD digital camera (1) is sorted picture shooting, the data acquisition of article, and the picture of taking and the data of collection are transferred to industrial computer (11) by switch (10);

Step 3, the image processing module of industrial computer (11) carries out image processing to being sorted item pictures, judges in image, whether there is target item: if do not had, return to step 2; If had, all target items are extracted, and enter next step;

Step 4, the robot control module of industrial computer (11) carries out after hand and eye calibrating CCD digital camera (1) and six axle articulated robot bodies (8), according to the transformational relation between image coordinate system, camera coordinates system and robot coordinate system, determine that each target item, at robot coordinate system's position and attitude information, is stored into the position of each target item obtaining and attitude information in file;

Step 5, target item rest area (7) is sorted in the target location that definite target item sorting is stacked, and ask for each target item position and the contrary solution of attitude motion by robot control module, carry out the movement locus planning of six axle articulated robot bodies (8);

Step 6, by switch (10), to electrical control cabinet (9), transmit control signal, electrical control cabinet (9) is handled six axle articulated robot bodies (8) according to the control signal of receiving and is carried out corresponding sorting operation, and guiding six axle articulated robot bodies (8) complete absorption, carrying and the stacking of all target items.

6. the robot method for sorting based on machine vision according to claim 5, is characterized in that, will be sorted feature selecting or the input industrial computer (11) of article described in step 1, is specially:

(1.1) if there is no required target item in database filelist, in interactive interface, target item feature is inputted and stored, form database file;

(1.2) if there is required target item in database filelist, directly selects and determine.

7. the robot method for sorting based on machine vision according to claim 5, is characterized in that, the image processing module of industrial computer (11) carries out image processing to being sorted item pictures described in step 3, and the step that wherein image is processed is as follows:

(3.1) to being sorted item pictures, carry out gray processing; Gray value linear scale convergent-divergent formula is as follows:

f(g)＝min(max(ag+b+0.5,0),2 ^b-1)

In formula, a is contrast, and b is brightness, and g is gray value;

(3.2) image after gray processing is carried out to Threshold segmentation; To in image, in the gray value of appointment tonal range, all choose in output area S:

S＝{(r,c)∈ ^R|g _min≤f _r,c≤g _max}

F _r,cfor coordinate in image is the gray value of (r, c) pixel; g _minfor the minimum gradation value for Threshold segmentation; g _maxfor the maximum gradation value for Threshold segmentation, R is image-region;

(3.3) image after Threshold segmentation is carried out to connected region extraction;

(3.4) image after connected region extraction is carried out to feature extraction and matching; Provincial characteristics selects area M to be:

$M=\left|R\right|=\underset{(r,c)\&Element;R}{\mathrm{\&Sigma;}}1$

M is the area in region, | R| is counting in region;

(3.5) according to the image after feature extraction and matching, judged whether target item, the elemental area value of given target item interval (m, n), institute's area that obtains is target item within this interval.

8. the robot method for sorting based on machine vision according to claim 5, it is characterized in that, the robot control module of industrial computer described in step 4 (11) carries out after hand and eye calibrating CCD digital camera (1) and six axle articulated robot bodies (8), according to the transformational relation between image coordinate system, camera coordinates system and robot coordinate system, determine that each target item is at robot coordinate system's position and attitude information, detailed process is as follows:

(4.1) hand and eye calibrating

Known calibration plate is fixedly mounted on to the tool tip of six axle articulated robot bodies (8), robot motion makes scaling board move to CCD digital camera (1) within sweep of the eye, obtain image and image is processed, obtain scaling board at the pose of camera coordinates system, and record the now pose of robot tool coordinate system end; Mobile robot, repeats above step and obtains N group data for N time, makes scaling board cover the regional in the camera visual field;

(4.2) Coordinate Conversion

N group data are processed, wherein tool coordinates system and robot coordinate system's transformational relation ^baseh _toolknown, the transformational relation of scaling board coordinate system and camera coordinates system ^camh _calknown:

^camH _cal= ^camH _base ^baseH _tool ^toolH _cal

According to above formula, by the processing to N group data, obtain the transformational relation between camera coordinates and robot coordinate system ^camh _base, the transformational relation between scaling board coordinate system and coordinate system ^toolh _calfor intermediate variable.

9. the robot method for sorting based on machine vision according to claim 5, it is characterized in that, the mode of determining sorting target item rest area (7) described in step 5 comprises two kinds: utilize robot control module's crawl teaching target location and store, or off-line is manually inputted target location.

10. the robot method for sorting based on machine vision according to claim 5, is characterized in that, asking for each target item position and attitude motion described in step 5, to learn contrary detailed process of separating as follows:

(5.1) in known six axle articulated robot bodies (8), each axle link parameters is as follows:

L ₆be that the first axle connecting rod axle center is that initial point is to the second distance of axle turning cylinder axle center in X-direction, l ₁be that the first axle connecting rod axle center is that initial point is to the second distance of axle turning cylinder axle center in Z-direction; l ₂be the length of the second axle connecting rod; l ₃be the 3rd axle connecting rod axle center to the four distances of axle connecting rod axle center in Z-direction; l ₄be the 3rd axle axle center to four shaft ends the distance in X-direction; l ₅be the length of the 5th axle connecting rod; l ₇be the first axle axle center to the three distances of axle connecting rod axle center in Y-axis, l in Gai robot ₇be 0;

Distal point pose is (x, y, z, A, B, C), and wherein (x, y, z) is the coordinate position under robot coordinate system, and (A, B, C) is the attitude angle under end robot coordinate system;

Solve the angle θ that in six axle articulated robot bodies (8), each axle rotates ₁, θ ₂, θ ₃, θ ₄, θ ₅, θ ₆, in following formula, c represents cos θ, s represents sin θ, the subscript of c, s represents corresponding angle, the double subscript of c, s represent corresponding angle and, i.e. c ₂₃represent cos (θ ₂+ θ ₃), s ₂₃represent sin (θ ₂+ θ ₃);

(5.2) solve θ ₁

According to robot end, put pose (x, y, z, A, B, C), can obtain the transition matrix T between robot coordinate system and tool coordinates system:

$T=\left[\begin{array}{cccc}{n}_x& o_x& a_x& x\\ n_y& o_y& a_y& y\\ n_z& o_z& a_z& z\\ 0& 0& 0& 1\end{array}\right]---\left(1\right)$

In formula,

$\left\{\begin{array}{c}{n}_x=c_B{c}_C\\ n_y=s_A{s}_B{c}_C-c_A{s}_C\\ n_z=c_A{s}_B{c}_C+s_A{s}_C\\ o_x=c_B{s}_C\\ o_y=s_A{s}_B{s}_C+c_A{c}_C\\ o_z=c_A{s}_B{s}_C-s_A{c}_C\\ a_x=-s_B\\ a_y=s_A{c}_B\\ a_z=c_A{c}_B\end{array}\right.---\left(2\right)$

$T_1^w=T_0^w*T_1^0=\left[\begin{array}{cccc}{c}_1& {-s}_1& 0& 0\\ s_1& c_1& 0& 0\\ 0& 0& 1& l_1\\ 0& 0& 0& 1\end{array}\right]---\left(3\right)$ In formula,

be the transformational relation between the first axis coordinate system and robot coordinate system, the like, w is robot coordinate system;

To formula (3) while premultiplication

:

${\left(T_1^w\right)}^{-1}*T=T*_3^2{}_2{}^{1}T*T_4^3*T_5^4*T_6^5*T_T^6---\left(4\right)$

$\left[\begin{array}{cccc}{c}_1& s_1& 0& 0\\ {-s}_1& c_1& 0& 0\\ 0& 0& 1& {-l}_1\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{cccc}{n}_x& o_x& a_x& x\\ n_y& o_y& a_y& y\\ n_z& o_z& a_z& z\\ 0& 0& 0& 1\end{array}\right]=T_T^1---\left(5\right)$

Order $T_T^1=\left[\begin{array}{cccc}{r}_{11}& r_{21}& r_{31}& q_{x1}\\ r_{12}& r_{22}& r_{32}& q_{y1}\\ r_{13}& r_{23}& r_{33}& q_{z1}\\ 0& 0& 0& 1\end{array}\right],$ In formula

r ₁₁＝c ₁n _x+s ₁n _y＝-c ₆(s ₅s ₂₃-c ₄c ₅c ₂₃)-s ₄c ₅s ₂₃ (6)

r ₂₁＝-s ₁n _x+c ₁n _y＝-c ₄c ₆-s ₄c ₅c ₆ (7)

r ₃₁＝n _z＝s ₄s ₆s ₂₃-c ₆(s ₅c ₂₃+c ₄c ₅s ₂₃) (8)

r ₁₂＝c ₁o _x+s ₁o _y＝s ₆(s ₅s ₂₃-c ₄c ₅c ₂₃)-s ₄c ₆c ₂₃ (9)

r ₂₂＝-s ₁o _x+c ₁o _y＝c ₅s ₄s ₆-c ₄c ₆ (10)

r ₃₂＝o _z＝s ₆(s ₅c ₂₃+c ₄c ₅s ₂₃)+s ₄c ₆s ₂₃ (11)

r ₁₃＝c ₁a _x+s ₁a _y＝-c ₅s ₂₃-c ₄s ₅c ₂₃ (12)

r ₂₃＝-s ₁a _x+c ₁a _y＝s ₄s ₅ (13)

r ₃₃＝c ₄s ₅s ₂₃-c ₅c ₂₃ (14)

q _x1＝l ₆+l ₂c ₂+l ₃c ₂₃-l ₄s ₂₃-l ₅(c ₅s ₂₃+c ₄s ₅c ₂₃) (15)

q _y1＝l ₅s ₄s ₅ (16)

q _z1＝-l ₂s ₂-l ₃s ₂₃-l ₄c ₂₃-l ₅(c ₅c ₂₃-c ₄s ₅s ₂₃) (17)

By formula (8) and formula (11), can be obtained:

${\mathrm{\&theta;}}_1=\arctan \left(\frac{y-a_y{l}_5}{x-a_x{l}_5}\right)---\left(18\right)$

(5.3) solve θ ₃

By θ ₁can obtain matrix

in every value, will after formula (10) and formula (12) square, be added:

${\mathrm{\&theta;}}_3=\arctan \left(\frac{l_3}{l_4}\right)-\arctan (k,\&PlusMinus;\sqrt{{l_3}^2+{l_4}^2-k^2})---\left(19\right)$

Wherein $k=\frac{{(q_{x1}-l_6-a_x{l}_5)}^2+{(q_{z1}-a_z{l}_5)}^2-{l_2}^2-{l_3}^2-{l_4}^2}{{2l}_2}---\left(20\right)$

(5.4) solve θ ₂

Formula (17) substitution formula (10) (12) simultaneous two formulas are obtained

${\mathrm{\&theta;}}_{23}={\mathrm{\&theta;}}_2+{\mathrm{\&theta;}}_3=\arctan \left(\frac{(q_{x1}-l_6-a_x{l}_5)(l_2{s}_3-l_4)-(q_{z1}-a_z{l}_5)(l_2{c}_3+l_3)}{(q_{x1}-l_6-a_x{l}_5)(l_2{c}_3+l_3)+(q_{z1}-a_z{l}_5)(l_2{s}_3-l_4)}\right)---\left(21\right)$

θ ₂＝θ ₂₃-θ ₃ (22)

(5.5) solve θ ₄

Simultaneous formula (7), (9):

c ₄s ₅＝a _zs ₂₃-a _xc ₂₃ (23)

Simultaneous formula (8), (10):

${\mathrm{\&theta;}}_4=\arctan \left(\frac{a_y}{a_z{s}_{23}-a_x{c}_{23}}\right)---\left(24\right)$

(5.6) solve θ ₅

Simultaneous formula (7) (9)

c ₅＝-a _xs ₂₃-a _zc ₂₃ (25)

Simultaneous formula (8), (19)

${\mathrm{\&theta;}}_5=\arctan \left(\frac{\frac{a_y}{\sin \left({\mathrm{\&theta;}}_4\right)}}{-(a_x{s}_{23}+a_z{c}_{23})}\right)---\left(26\right)$

If sin is (θ ₄)=0, simultaneous formula (7), (9) obtain

${\mathrm{\&theta;}}_5=\arctan \left(\frac{\frac{a_z{s}_{23}-a_x{c}_{23}}{\cos \left({\mathrm{\&theta;}}_4\right)}}{-(a_x{s}_{23}+a_z{c}_{23})}\right)---\left(27\right).$

Images