CN111452038B - High-precision workpiece assembly and assembly method of high-precision workpiece assembly - Google Patents

Abstract

The invention discloses a high-precision workpiece assembly and an assembly method thereof.A robot drives an image collector to respectively collect images of the workpiece assembly, the images collected by the image collector of a computer are used for accurately positioning the workpiece assembly, and the robot realizes the grabbing and assembly of the workpiece assembly according to the position information of the workpiece; in the assembly of the workpiece components, the computer accurately and stably controls the robot to repeatedly adjust the relative positions of the workpiece components through the PID controller according to the data detected by the six-dimensional sensor until the workpiece components are assembled; the robot finishes the assembly operation of the workpiece assembly and is automatically controlled by a computer, so that the assembly is more intelligent; the robot is controlled by the PID controller, so that the assembly process of the workpiece assembly is more stable and accurate, the PID control principle is simple and clear, the computer operation speed is high, the parameters of the PID controller can be adjusted according to historical docking data, the assembly process of the workpiece assembly is optimized, the production cost is reduced, and the production efficiency is improved.

Description

High-precision workpiece assembly and assembly method of high-precision workpiece assembly

technical field

The invention relates to the technical field of high-precision workpiece assembly assembly, in particular to a high-precision workpiece assembly and a method for assembling the high-precision workpiece assembly.

Background technique

In modern industrial production, the assembly process of most workpiece components is completed by robotic arms. Among them, when it comes to the assembly of high-precision workpiece components, the disadvantage of low accuracy of the visual positioning method often occurs that the positioning of the workpiece components is inaccurate, resulting in the inaccuracy of the robot. The workpiece components are clamped, so the subsequent assembly of the workpiece components cannot be completed, thereby reducing the stability of the assembly system, reducing the yield, and affecting the production efficiency; and when the computer is assembling the high-precision workpiece components, the computer also calibrates the assembly position many times. However, it is still impossible to complete the assembly of the workpiece components, and even cause the computer to crash, causing the assembly process to be stuck and prolonging the assembly cycle.

According to the deficiencies of the prior art, there is a need for a method that can precisely locate a high-precision workpiece assembly, and can also quickly, accurately and stably complete the assembly of the workpiece assembly.

SUMMARY OF THE INVENTION

In order to solve the above problems, the present invention proposes a high-precision workpiece assembly and an assembly method of the high-precision workpiece assembly. The assembly method of the workpiece assembly can precisely locate the workpiece assembly, and can also quickly, accurately and stably complete the assembly of the workpiece assembly. .

In order to achieve the above object, the specific technical solution adopted in the present invention is as follows: a method for assembling a high-precision workpiece assembly, the key point of which is to include a workpiece assembly, and the workpiece assembly includes a workpiece M and a workpiece N, and the workpiece M and the workpiece N Assemble as follows:

Preprocessing: place the workpiece M on the tooling to be grasped, and place the workpiece N on the docking tooling; install the image collector and the six-dimensional sensor on the robot, and set the image collector to take pictures of the workpiece M and the workpiece N respectively. point; set the alignment position and grasping position of the robot gripper on the workpiece M in the computer, set the pre-joint position and the docking position of the workpiece M and the workpiece N; set the initial point of the robot; set the workpiece M and the workpiece respectively The image ROI region of N; set the unit vector (x _e , y _e ).

S1: Initialization, the robot moves to the initial point, and the robot establishes communication with the computer.

构成向量与所述单位向量(x_e,y_e)之间的夹角。S2: The robot drives the image collector to move to the fixed photographing point of the workpiece M, collects the image of the workpiece M, transmits the image of the workpiece M to the computer, and the computer analyzes the image of the workpiece M and determines the center point P _{M of the workpiece M} The feature base image coordinates and rotation angle θ _M , the rotation angle θ _M is the center point P _M of the workpiece M that points to the center point of the image ROI area of the workpiece M

constitutes the angle between the vector and the unit vector (x _e , y _e ).

S3: The robot adjusts the gripper itself and moves to the alignment position of the workpiece _{M according to the feature base image coordinates and rotation angle θ M of} the center point PM of the workpiece M calculated in the previous step, and starts to descend until it reaches the grasping position. The robot controls the gripper to grasp the workpiece M.

构成向量与所述单位向量(x_e,y_e)之间的夹角。S4: The robot drives the image collector to move to the fixed photographing point of the workpiece N, collects the image of the workpiece N, transmits the image of the workpiece N to the computer, and the computer analyzes the image of the workpiece N and determines the center point P _{N of the workpiece N} The feature base image coordinates and rotation angle θ _N , the rotation angle θ _N is the center point P _N of the workpiece N pointing to the center point of the image ROI area of the workpiece N

constitutes the angle between the vector and the unit vector (x _e , y _e ).

S5: The robot moves the workpiece M to a pre-docking position, which is set directly above the workpiece N according to the feature base image coordinates and the rotation angle θ _N of the center point P _N of the workpiece N calculated in the previous step .

S6: The robot controls the workpiece M to connect with the workpiece N, the six-dimensional sensor detects the detection data when the workpiece M and the workpiece N are docked, and the computer controls the robot to repeatedly adjust the position of the workpiece M according to the detection data until the detection data is smaller than the installation threshold, and controls the workpiece. M is docked with the workpiece N, wherein the installation threshold includes at least a torque threshold and a force threshold.

S7: The robot controls the gripper to release the workpiece M, and returns to the initial point for the next assembly.

With the above design, the computer collects the images of the workpiece M and the workpiece N through the image collector, and determines the center point feature base for alignment in the workpiece M and the workpiece N according to the transformation relationship between the robot and the image collector. The image coordinates and the center base coordinates of the image ROI area can determine the rotation angle of the two workpieces; setting the image ROI area on the workpiece can be used to calibrate the workpiece position to facilitate workpiece grasping or workpiece docking, and also set the core components of the workpiece as The image ROI area is used to detect the production quality of the workpiece; the robot realizes the grasping of the workpiece according to the image coordinates and rotation angle of the center point feature of the two workpieces. The computer uses the detection data of the six-dimensional sensor, and the robot repeatedly adjusts the position of the workpiece M. , until the workpiece M and the workpiece N are docked; using this method to assemble the workpiece components to realize the automatic process, and the workpiece M and the workpiece N are adjusted in the set docking position, and the docking is completed after the docking threshold is met, reducing the friction of the docking part of the workpiece components It can prevent the workpiece components from being damaged and scrapped directly and completely, thereby reducing the product loss rate, thereby reducing the production cost.

工件N的中心点P_N、工件N的图像ROI区域中心点

机器人初始点、图像采集器的位置通过建立空间坐标系进行标定。Further description, the center point P _M of the workpiece M, the center point of the image ROI area of the workpiece M

The center point P _N of the workpiece N, the center point of the image ROI area of the workpiece N

The initial point of the robot and the position of the image collector are calibrated by establishing a space coordinate system.

Wherein, an image coordinate system is set in the picture collected at any position in the image collector, and a feature image coordinate is correspondingly set for all the features in the corresponding picture, and the robot is set in the base coordinate system, and the The feature image coordinates are transformed by the coordinate system transformation matrix A _ij to obtain the feature base image coordinates in the base coordinate system.

Using the above scheme, the computer can calibrate the positions of the workpiece M, the workpiece N, the robot and the image collector, which is convenient for positioning; the position calibration of the center point of the workpiece M and the workpiece N and the center point of the image ROI area is used for grasping the workpiece. and docking.

Further description, in steps S2 and S4, the feature base image coordinates of the center point of the workpiece and the feature base image coordinates of the center point of the image ROI area are both to obtain the centroid coordinates in the corresponding contour in the image, specifically:

S-A1: The computer performs lens distortion correction on the image collected by the image collector to obtain the image after lens distortion correction, specifically:

畸变校正后的像素点的坐标为(u，v)，(k₁，k₂，k₃，p₁，p₂)是畸变系数，r是图像采集器镜头半径，计算机通过棋盘格获取W₁组相对应的像素点，利用最小二乘法就求出所有的畸变系数，得到镜头畸变校正后的图像。Among them, the original coordinates of the pixel points before distortion correction are

The coordinates of the pixel point after distortion correction are (u, v), (k ₁ , k ₂ , k ₃ , p ₁ , p ₂ ) is the distortion coefficient, r is the radius of the image collector lens, and the computer obtains W ₁ through the checkerboard For the corresponding pixel points of the group, all the distortion coefficients are obtained by the least square method, and the image after lens distortion correction is obtained.

S-A2: The computer converts the image after lens distortion correction from RGB color space to HSV color space, and performs multiple threshold segmentation and logical AND operation on the converted image to obtain the rough outline of the workpiece.

S-A3: The computer smoothes the rough outline of the workpiece, converts the converted HSV image into a binary image, and then performs image morphological processing of image erosion and image expansion on the binary image in turn:

Image erosion:

Image inflation:

是处理过后的图像，f(x,y)是原始图像，

是结构元素，使图像中工件轮廓更为光滑。in,

is the processed image, f(x,y) is the original image,

It is the structural element that makes the contour of the workpiece in the image smoother.

S-A4: The computer uses the Canny algorithm to extract all the contours in the image obtained in step S-A3, and then obtains the precise contours in the image according to the area threshold.

S-A5: The computer uses the gray centroid method to obtain the centroid coordinates of the center point of the precise contour in the image in the ROI area f

为图像坐标系中的特征图像坐标。Among them, the center of mass coordinates

is the feature image coordinate in the image coordinate system.

With the above solution, the computer obtains the characteristic image coordinates of the workpiece center point and the image ROI area center point from the workpiece image, and locates the workpiece.

To further describe, in step S-A2, the computer converts the image after lens distortion correction from the RGB color space to the HSV color space. The specific steps are:

S-B1, calculate the conversion factor:

Δ= _Cmax - _Cmin

Among them, (R _P , G _P , B _P ) is the pixel value of point P in the RGB color space.

S-B2, H calculation:

S-B3, S calculation:

S-B4, V calculation:

VP _{= Cmax}

Thus, the computer can obtain the value (H _P , S _P , V _P ) of point P in the HSV color space.

S-B5, the computer extracts the color of the outline as needed, and sets the color range in the HSV color space.

Thereby, the image converted into HSV is obtained, and then multiple thresholds are divided, and the logical AND operation is performed to obtain the image containing the rough outline of the workpiece.

Using the above scheme, the image after computer lens distortion correction is converted from RGB color space to HSV color space, which can make the contour of the workpiece in the image express the color, tone, and degree of vividness more intuitively, and make it easier for the computer to obtain the contour of the workpiece; among them, you can According to the need to obtain the color of the contour, set the color range of the HSV color space, and then the contour can be extracted into the image as required.

Describing further, the coordinate system conversion matrix A _ij is the coordinate system conversion coefficient in the Eye In Hand hand-eye calibration, and the coordinate system conversion coefficient is obtained by the DLT method, specifically:

S-C1: The computer collects the characteristic image coordinates of at least three points in the image, and establishes the relationship between the image coordinate system and the base coordinate system according to the working principle of the image collector:

where (u, v) are the feature image coordinates, (X _r , Y _r , Z _r , 1) are the base coordinates of the robot, f is the focal length, and dx and dy represent the physical dimensions of each pixel on the X and Y axes, respectively , (u ₀ , v ₀ ) are the coordinates of the principal point.

S-C2: Since the robot drives the image collector to take pictures of the workpiece at the same height, the Z value of the characteristic image coordinate in the image is a constant, and the above formula is simplified to get:

为坐标系转换矩阵A_ij。in,

is the coordinate system transformation matrix A _ij .

S-C3: The computer brings the collected feature image coordinates into S-C21:

Bring the coordinate value of the X-axis in the feature image coordinates collected in step S-C1 into S-C21:

Obtain the first row of coefficients in the coordinate system conversion matrix A _ij ; Similarly, the coordinates of the Y-axis in the collected image coordinates are brought into S-C21 to obtain a complete coefficient matrix A _ij ; The characteristic image coordinates of the image coordinate system take the left The standard system transformation matrix A _ij is converted into the feature base image coordinates of the base coordinate system.

Using the above scheme, the computer obtains the coordinate system transformation matrix between the base coordinate system where the robot is located and the image coordinate system in the image collector through the DLT method, and realizes the coordinate transformation between the base coordinate system and the image collector. The feature map line coordinates in the image can be directly converted into the feature base image coordinates in the base coordinate system for positioning.

组成的有向线段

所述旋转角θ_X为所述有向线段

与单位向量(x_e,y_e)之间的夹角，所述旋转角具体为：Further description, the workpiece center point P _X on the workpiece to the image ROI area center point

directed line segment

The rotation angle θ _X is the directional line segment

The included angle with the unit vector (x _e , y _e ), the rotation angle is specifically:

Among them, the robot adjusts the workpiece docking position according to the rotation angle θ _X ; or the clamping position; or the alignment position; or the pre-docking position.

Using the above scheme, the computer obtains the rotation angles of the workpiece M and the workpiece N respectively, and the computer controls the robot to adjust the position of the workpiece M through the values of the two rotation angles, so that the two rotation angles are consistent, that is, the workpiece M and the workpiece N are aligned.

Describing further, in step S6, the robot controls the workpiece M and the workpiece N to dock, the six-dimensional sensor detects the detection data of the docking part of the workpiece M and the workpiece N, and the computer controls the robot to repeatedly adjust the position of the workpiece M according to the detection data, until the detection data is smaller than the installation. After the threshold value, the specific steps for controlling workpiece M and workpiece N to complete the docking are:

S-D1: The robot moves the workpiece M downward from the pre-docking position until it reaches the docking position, and the computer reads the detection data of the six-dimensional sensor.

S-D2: The computer judges whether the detection data of the six-dimensional sensor are all smaller than the installation threshold; if so, go to step S-D5; if not, go to step S-D3.

S-D3: The computer analyzes the detection data of the six-dimensional sensor, the PID controller receives the detection data of the six-dimensional sensor and gives the current adjustment value, the robot moves the workpiece M upward W ₂ , the robot adjusts the position of the workpiece M according to the current adjustment value, the robot Move the workpiece M downward, and then reach the docking position, and the computer reads the detection data of the six-dimensional sensor.

S-D4: The computer judges whether the detection data of the six-dimensional sensor are all smaller than the installation threshold; if so, go to step S-D5; if not, return to step S-D3.

S-D5: The robot moves the workpiece M down W ₃ to complete the docking of the workpiece M and the workpiece N.

With the above solution, the robot repeatedly adjusts the position of the workpiece M through the PID controller, and finally makes the workpiece M and the workpiece N accurately docked; through the stability and high cost performance of the PID controller, the workpiece assembly system is more stable and costs are reduced.

Describing further, the described concrete steps of adjusting workpiece M by PID controller according to the adjustment strategy are:

S-D41: The control principle of the computer according to the PID control:

Δu(k)=K _P [e(k)-e(k-1)]+K _I e(k)+K _D [e(k)-2e(k-1)+e(k-2)]

Among them, Δu(k) is the current adjustment amount, e(k) is the current error, e(k-1) is the last error, e(k-2) is the last error, K _P proportional coefficient, K _I integral coefficient, K _D is the differential coefficient.

S-D42: In order to shorten the setting period, the computer removes the differential link, and the above formula is simplified to:

Δu(k)=K _P [e(k)-e(k-1)]+K _I e(k);

S-D43: Of which:

e(k)=D(k)-D

Among them, D(k) is the current degree of the six-dimensional force sensor, and D is the ideal value.

S-D44: In order to avoid noise response, the computer will continuously obtain the detection data of W ₄ six-dimensional sensors and take the average value as D(k), specifically:

S-D45: Set D=0, the above formula is simplified to:

Δu(k)=K _P [D(k)-D(k-1)]+K _I D(k)

The robot repeatedly adjusts the position of the workpiece M according to the current adjustment amount Δu(k) until the detection data of the six-dimensional sensor is smaller than the installation threshold.

Using the above scheme, the robot is controlled by PID, which has good applicability in the control process; the PID control principle is simple and clear, each control parameter is relatively independent, the computer operation speed is fast, and the deviation can be dynamically corrected, the response is rapid, and the efficiency of the workpiece component docking is improved. ; The PID controller can correct the parameters in the PID controller according to the historical docking data of the workpiece components to optimize the entire docking process of the workpiece components.

A high-precision workpiece assembly, the key is that it includes the workpiece M and the workpiece N, the workpiece M is a cylinder, one bottom surface of the workpiece M is a clamped surface, and the other bottom surface of the workpiece M is a butt joint At least three limit flanges extend radially from the clamped surface of the workpiece M, the limit flanges open limit through holes along the axial direction of the workpiece M, and the abutting surface of the workpiece M extends along the axial direction of the workpiece M. At least three butt flanges extend radially, the butt flanges are provided with butt through holes along the axial direction of the workpiece M, and at least one butt blind hole is opened on the butt surface of the cylinder.

The workpiece N is a cylinder, and a bottom surface of the workpiece N is a butted surface, and the butted surface is adapted to the size and shape of the butted surface, and the butted surface of the workpiece N extends radially out at least three times. A butted flange is provided with a butted through hole along the N-axis of the workpiece, the diameter of the butted through hole is adapted to the diameter of the butted through hole, and the butted surface is also A butt cylinder coaxial with the workpiece N is provided, the butt cylinder and the workpiece N are integrally formed, the bottom surface radius of the butt cylinder is the same as the diameter of the butt blind hole, and the upper bottom surface of the butt cylinder is provided with a mounting plate. Blind hole.

The workpiece M is butted with the workpiece N through a butting blind hole and a butting cylinder, and the setting positions of the abutting through holes are in a one-to-one correspondence with the setting positions of the butted through holes.

The beneficial effects of the invention are as follows: the robot drives the image collector to collect the images of the workpiece components respectively, the images collected by the computer image collector precisely locate the workpiece components, and the robot realizes grasping and assembling of the workpiece components according to the workpiece position information; According to the detection data of the six-dimensional sensor, the computer controls the robot to repeatedly adjust the relative position between the workpiece components through the PID controller accurately and smoothly until the workpiece components are assembled; the robot completes the assembly operation of the workpiece components is automatically controlled by the computer, Make the assembly more intelligent; controlling the robot through the PID controller can make the assembly process of the workpiece components more stable and accurate, and the PID control principle is simple and clear, the computer operation speed is fast, and the parameters of the PID controller can be adjusted according to the historical docking data. Optimize the assembly process of workpiece components, reduce production costs and improve production efficiency.

Description of drawings

Fig. 1 is the flow chart of the assembling method of the high-precision workpiece assembly in the present invention;

Fig. 2 is the flow chart that robot controls workpiece M and workpiece N to complete docking among the present invention;

Fig. 3 is the schematic diagram of the high-precision workpiece assembly in the present invention;

Fig. 4 is the sectional view of the high-precision workpiece assembly in the present invention that is not butted;

FIG. 5 is a cross-sectional view of the high-precision workpiece assembly in the present invention after the docking is completed.

Detailed ways

The specific embodiments and working principles of the present invention will be further described in detail below with reference to the accompanying drawings.

In this embodiment, W ₁ =15, W ₂ =10 mm, W ₃ =4 mm, and W ₄ =100.

As can be seen from Figure 1, a method for assembling a high-precision workpiece assembly includes a workpiece assembly, and the workpiece assembly includes a workpiece M and a workpiece N, and the workpiece M and the workpiece N are assembled according to the following steps:

Preprocessing: place the workpiece M on the tooling to be grasped, and place the workpiece N on the docking tooling; install the image collector and the six-dimensional sensor on the robot, and set the image collector to take pictures of the workpiece M and the workpiece N respectively. point; set the alignment position and grasping position of the robot gripper on the workpiece M in the computer, set the pre-joint position and the docking position of the workpiece M and the workpiece N; set the initial point of the robot; set the workpiece M and the workpiece respectively The image ROI region of N; set the unit vector (x _e , y _e ).

S1: Initialization, the robot moves to the initial point. In this embodiment, the robot establishes Socket communication with the computer.

构成向量与所述单位向量(x_e,y_e)之间的夹角。S2: The robot drives the image collector to move to the fixed photographing point of the workpiece M, collects the image of the workpiece M, transmits the image of the workpiece M to the computer, and the computer analyzes the image of the workpiece M and determines the center point P _{M of the workpiece M} The feature base image coordinates and rotation angle θ _M , the rotation angle θ _M is the center point P _M of the workpiece M that points to the center point of the image ROI area of the workpiece M

constitutes the angle between the vector and the unit vector (x _e , y _e ).

S3: The robot adjusts the gripper itself and moves to the alignment position of the workpiece _{M according to the feature base image coordinates and rotation angle θ M of} the center point PM of the workpiece M calculated in the previous step, and starts to descend until it reaches the grasping position. The robot controls the gripper to grasp the workpiece M.

构成向量与所述单位向量(x_e,y_e)之间的夹角。S4: The robot drives the image collector to move to the fixed photographing point of the workpiece N, collects the image of the workpiece N, transmits the image of the workpiece N to the computer, and the computer analyzes the image of the workpiece N and determines the center point P _{N of the workpiece N} The feature base image coordinates and rotation angle θ _N , the rotation angle θ _N is the center point P _N of the workpiece N pointing to the center point of the image ROI area of the workpiece N

constitutes the angle between the vector and the unit vector (x _e , y _e ).

S5 : The robot moves the workpiece M to a pre-docking position, which is set just above the workpiece N, according to the feature base image coordinates and the rotation angle θ _B of the center point P _N of the workpiece N calculated in the previous step.

S6: The robot controls the workpiece M to connect with the workpiece N, the six-dimensional sensor detects the detection data when the workpiece M and the workpiece N are docked, and the computer controls the robot to repeatedly adjust the position of the workpiece M according to the detection data until the detection data is smaller than the installation threshold, and controls the workpiece. M is docked with the workpiece N, wherein the installation threshold includes at least a torque threshold and a force threshold. In this embodiment, the torque threshold is 10NM and the force threshold is 30N.

S7: The robot controls the gripper to release the workpiece M, and returns to the initial point for the next assembly.

工件N的中心点P_N、工件N的图像ROI区域中心点

机器人初始点、图像采集器的位置通过建立空间坐标系进行标定。It can be seen in conjunction with Fig. 1 that the center point P _M of the workpiece M and the center point of the image ROI area of the workpiece M

The center point P _N of the workpiece N, the center point of the image ROI area of the workpiece N

The initial point of the robot and the position of the image collector are calibrated by establishing a space coordinate system.

Among them, the image collected at any position in the image collector is set with an image coordinate system, and all features in the corresponding picture are set with a corresponding feature image coordinate, which is set in the base coordinate system in the robot, and the feature image coordinates are set through the coordinate system. After the transformation matrix A _ij is transformed, the feature base image coordinates in the base coordinate system are obtained.

It can be seen in conjunction with Fig. 1 that in steps S2 and S4, the feature base image coordinates of the center point of the workpiece and the feature base image coordinates of the center point of the image ROI area are both to obtain the centroid coordinates in the corresponding contour in the image, specifically:

S-A1: The computer performs lens distortion correction on the image collected by the image collector to obtain the image after lens distortion correction, specifically:

畸变校正后的像素点的坐标为(u，v)，(k₁，k₂，k₃，p₁，p₂)是畸变系数，r是图像采集器镜头半径，本实施例中，计算机通过棋盘格获取15组相对应的像素点，利用最小二乘法就求出所有的畸变系数，得到镜头畸变校正后的图像。Among them, the original coordinates of the pixel points before distortion correction are

The coordinates of the pixel point after distortion correction are (u, v), (k ₁ , k ₂ , k ₃ , p ₁ , p ₂ ) is the distortion coefficient, and r is the radius of the image collector lens. The checkerboard obtains 15 groups of corresponding pixel points, and uses the least squares method to obtain all the distortion coefficients to obtain the image after lens distortion correction.

S-A2: The computer converts the image after lens distortion correction from RGB color space to HSV color space, and performs multiple threshold segmentation and logical AND operation on the converted image to obtain the rough outline of the workpiece.

S-A3: The computer smoothes the rough outline of the workpiece, converts the converted HSV image into a binary image, and then performs image morphological processing of image erosion and image expansion on the binary image in turn:

Image erosion:

Image inflation:

是处理过后的图像，f(x,y)是原始图像，

是结构元素，使图像中工件轮廓更为光滑。in,

is the processed image, f(x,y) is the original image,

It is the structural element that makes the contour of the workpiece in the image smoother.

S-A4: The computer uses the Canny algorithm to extract all the contours in the image obtained in step S-A3, and then obtains the precise contours in the image according to the area threshold.

S-A5: The computer uses the gray centroid method to obtain the centroid coordinates of the center point of the precise contour in the image in the ROI area f

为图像坐标系中的特征图像坐标。Among them, the center of mass coordinates

is the feature image coordinate in the image coordinate system.

It can be seen from Figure 1 that in step S-A2, the computer converts the image after lens distortion correction from the RGB color space to the HSV color space. The specific steps are:

S-B1, calculate the conversion factor:

Δ= _Cmax - _Cmin

Among them, (R _P , G _P , B _P ) is the pixel value of point P in the RGB color space;

S-B2, H calculation:

S-B3, S calculation:

S-B4, V calculation:

VP _{= Cmax}

Thus, the computer can obtain the value (H _P , S _P , V _P ) of point P in the HSV color space.

S-B5, the computer extracts the color of the outline as needed, and sets the color range in the HSV color space.

Thereby, the image converted into HSV is obtained, and then multiple thresholds are divided, and the logical AND operation is performed to obtain the image containing the rough outline of the workpiece.

In this embodiment, the color of the workpiece M and the workpiece M are both black, the color of the docking tool and the tool to be grasped are both red, and the image ROI areas of the workpiece M and the workpiece N are respectively set on the limit flange and the butted flange. , and the color of the image ROI area in workpiece M and workpiece N is green, when obtaining the image coordinates of the center point of workpiece M and workpiece N, in the HSV color space, the red range is set as:

When obtaining the image coordinates of the center point of the image ROI area of workpiece M and workpiece N, in the HSV color space, the green range is set as:

It can be seen from Figure 1 that the coordinate system transformation matrix A _ij is the coordinate system transformation matrix in the Eye In Hand hand-eye calibration, and the coordinate system transformation matrix is obtained by the DLT method, specifically:

S-C1: In this embodiment, the computer collects the characteristic image coordinates of 9 points in the image, and establishes the relationship between the image coordinate system and the base coordinate system according to the working principle of the image collector:

where (u, v) are the feature image coordinates, (X _r , Y _r , Z _r , 1) are the base coordinates of the robot, f is the focal length, and dx and dy represent the physical dimensions of each pixel on the X and Y axes, respectively , (u ₀ , v ₀ ) are the coordinates of the principal point.

S-C2: Since the robot drives the image collector to take pictures of the workpiece at the same height, the Z value of the characteristic image coordinate in the image is a constant, and the above formula is simplified to get:

为坐标系转换矩阵A_ij。in,

is the coordinate system transformation matrix A _ij .

S-C3: The computer brings the collected feature image coordinates into S-C21:

Bring the coordinate value of the X-axis in the feature image coordinates collected in step S-C1 into S-C21:

Obtain the first row of coefficients in the coordinate system conversion matrix A _ij ; Similarly, the coordinates of the Y-axis in the collected image coordinates are brought into S-C21 to obtain a complete coefficient matrix A _ij ; The characteristic image coordinates of the image coordinate system take the left The standard system transformation matrix A _ij is converted into the feature base image coordinates of the base coordinate system.

组成的有向线段

旋转角θ_X为有向线段

与单位向量(x_e,y_e)之间的夹角，旋转角具体为：Combining with Figure 1, it can be seen that the workpiece center point P _X on the workpiece to the center point of the image ROI area

directed line segment

The rotation angle θ _X is the directed line segment

The included angle with the unit vector (x _e , y _e ), the rotation angle is specifically:

Among them, the robot adjusts the workpiece docking position according to the rotation angle θ _X ; or the clamping position; or the alignment position; or the pre-docking position.

As can be seen from Figure 1 and Figure 2, in step S6, the robot controls the workpiece M to connect with the workpiece N, the six-dimensional sensor detects the detection data of the butting part of the workpiece M and the workpiece N, and the computer controls the robot to repeatedly adjust the position of the workpiece M according to the detection data. , until the detection data is less than the installation threshold, the specific steps for controlling the workpiece M and the workpiece N to complete the docking are:

S-D1: The robot moves the workpiece M downward from the pre-docking position until it reaches the docking position, and the computer reads the detection data of the six-dimensional sensor.

S-D2: The computer judges whether the detection data of the six-dimensional sensor are all smaller than the installation threshold; if so, go to step S-D5; if not, go to step S-D3.

S-D3: The computer analyzes the detection data of the six-dimensional sensor, the PID controller receives the detection data of the six-dimensional sensor and gives the current adjustment value, the robot moves the workpiece M upward by 10mm, and the robot adjusts the position of the workpiece M according to the current adjustment value. Move the workpiece M down, and then reach the docking position, and the computer reads the detection data of the six-dimensional sensor.

S-D4: The computer judges whether the detection data of the six-dimensional sensor are all smaller than the installation threshold; if so, go to step S-D5; if not, return to step S-D3.

S-D5: The robot moves the workpiece M down by 4mm to complete the docking between the workpiece M and the workpiece N.

Combining Figure 1 and Figure 2, it can be seen that the specific steps for adjusting the workpiece M through the PID controller according to the adjustment strategy are:

S-D41: The control principle of the computer according to the PID control:

Δu(k)=K _P [e(k)-e(k-1)]+K _I e(k)+K _D [e(k)-2e(k-1)+e(k-2)]

Among them, Δu(k) is the current adjustment amount, e(k) is the current error, e(k-1) is the last error, e(k-2) is the last error, K _P proportional coefficient, K _I integral coefficient, K _D is the differential coefficient.

S-D42: In order to shorten the setting period, the computer removes the differential link, and the above formula is simplified to:

Δu(k)=K _P [e(k)-e(k-1)]+K _I e(k);

S-D43: Of which:

e(k)=D(k)-D

Among them, D(k) is the current degree of the six-dimensional force sensor, and D is the ideal value.

S-D44: In order to avoid noise response, the computer will obtain the average value of the detection data of 100 six-dimensional sensors as D(k), specifically:

S-D45: Set D=0, the above formula is simplified to:

Δu(k)=K _P [D(k)-D(k-1)]+K _I D(k)

The robot repeatedly adjusts the position of the workpiece M according to the current adjustment amount Δu(k) until the detection data of the six-dimensional sensor is smaller than the installation threshold.

3-5, it can be seen that a high-precision workpiece assembly includes a workpiece M and a workpiece N, the workpiece M is a cylinder, one bottom surface of the workpiece M is the clamped surface, and the other bottom surface of the workpiece M is the butting surface. , In this embodiment, 12 limit flanges 11 extend radially from the clamped surface of the workpiece M, and the limit flanges open limit through holes 12 along the axial direction of the workpiece M. In this embodiment, the workpiece M Four butt flanges 13 extend radially from the abutting surface of the cylindrical body, the butt flanges are provided with butt-connecting holes 14 along the axial direction of the workpiece M, and at least one butt-blind hole 15 is opened on the butting surface of the cylinder.

The workpiece N is a cylinder, and a bottom surface of the workpiece N is a butted surface, and the butted surface is adapted to the size and shape of the butted surface. In this embodiment, the butted surface of the workpiece N extends radially with four butted surfaces. The flange 21, the butted flange is provided with a through hole 22 along the axial direction of the workpiece N, the diameter of the through hole 22 is adapted to the diameter of the through hole 14, and the butted surface is also provided with the same diameter as the workpiece N. The abutment cylinder 23 of the shaft is integrally formed with the workpiece N, the bottom surface radius of the butt cylinder 23 is the diameter of the butt blind hole 15 , and a blind installation hole 24 is opened on the upper bottom surface of the butt cylinder 23 .

The workpiece M is matched with the butting cylinder 23 through the butting blind hole 15 to be butted with the workpiece N, and the setting positions of the opposing through holes 14 are in a one-to-one correspondence with the setting positions of the butted through holes 22 .

The working principle of the present invention:

Positioning of workpiece components:

The computer collects 9 reference points according to the DLT method according to the installation position of the image collector on the robot and the coordinates of the base coordinate system where the robot is located, and calculates the coordinate system transformation matrix A _ij .

The computer will remove the distortion and distortion caused by the lens of the image collector through the lens distortion correction of the image of the workpiece M collected from the fixed position.

The computer converts the distortion-corrected image from the RGB color space to the HSV color space, so that the outline of the workpiece in the image can express the color, tone, and vividness more intuitively. The colors of the tooling to be grabbed are all red, the image ROI areas of workpiece M and workpiece N are respectively set on the limit flange and the butted flange, and the color of the image ROI areas in workpiece M and workpiece N are both green. Determine the value range of different colors, and extract the required contour. If the value range of the HSV color space is set as the red value range, the converted HSV image is divided by multiple thresholds, and the logical AND operation is performed to obtain the workpiece M. Or the image of the rough outline of workpiece N, if the value range of the HSV color space is set to the green value range, the rough outline of the image ROI area in workpiece M and workpiece N can be obtained, and then the converted HSV image is divided by multiple thresholds. , and perform logical AND operation to obtain an image containing the rough outline of the image ROI area in workpiece M or workpiece N.

The computer converts the obtained rough outline image into a binary image, and sequentially performs image erosion and image expansion on the binary image to make the contours in the image smoother; then use the Canny algorithm to extract all the contours of the image, and then use the Canny algorithm to extract all the contours of the image, and then according to the area threshold Get precise contours in the image.

The computer uses the gray centroid method to obtain the centroid coordinates of the center point of the precise contour in the image in the ROI area, that is, the characteristic image coordinates of the workpiece center point and the characteristic image coordinates of the center point of the workpiece image ROI area, and then the characteristics of the workpiece center point are obtained. The image coordinates and the feature image coordinates of the center point of the ROI area of the workpiece image are converted into the feature base image coordinates in the base coordinate system according to the coordinate system transformation matrix A _ij .

Grabbing of workpiece components:

The center point of the workpiece and the center point of the image ROI area form the directional line segment of the workpiece, and the angle between the directional line segment and the set unit vector is calculated to obtain the rotation angle of the workpiece

的特征基图像坐标以及旋转角，控制机器人抓取工件M，再根据工件N中心点P_N的特征基图像坐标和图像ROI区域中心点

的特征基图像坐标以及旋转角，让工件M位于工件N的正上方的预对接位置。According to the feature base image coordinates of the center point P _M of the workpiece M and the center point of the image ROI area, the computer

According to the feature base image coordinates and rotation angle, control the robot to grab the workpiece M, and then according to the feature base image coordinates of the center point P _N of the workpiece N and the center point of the image ROI area

The feature base image coordinates and rotation angle of , let the workpiece M be located in the pre-docking position just above the workpiece N.

Assembly of workpiece components:

The robot moves the workpiece M down to the docking position to make the workpiece M fully contact with the workpiece N. The six-dimensional sensor detects the torque and force between the workpiece M and the workpiece N. The computer reads the detection data of the six-dimensional sensor and compares the settings. The set threshold value is the torque threshold value of 10NM and the force threshold value of 30N; if it exceeds the set threshold value, the computer formulates an adjustment strategy through the PID controller, and the robot moves the workpiece M up by 10mm, and the robot adjusts according to the current adjustment value given by the PID controller. The position of the workpiece M, and then move the workpiece M to the docking position, the six-dimensional sensor detects again, and the computer determines the detection data. If it is greater than the set threshold, the robot adjusts the position of the workpiece M according to the PID controller until the six-dimensional sensor If the detected data is less than the set threshold, if not, the robot will move the workpiece M down by 4mm to complete the docking between the workpiece M and the workpiece N. The robot will return to the initial point and wait for the next workpiece assembly docking.

Claims (9)

1. the assembling method of a high-precision workpiece assembly, is characterized in that, comprises workpiece assembly, and described workpiece assembly comprises workpiece M and workpiece N, and described workpiece M and workpiece N are assembled according to the following steps: Preprocessing: place the workpiece M on the tooling to be grasped, and place the workpiece N on the docking tooling; install the image collector and the six-dimensional sensor on the robot, and set the image collector to take pictures of the workpiece M and the workpiece N respectively. point; set the alignment position and grasping position of the robot gripper on the workpiece M in the computer, set the pre-joint position and the docking position of the workpiece M and the workpiece N; set the initial point of the robot; set the workpiece M and the workpiece respectively The image ROI area of N; set the unit vector (x _e , y _e ); S1: Initialization, the robot moves to the initial point, and the robot establishes communication with the computer; S2: The robot drives the image collector to move to the fixed photographing point of the workpiece M, collects the image of the workpiece M, transmits the image of the workpiece M to the computer, and the computer analyzes the image of the workpiece M and determines the center point P _{M of the workpiece M} The feature base image coordinates and rotation angle θ _M , the rotation angle θ _M is the center point P _M of the workpiece M that points to the center point of the image ROI area of the workpiece M

the included angle between the constituent vector and the unit vector (x _e , y _e ); S3: The robot adjusts the gripper itself and moves to the alignment position of the workpiece _{M according to the feature base image coordinates and rotation angle θ M of} the center point PM of the workpiece M calculated in the previous step, and starts to descend until it reaches the grasping position. The robot controls the gripper to grasp the workpiece M; S4: The robot drives the image collector to move to the fixed photographing point of the workpiece N, collects the image of the workpiece N, transmits the image of the workpiece N to the computer, and the computer analyzes the image of the workpiece N and determines the center point P _{N of the workpiece N} The feature base image coordinates and rotation angle θ _N , the rotation angle θ _N is the center point P _N of the workpiece N pointing to the center point of the image ROI area of the workpiece N

the included angle between the constituent vector and the unit vector (x _e , y _e ); S5: The robot moves the workpiece M to the pre-docking position according to the feature base image coordinates and rotation angle θ _N of the center point P _N of the workpiece N calculated in the previous step; S6: The robot controls the workpiece M to connect with the workpiece N, the six-dimensional sensor detects the detection data when the workpiece M and the workpiece N are docked, and the computer controls the robot to repeatedly adjust the position of the workpiece M according to the detection data until the detection data is smaller than the installation threshold, and controls the workpiece. M is docked with the workpiece N, wherein the installation threshold at least includes a torque threshold and a force threshold; S7: The robot controls the gripper to release the workpiece M, and returns to the initial point for the next assembly. 2. The assembling method of high-precision workpiece assembly according to claim 1, wherein the center point P _M of the workpiece M, the image ROI area center point of the workpiece M

The center point P _N of the workpiece N, the center point of the image ROI area of the workpiece N

The initial point of the robot and the position of the image collector are calibrated by establishing a space coordinate system; Wherein, an image coordinate system is set in the picture collected at any position in the image collector, and a feature image coordinate is correspondingly set for all the features in the corresponding picture, and the robot is set in the base coordinate system, and the The feature image coordinates are converted by the coordinate system transformation matrix A _ij to obtain the feature base image coordinates in the base coordinate system. 3. the assembling method of high-precision workpiece assembly according to claim 2, is characterized in that, in step S2 and S4, the feature base image coordinates of the center point of the workpiece, the feature base image coordinates of the center point of the image ROI area are determined to be obtained. Take the centroid coordinates in the corresponding contour in the image, specifically: S-A1: The computer performs lens distortion correction on the image collected by the image collector to obtain the image after lens distortion correction, specifically:

Among them, the original coordinates of pixels before distortion correction are

The coordinates of the pixel point after distortion correction are (u, v), (k ₁ , k ₂ , k ₃ , p ₁ , p ₂ ) is the distortion coefficient, r is the radius of the image collector lens, and the computer obtains W ₁ through the checkerboard For the corresponding pixel points of the group, all the distortion coefficients are obtained by the least square method, and the image after lens distortion correction is obtained; S-A2: The computer converts the image after lens distortion correction from RGB color space to HSV color space, and performs multiple threshold segmentation and logical AND operation on the converted image to obtain the rough outline of the workpiece; S-A3: The computer smoothes the rough outline of the workpiece, converts the converted HSV image into a binary image, and then performs image morphological processing of image erosion and image expansion on the binary image in turn: Image erosion:

Image inflation:

in,

is the processed image, f(x,y) is the original image,

It is a structural element that makes the contour of the workpiece in the image smoother; S-A4: The computer uses the Canny algorithm to extract all the contours in the image obtained in step S-A3, and then obtains the precise contours in the image according to the area threshold; S-A5: The computer uses the gray centroid method to obtain the centroid coordinates of the center point of the precise contour in the image in the ROI area f

Among them, the center of mass coordinates

is the feature image coordinate in the image coordinate system. 4. the assembling method of high-precision workpiece assembly according to claim 3, is characterized in that, in step S-A2, computer converts the image after lens distortion correction from RGB color space to HSV color space The concrete steps are: S-B1, calculate the conversion factor:

Δ= _Cmax - _Cmin Among them, (R _P , G _P , B _P ) is the pixel value of point P in the RGB color space; S-B2, H calculation:

S-B3, S calculation:

S-B4, V calculation: VP _{= Cmax} Thus, the computer can obtain the value of point P in the HSV color space (H _P , S _P , _VP ); S-B5, the computer extracts the color of the outline as needed, and sets the color range in the HSV color space; Thereby, the image converted into HSV is obtained, and then multiple thresholds are divided, and the logical AND operation is performed to obtain the image containing the rough outline of the workpiece. 5. the assembling method of high-precision workpiece assembly according to claim 2, is characterized in that, described coordinate system conversion matrix A _ij is the coordinate system conversion matrix in Eye In Hand hand-eye calibration, and this coordinate system conversion matrix adopts DLT method Ask for, specifically: S-C1: The computer collects the characteristic image coordinates of at least three points in the image, and establishes the relationship between the image coordinate system and the base coordinate system according to the working principle of the image collector:

where (u, v) are the feature image coordinates, (X _r , Y _r , Z _r , 1) are the base coordinates of the robot, f is the focal length, and dx and dy represent the physical dimensions of each pixel on the X and Y axes, respectively , (u ₀ , v ₀ ) is the principal point coordinate; S-C2: Since the robot drives the image collector to take pictures of the workpiece at the same height, the Z value of the characteristic image coordinate in the image is a constant, and the above formula is simplified to get:

in,

is the coordinate system transformation matrix A _ij ; S-C3: The computer brings the collected feature image coordinates into S-C21: Bring the coordinate value of the X-axis in the feature image coordinates collected in step S-C1 into S-C21:

Obtain the first row of coefficients in the coordinate system conversion matrix A _ij ; Similarly, the coordinates of the Y-axis in the collected image coordinates are brought into S-C21 to obtain a complete coefficient matrix A _ij ; The characteristic image coordinates of the image coordinate system take the left The standard system transformation matrix A _ij is converted into the feature base image coordinates of the base coordinate system. 6. The assembling method of high-precision workpiece assembly according to claim 1, wherein the workpiece center point P _X on the workpiece is to the image ROI area center point

directed line segment

The rotation angle θ _X is the directional line segment

The included angle with the unit vector (x _e , y _e ), the rotation angle is specifically:

Among them, the robot adjusts the workpiece docking position according to the rotation angle θ _X ; or the clamping position; or the alignment position; or the pre-docking position. 7. the assembling method of high-precision workpiece assembly according to claim 1, is characterized in that, in step S6, robot controls workpiece M and workpiece N to carry out butt joint, six-dimensional sensor detects the detection data of workpiece M and workpiece N docking part, computer Control the robot to repeatedly adjust the position of the workpiece M according to the detection data until the detection data is less than the installation threshold, and the specific steps for controlling the workpiece M and the workpiece N to complete the docking are as follows: S-D1: The robot moves the workpiece M downward from the pre-docking position until it reaches the docking position, and the computer reads the detection data of the six-dimensional sensor; S-D2: The computer judges whether the detection data of the six-dimensional sensor is less than the installation threshold; if so, go to step S-D5; if not, go to step S-D3; S-D3: The computer analyzes the detection data of the six-dimensional sensor, the PID controller receives the detection data of the six-dimensional sensor and gives the current adjustment value, the robot moves the workpiece M upward W ₂ , the robot adjusts the position of the workpiece M according to the current adjustment value, the robot Move the workpiece M downward, and then reach the docking position, and the computer reads the detection data of the six-dimensional sensor; S-D4: The computer determines whether the detection data of the six-dimensional sensor is less than the installation threshold; if so, go to step S-D5; if not, return to step S-D3; S-D5: The robot moves the workpiece M down W ₃ to complete the docking of the workpiece M and the workpiece N. 8. the assembling method of high-precision workpiece assembly according to claim 7, is characterized in that, the described concrete steps of adjusting workpiece M by PID controller according to adjustment strategy are: S-D41: The control principle of the computer according to the PID control: Δu(k)=K _P [e(k)-e(k-1)]+K _I e(k)+K _D [e(k)-2e(k-1)+e(k-2)] Among them, Δu(k) is the current adjustment amount, e(k) is the current error, e(k-1) is the last error, e(k-2) is the last error, K _P proportional coefficient, K _I integral coefficient, K _D is the differential coefficient; S-D42: In order to shorten the setting period, the computer removes the differential link, and the above formula is simplified to: Δu(k)=K _P [e(k)-e(k-1)]+K _I e(k); S-D43: Of which: e(k)=D(k)-D Among them, D(k) is the current degree of the six-dimensional force sensor, and D is the ideal value; S-D44: In order to avoid noise response, the computer will continuously obtain the detection data of W ₄ six-dimensional sensors and take the average value as D(k), specifically:

S-D45: Set D=0, the above formula is simplified to: Δu(k)=K _P [D(k)-D(k-1)]+K _I D(k) The robot repeatedly adjusts the position of the workpiece M according to the current adjustment amount Δu(k) until the detection data of the six-dimensional sensor is smaller than the installation threshold. 9. A high-precision workpiece assembly, characterized in that it comprises the workpiece M and the workpiece N, the workpiece M is a cylinder, a bottom surface of the workpiece M is a clamped surface, and another bottom surface of the workpiece M is a cylinder. For the butting surface, the clamped surface of the workpiece M extends radially with at least three limit flanges (11), and the limit flanges open limit through holes (12) along the axial direction of the workpiece M, At least three butt flanges (13) extend from the butt surface of the workpiece M in the radial direction, the butt flanges are provided with butt holes (14) along the axial direction of the workpiece M, and the butt surfaces of the cylindrical body At least one blind butt hole (15) is opened on it; The workpiece N is a cylinder, and a bottom surface of the workpiece N is a butted surface, and the butted surface is adapted to the size and shape of the butted surface, and the butted surface of the workpiece N extends radially out at least three times. A butted flange (21) is provided with a butted through hole (22) along the N-axis of the workpiece, and the diameter of the butted through hole (22) is the same as that of the butted through hole (22). 14) The hole diameter is adapted, and the butted surface is also provided with a butt cylinder (23) coaxial with the workpiece N, the butt cylinder and the workpiece N are integrally formed, and the bottom surface radius of the butt cylinder (23) is the same as that of the workpiece N. the diameter of the blind butt hole (15), and the blind installation hole (24) is opened on the upper bottom surface of the butt cylinder (23); The workpiece M is matched with the workpiece N through the butting blind hole (15) and the butting cylinder (23), and the setting position of the butting through hole (14) is the same as that of the butted through hole (22). The setting positions correspond one-to-one.

Images