JPWO2009025271A1 - Robot control apparatus and control method - Google Patents

Abstract

The operator does not need to teach the position and posture of the measurement operation for each workpiece, and the three-dimensional position and posture are acquired with high accuracy to correct the deviation of the workpiece position and posture. A robot 101 including a position detection sensor 106 that acquires an actual measurement position of an edge of a rectangular parallelepiped workpiece 105 provided in an end effector 104 is taught in the following teaching steps (1) to (3) and the following reproduction steps (4) to Replay teaching in (6). (1) The first position and posture immediately before the end effector 104 grips the workpiece 105 are roughly taught. (2) Based on the first position and orientation and the shape dimension data of the workpiece 105, the calculated position and orientation of the workpiece 105 are obtained. (3) Based on the position and orientation of the workpiece 105, a robot operation command for actually measuring the workpiece position is generated. (4) The robot operates based on the operation command of the robot, and obtains the position and orientation of the workpiece 105 by actual measurement. (5) A deviation amount between the calculated position and orientation of the workpiece 105 and the actually measured position and orientation is obtained. (6) The position and posture of the grip position of the end effector 104 are corrected based on the deviation amount.

Description

  The present invention relates to a technology of a robot used for assembling parts.

The part assembly process performed at a car or home appliance manufacturing factory includes a work for fitting workpieces. When an assembly process including such a fitting operation is automated using a robot, the fitting operation needs to be taught to the robot. This teaching includes online teaching and offline teaching.
As an online teaching method, for example, there is a method in which an operator guides a robot and teaches an actual operation or procedure. As an off-line teaching method, there is a method in which a work program is created in advance by an external personal computer or the like, and is transferred to a robot controller for reproduction operation.
However, in the on-line teaching, there is a possibility that the position where the teaching is actually performed and the position where the work is performed at the time of the reproduction operation are shifted due to the positioning error of the workpiece. In offline teaching, there is a case where the origin of teaching data and the actual robot origin are misaligned, which makes it difficult to correct the teaching point position.
Therefore, Patent Document 1 discloses a method for teaching a teaching position with high accuracy by online teaching. In Patent Document 1, a position detection sensor for detecting the position of a workpiece is provided on a robot end effector, and the position detection sensor irradiates measurement light to the position detection region, and diffusely reflected light or regular reflected light returned from the position detection region. The position information of the workpiece is acquired based on the above.
In Patent Documents 2 and 3, (i) a robot is moved based on an operation file for measurement, a reference workpiece is sensed by a distance sensor and a measurement distance is stored, and then (ii) an actual workpiece is sensed to determine a measurement distance. The work operation file is corrected from the difference between the measurement distance and the measurement distance of the reference workpiece.
In Patent Document 4, (i) reference data is measured by teaching in advance the movement of the distance sensor relative to the workpiece, (ii) the work procedure and operation are taught in this state, and (iii) the distance measurement sensor is used when the workpiece is replaced. The workpiece is measured from three directions and corrected by comparison with the reference data.
Patent document 5 equips the wrist of a robot with the distance sensor which measures the distance to the specific point on a workpiece | work non-contactingly. The distance data obtained from the distance measured by the distance sensor is converted into a visual image instruction route and displayed on the screen, and the desired position indicated by the operator is converted into position data and indicated. A desired instruction route is acquired from the obtained distance data.
Patent Document 6 discloses an example in which the entire workpiece is sensed by a distance sensor and automatically input to an offline teaching program to create a welding line search job and a welding job.

JP-A-5-309590 (4th page, FIG. 1) JP-A-6-210580 (second page, FIG. 1) Japanese Patent Laid-Open No. 1-252381 (2nd page, FIG. 1) Japanese Patent Publication No. 6-008730 (2nd page, Fig. 1) JP-A-6-348330 (page 3, FIG. 1) Japanese Patent Laid-Open No. 9-183087 (page 4, FIG. 1)

In Patent Documents 1 and 2, it is impossible to measure the deviation of the posture of the workpiece by the method of scanning the workpiece in the cross direction or only in one direction. Also, it is cumbersome and time consuming to teach the operator how to scan the workpiece for each workpiece. In addition, an efficient three-dimensional position and orientation measurement operation itself is not clear, and an operator needs trial and error for each workpiece.
For Patent Documents 3 and 4, it is necessary to create a measurement operation file (job) for each workpiece, and the measurement direction with respect to the workpiece also needs to be measured from three directions. The measurement posture cannot be taken depending on the degree and interference with the surrounding environment.
With respect to Patent Document 5, it takes time for the operator to instruct how to scan the workpiece for each workpiece, and it takes a lot of time to recognize the workpiece shape.
In Patent Document 6, it takes time to scan the entire workpiece. It is necessary to input the entire size in advance. Two sensing steps are required: full scan and weld line scan.
The present invention has been made in view of such problems, and it is not necessary for the operator to teach the position and orientation of the measurement operation for each workpiece, and the three-dimensional position and orientation are obtained with high accuracy. The purpose is to correct the deviation of the position and posture of the workpiece.

In order to solve the above problem, the present invention is configured as follows.
The invention according to claim 1 is a memory for storing a first position and posture immediately before the end effector grips the workpiece, a workpiece dimension input unit for inputting dimension data of the workpiece, and the first effector. An operation pattern generator for generating an operation command for a position detection sensor provided in an end effector to measure the position and orientation of the workpiece based on the position and orientation and the input dimension data of the workpiece; The robot is operated based on an operation command generated by an operation pattern generator, the position and orientation of the workpiece are measured, and the workpiece obtained based on the first position and orientation and shape data of the workpiece A deviation amount between the calculated position and posture and the actually measured position and posture is obtained, and the end error is calculated based on the deviation amount. And control means for correcting the position and orientation of Ekuta, it is characterized in that it comprises a.
The invention according to claim 2 is characterized in that the position detection sensor is a one-dimensional position detection sensor.
The invention according to claim 3 is characterized in that the position detection sensor is a two-dimensional position detection sensor.
The invention described in claim 4 is characterized in that when the workpiece teaching position is stored, one point of the same position and posture as the end effector acting on the workpiece is registered in actual work. .
The invention described in claim 5 is characterized in that the workpiece dimension data is input from a teaching device.
The invention described in claim 6 is characterized in that the workpiece dimension data is data based on CAD data.
According to a seventh aspect of the present invention, the workpiece has a rectangular parallelepiped shape, and the operation command is determined by the long sides of the position detection sensor that are separated by a predetermined distance from four vertices of a rectangle that is a measurement target surface of the workpiece. It is an operation command for measuring the positions on the upper four edges and the positions on the two edges which are the midpoints of the two short sides of the rectangle.
According to an eighth aspect of the present invention, the workpiece has a rectangular parallelepiped shape, and the operation command is received while the two-dimensional position detection sensor moves in a short side direction of a rectangle serving as a measurement target surface of the workpiece. It is an operation command for measuring a three-dimensional shape of a workpiece.
According to a ninth aspect of the present invention, the workpiece has a cylindrical shape having a measurement target surface as an upper surface, and the operation command is generated by a parallel straight line that is equidistant from the circle center of the upper surface in the opposite direction, and the upper surface. Is an operation command for measuring the positions on the four edges where the crossing points.
In the invention according to claim 10, the work is in a cylindrical shape,
The operation command is an operation command in which the two-dimensional position detection sensor scans an upper surface, which is a measurement target surface of the workpiece, in a certain direction.
In the invention described in claim 11, when the amount of positional deviation of the workpiece exceeds a preset threshold value, the robot is operated again based on the operation command to obtain the position and posture by actual measurement of the workpiece, Control means for recalculating the amount of deviation between the calculated position and posture of the workpiece obtained based on the first position and posture and the shape data of the workpiece and the position and posture measured again. It is characterized by that.
According to a twelfth aspect of the present invention, when the amount of positional deviation of the workpiece exceeds a preset threshold, the motion pattern generator changes the motion direction to a reverse or orthogonal direction, and the position detection sensor Regenerates an operation command for detecting the actual measurement position of the workpiece, operates the robot based on the regenerated operation command, obtains a position and orientation by actual measurement of the workpiece, and determines the first position. And control means for recalculating the amount of deviation between the calculated position and posture of the workpiece obtained based on the posture and the shape dimension data of the workpiece and the position and posture obtained by the re-measurement. Is.
The invention according to claim 13 includes a robot provided with an end effector for gripping a workpiece, and a control device for controlling the robot, wherein the control device detects a position and a posture of the workpiece. A sensor, a memory for storing a first position and orientation immediately before the end effector grips the workpiece, a workpiece dimension input unit for inputting dimension data of the workpiece, and the first position and orientation; An operation pattern generator for generating an operation command for a position detection sensor provided in an end effector to measure the position of the workpiece based on the input dimension data of the workpiece, and the operation pattern generator A command generator for generating a motion command for the robot based on the generated motion pattern, the first position and posture and The amount of deviation between the position and posture of the workpiece calculated from the shape data of the workpiece and the position and posture actually measured by the position detection sensor is obtained, and the end effector of the end effector is obtained based on the amount of deviation. The position and orientation are corrected.
The invention described in claim 14 is characterized in that the position detection sensor is provided in the end effector.
The invention described in claim 15 is characterized in that the position detection sensor is a one-dimensional position detection sensor.
The invention described in claim 16 is characterized in that the position detection sensor is a two-dimensional position detection sensor.
In the invention described in claim 17, the workpiece has a rectangular parallelepiped shape, and the operation command is a long side where the position detection sensor is separated by a predetermined distance from each of four vertices of a rectangle which is a measurement target surface of the workpiece. It is an operation command for measuring the positions on the upper four edges and the positions on the two edges which are the midpoints of the two short sides of the rectangle.
In the invention described in claim 18, the workpiece has a cylindrical shape having a measurement target surface as an upper surface,
The operation command is an operation command for measuring positions on four edges at which a parallel straight line that is equidistantly spaced in the opposite direction from the circle center of the upper surface and the upper surface intersect each other. is there.
The invention described in claim 19 is characterized in that a robot including an end effector for gripping a workpiece and a position sensor for detecting the position of the workpiece is taught and reproduced in the following steps. Is. (1) A first position and posture immediately before the end effector grips the workpiece are roughly taught. (2) The calculation position and orientation of the workpiece are obtained based on the first position and orientation and the shape data of the workpiece. (3) Based on the position and posture of the workpiece, an operation command of the robot for actually measuring the workpiece position is generated. (4) The robot is operated based on the operation command, and the position of the workpiece is measured by the position detection sensor. (5) The amount of deviation between the calculated position and orientation of the workpiece and the position and orientation of the workpiece measured by the position detection sensor is obtained. (6) The position and posture of the end effector at the gripping position are corrected based on the amount of deviation.
The invention according to claim 20 is characterized in that the position detection sensor is provided in the end effector.

According to the invention described in claims 1, 13, and 14, there is an effect that an accurate workpiece position can be automatically acquired if the worker teaches the workpiece position roughly once.
According to the second and fifteenth aspects of the present invention, since the three-dimensional position and posture of the workpiece can be measured with a one-dimensional distance sensor, there is an effect that it is inexpensive and signal processing is simple.
According to the invention described in claims 3 and 16, since the three-dimensional position and orientation of the workpiece can be measured by the two-dimensional distance sensor, the measurement operation time is short, the cost is low, and the signal processing is simple. is there.
According to the fourth aspect of the present invention, since the measurement operation can be generated by registering only one position and posture at which the end effector acts on the work in actual work, for example, gripping the work without gripping the work in assembly work. The correct detailed workpiece position can be acquired simply by registering the position of the posture. Therefore, there is an effect that troublesome teaching work such as actually gripping the workpiece is not required.
According to the fifth aspect of the present invention, the workpiece dimension data has only to be input as a numerical value for the outer dimension of the workpiece.
According to the sixth aspect of the present invention, since CAD data is used as the workpiece dimension data, there is an effect that it is possible to generate an operation pattern that is accurate and takes into account surface irregularities.
According to the invention described in claims 7 and 17, when the workpiece shape is a rectangular parallelepiped, the position detection sensor has four edge positions on the long side of the workpiece and two edge positions on the short side. Is an operation pattern for measuring. Therefore, there is an effect that the three-dimensional position and orientation can be measured with a small number of operations.
According to the invention described in claim 8, when the workpiece shape is a rectangular parallelepiped, the two-dimensional distance sensor moves in the direction of the short side of the rectangle that becomes the measurement target surface of the workpiece, and the three-dimensional shape of the workpiece. It is an operation pattern for measuring a shape. Therefore, there is an effect that a three-dimensional position and posture can be measured with only one measurement operation.
According to the invention described in claims 9 and 18, the operation pattern when the workpiece shape is a cylinder is obtained by measuring twice the workpiece cross section that is equally spaced from the center of the circle from one direction with respect to the workpiece. In order to calculate the amount of positional deviation of the workpiece, there is an effect that the amount of positional deviation of the workpiece can be calculated by measuring the three-dimensional position with a small number of measurements of only two measurements.
According to the tenth aspect of the present invention, the operation pattern when the workpiece shape is a cylinder is measured continuously once from one direction with respect to the workpiece by the two-dimensional sensor, and the workpiece data is measured from the measurement data. Since the positional deviation amount is calculated, there is an effect that the three-dimensional position can be measured by only one measurement and the workpiece positional deviation amount can be calculated.
According to the eleventh aspect of the present invention, when the amount of positional deviation of the workpiece exceeds a preset threshold value, the position and orientation of the workpiece are actually obtained again based on the operation command, and the first position and Since the amount of deviation between the calculated position and posture of the workpiece obtained based on the posture and the shape dimension data of the workpiece and the position and posture obtained by re-measurement is obtained again, there is an effect of preventing work failure due to a measurement error.
According to the twelfth aspect of the present invention, when the amount of positional deviation of the workpiece exceeds a preset threshold, the movement pattern generator changes the moving direction of the position detection sensor to the reverse direction or the orthogonal direction, The position detection sensor regenerates an operation command for detecting the actual measurement position of the workpiece. Based on this operation command, the robot is operated to obtain the position and orientation of the workpiece, and the deviation amount between the calculated position and orientation and the actually measured position and orientation is obtained again. Therefore, even when mistakes frequently occur in the measurement in the same direction, there is an effect that it is possible to prevent a work failure due to the measurement mistake.
According to the 19th and 20th aspects of the present invention, if the worker roughly teaches the workpiece position, there is an effect that an accurate workpiece position can be automatically obtained.

The block diagram of the robot system in 1st Example of this invention Detailed view of position teaching example in the first embodiment of the present invention The figure which shows the measurement operation | movement in 1st Example of this invention. The figure which shows the calculation process in 1st Example of this invention. The figure which shows the measurement operation | movement in 2nd Example of this invention. The figure which shows the measurement operation | movement in 3rd Example of this invention. The figure which shows the measurement operation | movement in 4th Example of this invention. Work flow chart in the fifth embodiment of the present invention

Explanation of symbols

101 Robot 102 Controller 1021 Command Generation Unit 1022 Servo Control Unit 1023 Memory 1024 Work Dimension Input Unit 1025 Operation Pattern Generation Unit 103 Teaching Device 104 End Effector (Gripper)
105 Work 106 Position detection sensor

  Hereinafter, specific examples of the present invention will be described with reference to the drawings.

FIG. 1 is a configuration diagram of a robot system according to a first embodiment of the present invention. This system includes a robot 101, a controller 102 that controls the operation of the robot 101, a teaching device 103, an end effector 104 (herein referred to as a gripper) disposed at the hand of the robot, and position detection disposed on or around the gripper 104. A sensor 106 is provided. In this embodiment, the position detection sensor is a two-dimensional sensor such as a two-dimensional laser displacement sensor for detecting the workpiece 105 or a one-dimensional sensor such as a laser displacement sensor.
In the controller 102, a memory 1023 for storing teaching data or the like by inputting an operator's enter key or the like from the teaching device 103 during teaching work, a workpiece dimension input unit 1024 for inputting workpiece dimension data, and storage in the memory 1023. The motion pattern generation unit 1025 generates a robot motion pattern so that the position detection sensor can perform a specific measurement operation from the workpiece teaching position and the workpiece dimension data input from the workpiece dimension input unit 1024, and the teaching stored in the memory 1023 A command generation unit 1021 that generates a robot operation command from data or an operation pattern generated by the operation pattern generation unit 1025, and a servo control unit 1022 that drives a motor mounted on the robot based on the operation command are provided. . Data is input to the workpiece dimension input unit 1024 via a display on the teaching device 103 and key operations.

The process from the teaching work of the assembly / fitting work using the robot system to the regeneration operation will be described. A case is assumed in which an operation of gripping a rectangular parallelepiped shaped workpiece (including a cube) shown in FIG. 2 is taught. FIG. 2 shows a state where the gripper 104 is positioned above the workpiece 105.
A procedure for performing only the teaching of the rough position of the gripping position at the time of teaching, performing the scanning operation for each workpiece during the reproduction operation, and obtaining the gripping position correction amount will be described.
The teaching work can be divided into the following four stages (A to D).

A. Rough Position Teaching The operator guides the robot 101 and roughly teaches the position and posture immediately before the gripper 104 actually grips the workpiece 105, whereby the teaching data is stored in the memory 1023 in the controller 102. Is done. With this teaching data, the approximate position and posture of the workpiece 105 relative to the robot 101 can be obtained. At this time, as shown in FIG. 2, it is not necessary to actually position the work at the center position of the gripper opening and closing, and a rough teaching to the extent that the work 105 is inserted between the gripper claws may be used. Therefore, it is not necessary to match the center of the workpiece 105 and the gripper 104 so that the workpiece is actually gripped, and the burden on the operator can be reduced.

B. Input of Work Shape Dimension An operator inputs a work shape dimension from a result of measuring a drawing or an actual work through a display on the teaching device 103 and key operation. The input shape dimension is stored in the workpiece dimension input unit 1024 in the controller 102. Specifically, when the gripper 104 grips the center position WC (Work Center) of the workpiece 105, the external dimensions of the rectangular parallelepiped and the workpiece grip height position (Z-axis direction position from the lower surface of the workpiece 105) are input. Just do it. However, when the gripping center position HC (Hand Center) of the gripper portion of the gripper 104 matches the workpiece center position WC, it is not necessary to input the workpiece gripping height position. On the other hand, when the gripper 104 does not grip the center position WC of the workpiece 105, the offset amount (shift amount in the X-axis or Y-axis direction) is input.

C. Command addition The operator adds a scan operation command before the operation command registered as the gripping position in the work program. Alternatively, a tag command for performing a scanning operation may be added to the operation command.

D. Generation of scan motion When a command for scan motion is added, the motion pattern generation unit 1025 calculates the robot from the position and orientation of the gripping position stored in the memory 1023 and the shape dimension of the workpiece stored in the workpiece dimension input unit 1024. An operation pattern for moving the position detection sensor 106 mounted on 101 is automatically generated. Specifically, it is generated by the following procedure.

D-1. Calculate the workpiece position and posture in robot coordinates Since the gripping position and gripping posture of the gripper 104 are stored as the respective axis angles of the robot, the position and posture of the gripper 104 at the gripping center position HC in the robot coordinate system Ask. This is obtained by converting it into a robot coordinate value by forward conversion or the like. The position and posture of a rectangular parallelepiped workpiece are expressed using eight points of the workpiece. Therefore, from the position and posture of the gripper 104 in the robot coordinate system at the grip center position HC and the workpiece shape dimension, the calculation position and posture of the workpiece 105 are represented by eight points in the robot coordinates.
The calculation position and posture of the workpiece 105 are deviated from the actual position and posture of the workpiece. Accordingly, the eight vertices representing the calculation position and posture of the workpiece 105 are approximate positions of the actual eight vertices of the workpiece 105.
Note that the position and orientation of the workpiece 105 may be expressed by a method other than the eight vertices of the workpiece 105.

D-2. Teaching edge position detection operation on the long side (teaching the scanning operation reference point in the long side direction)
FIG. 3 shows a view of the workpiece 105 as viewed from above. The XYZ axes represent the work coordinate system. The upper surface of the workpiece 105 is a measurement surface of the position detection sensor 106. The long sides W1 and W2 of the workpiece 105 are equally divided into four, and the positions of 1/4 from both ends are set to a, b, c, d, respectively, and coordinate data is obtained from the positions of the four corners of the robot coordinates. A virtual point located outside the position deviation allowable error ΔY of the workpiece 105 from these positions a to d is defined as a ₁ , b ₁ , c ₁ , d ₁ (b ₁ , d ₁ are not shown), Points that are positioned above the height adjustment amount ΔZ so that these positions a _{1 to} d ₁ fall within the measurement distance of the position detection sensor 106 are defined as a ₂ , b ₂ , c ₂ , and d ₂ .
As shown in the cross section A, when the position detection sensor 106 is a two-dimensional sensor, the sensor scanning direction is taken in the Y-axis direction, and the sensor distance measuring direction is taken in the Z-axis direction. Next, when performing scanning operation in the short side direction, a ₁ and c ₁ , b ₁ and d ₁ can be measured (a ₁ and c ₁ etc. are inside the dotted line that is the measurement range in one scan) And so on, the sensor measurement ranges of the position detection sensor 106 are respectively appropriate positions a ₃ and b ₃ (b ₃ is not shown, but in the cross section B between a ₂ and c ₂ and b ₂ and d ₂ , a ₃ is a position corresponding to. become) position and determining the height. Then, the two points a ₃ and b ₃ are stored in the memory 1023 as teaching points serving as a reference for the scanning operation.
Here, the long sides W1 and W2 are equally divided into four to determine the scanning operation points a to d. However, as long as the distance from the four corners is larger than the positional deviation allowable error ΔX, any scanning operation may be performed. . The reason why the positional deviation allowable error ΔX is taken into consideration is to prevent the position detection sensor 106 from scanning the short side of the workpiece 105 when the positional deviation amount in the X-axis direction is too large.
When the position detection sensor 106 is a one-dimensional sensor, the scanning operation in the short side direction is a straight line that passes through a ₂ and c ₂ , and b ₂ and d _2. The positions a _{2 to} d ₂ are stored in the memory 1023 as teaching points.

D-3. Teaching edge position detection operation on the short side (teaching the scanning operation reference point in the short side direction)
A position obtained by dividing the short side D1 and D2 of the workpiece 105 into two equal parts is obtained as e and f, respectively, and a straight line passing through e and f is obtained. On this straight line, the virtual points e ₁ and f ₁ located outside are set by the positional deviation allowable error ΔX of the workpiece 105, and these positions e ₁ and f ₁ are within the measurement distance of the position detection sensor 106. As shown, e ₂ and f ₂ are points positioned above the height adjustment amount ΔZ.
As shown in the cross section C, when the position detection sensor 106 is a two-dimensional sensor, the sensor scanning direction is taken in the X-axis direction, the sensor distance measuring direction is taken in the Z-axis direction, and e _{1 is} a scanning operation in the long side direction. And an appropriate position e ₃ between e ₂ and f ₂ is obtained as the sensor measurement position of the position detection sensor 106 so as to pass through f ₁ and f ₁ . Then, as taught point this e ₃ as a reference for scanning operation and stored in the memory 1023.
Here, the short side D is equally divided into two to determine the points e and f of the scanning operation. However, as long as the distance from the four corners is larger than the positional deviation allowable error ΔY, any scanning operation point may be used. The reason why the positional deviation allowable error ΔY is taken into account is that the position detection sensor 106 cannot scan the long side of the workpiece 105 when the positional deviation amount in the Y-axis direction is too large.
When the position detection sensor 106 is a one-dimensional sensor, the scanning operation in the long side direction is a straight line passing through e ₂ and f ₂ , and therefore the sensor measurement start and end positions e ₂ and f _{2 of the} position detection sensor 106 are detected. Are stored in the memory 1023 as teaching points.
The teaching position for the scanning operation can be automatically generated by the above procedure.

  The procedure during the regeneration operation can be divided into the following three stages (E to G).

E. Scan operation during regenerative operation During the regenerative operation, the work program is executed, and the scan operation command is processed immediately before gripping the work 105 with the gripper 104, so that the position detection sensor 106 measures the work 105 three times. To perform the operation. Specifically, when the position detection sensor 106 is a two-dimensional sensor, scanning is performed once at each of the positions a ₃ , b _3, and e ₃ . In the case where the position detection sensor 106 is a one-dimensional sensor, measurement is performed by scanning in a linear motion from a ₂ to c ₂ , from b ₂ to d ₂ , and from e ₂ to f ₂ .

F. Correction amount calculation By the measurement of the position detection sensor 106 three times, the actual edge measurement positions a _{m to} f _m with respect to the edge positions a to f in the calculation of the long side and the short side of the workpiece 105 in the robot coordinates are obtained. The difference between the edge calculation positions a to f and the actual measurement positions a _{m to} f _m is the amount of workpiece displacement.
Specifically, obtains a rectangle having vertices operational position a~d and the actual measurement position _a m to d _m respectively, the following equation (1) to (3), the deviation amount of the posture of the workpiece 105 dRx, dRy , DRz can be obtained.
The inclinations of the workpiece 105 around the x, y, and z axes of the robot coordinates obtained using the edge calculation positions a to f are respectively obtained using dRx1, dRy1, and dRz1, and similarly the actual measurement positions a _{m to} f _m. Are dRx2, dRy2, and dRz2, respectively. As shown in FIG. 4A, dRx1 is obtained from the angle formed by the edge calculation positions a and c with respect to the Y axis. dRx2 is determined from the angle of the actual measurement position _{a m} and _{c m} against Y axis. The difference between dRx1 and dRx2 is the posture deviation amount dRx. In the same manner, dRy and dRz can also be obtained.

dRx = dRx1-dRx2 = tan ⁻¹ ((az−cz) / (ay−cy))
−tan ⁻¹ ((az _m −cz _m ) / (ay _m −cy _m )) Formula (1)

dRy = dRy1-dRy2 = tan ⁻¹ ((fz-ez) / (fx-ex))
−tan ⁻¹ ((fz _m −ez _m ) / (fx _m −ex _m )) Equation (2)

dRz = dRz1-dRz2 = tan ⁻¹ ((by-ay) / (bx-ax))
−tan ⁻¹ ((by _m −ay _m ) / (bx _m −ax _m )) Equation (3)

here,
ax, ax _m : calculation position of edge position a in X direction, actual measurement position ay, ay _m : calculation position of edge position a in Y direction, actual measurement position az, az _m : calculation position of edge position a in Z direction , the actual measurement position bx, bx _m: calculating the position of the X-direction edge position b, the actual measurement position by, by _m: Y direction of the operation position of the edge position b, the actual measurement position cy, cy _m: edge position c Y Directional calculation position, actual measurement position cz, cz _m : Z position calculation position of edge position c, actual measurement position ex, ex _m : X position calculation position of edge position e, actual measurement position ez, ez _m : edge Calculation position of position e in Z direction, actual measurement positions fx, fx _m : calculation position of edge position f in X direction, actual measurement positions fz, fz _m : calculation position of edge position f in Z direction, actual measurement position

Next, displacement amounts dX, dY, dZ of the position of the workpiece 105 are obtained. First, as shown in FIG. 4B, the values of the actual measurement positions a _{m to} f _m are corrected using the previously obtained posture deviation amounts dRx to dRz, and the same posture as the edge calculation positions a to f. The actual measurement correction positions a _{m1 to} f _m1 corrected to are obtained. Thereafter, as shown in FIG. 4 (c), the deviation amount of the points is obtained for each axis of the robot coordinate system, and the total number of points is averaged for each axis, and the following equations (4) to (6) are obtained. The amount of positional deviation at each point can be obtained.

dX = ((ax _m 1-ax) + (bx _m 1-bx) + (cx _m 1-cx) + (dx _m 1-dx)
+ (Ex _m 1-ex) + (fx _m 1-fx)) / 6 Formula (4)

dY = ((ay _m 1-ay) + (by _m 1-by) + (cy _m 1-cy) + (dy _m 1-dy)
+ (Ey _m 1-ey) + (fy _m 1-fy)) / 6 Formula (5)

dZ = ((az _m 1-az) + (bz _m 1-bz) + (cz _m 1-cz) + (dz _m 1-dz)
+ (Ez _m 1-ez) + (fz _m 1-fz)) / 6 Formula (6)

here,
bz: Z position calculation position cx of edge position b: X position calculation position ey of edge position c: Y position calculation position fy of edge position e: Y position calculation position dx, dx _{m of} edge position f 1: Calculation position of edge position d in X direction, actual measurement posture correction position dy, dy _m 1: Calculation position of edge position d in Y direction, actual measurement posture correction position dz, dz _m 1: Calculation of edge position d in Z direction Position, actual measurement posture correction position ax _m 1: Actual measurement posture correction position ay _{m of} edge position a in the X direction Actual measurement posture correction position az _{m of} edge position a in the Y direction 1: Z direction of edge position a in the Z direction Actual measurement posture correction position bx _m 1: Actual measurement posture correction position by _{m in} the X direction of the edge position b 1: Actual measurement posture correction position cy _{m in} the Y direction of the edge position b 1: Actual measurement in the Y direction of the edge position c attitude correction position cz _m 1: edge position c Direction of the actual measurement posture correction position ey _m 1: actual measurement posture correction position in the Y direction edge position e ez _m 1: actual measurement posture correction position in the Z direction edge position e fy _m 1: edge position f in the Y-direction Actual measurement posture correction position fz _m 1: Actual measurement posture correction position in the Z direction of the edge position f

Note that the calculation load may be reduced by reducing the representative point to one point to obtain the positional deviation amount. Rather than obtaining the actual measurement correction positions a _{m1 to} f _m1 corrected to the same posture as the edge calculation positions a to f by correcting the values of the actual measurement positions a _{m to} f _m, the values of the edge calculation positions a to f are calculated. correcting the edge mended to the actual measurement position a _m ~f _m the same posture calculating corrected position may be determined.

G. Correction operation during reproduction operation In the gripping operation after the command of the scanning operation, the posture deviation amounts dRx, dRy, dRz of the workpiece 105 and the positional deviation amounts dX, dY, obtained by the above formulas (1) to (6). The work 105 can be accurately gripped by the gripper 104 by correcting the position and posture of the gripping position using dZ.

In this embodiment, the operator manually inputs the workpiece dimension data from the workpiece dimension input unit 1024. However, detailed data may be input using CAD data. In this case, when measuring the long side cross-section, it is possible to select a position avoiding the unevenness as the edge measurement position in consideration of the unevenness on the workpiece.
Further, in this embodiment, only a rough teaching of the gripping position is executed at the time of teaching, and a scanning operation is performed for each workpiece during the reproduction operation, and the gripping operation is performed by obtaining the gripping position correction amount. This procedure is applied when the workpiece positioning error is large even during the regeneration operation. However, if the positioning error is small, it is taught roughly at the time of teaching, immediately after that, a scan operation command is generated, and the scanning operation is performed on the spot to obtain a detailed teaching position, and the correction amount of the teaching position is calculated. Reflect in work program. A configuration in which the scanning operation is not performed during the regeneration operation may be used.

  According to this embodiment, one work teaching position is stored, and dimension data of the work is input. From the workpiece teaching position and workpiece dimension data, an operation pattern of the robot is generated so that the position detection sensor arranged in the gripper 104 can perform a specific measurement operation, and the robot is operated based on the operation pattern. Therefore, the position of the workpiece can be acquired by the position detection sensor, and the amount of deviation between the position and orientation of the workpiece can be calculated and corrected. Therefore, if the worker roughly teaches the workpiece position, it is possible to automatically acquire and correct the workpiece position automatically.

In the first embodiment, the one-dimensional or two-dimensional position detection sensor performs the measurement operation three times in total, that is, the scan operation in the short side direction twice and the scan operation in the long side direction once. The amount of displacement between position and posture was obtained. In this embodiment, a case where a two-dimensional position detection sensor is used for dynamic measurement using a three-dimensional sensor will be described.
As shown in FIG. 5, when the position detection sensor 106 is a two-dimensional sensor, the position detection sensor 106 is moved at a right angle to the long side of the workpiece 105 to continuously measure the entire workpiece. Then, the three-dimensional shape of the workpiece 105 is obtained. Thereby, the correction amount of the position and orientation can be obtained. Start and end positions in this case is _{e 31} and _{e 32.} e ₃₁ and e ₃₂ are positions of the points where the sensor measurement position e ₃ obtained in the first embodiment is arranged outside the workpiece 105 by the amount of allowable deviation ΔY of the workpiece 105 in the Y-axis direction. The position detection sensor 106 continuously scans the positions e ₃₁ to e ₃₂ to acquire three-dimensional data.
From the measurement data of the three-dimensional shape of the workpiece 105, straight lines of the long side and the short side of the workpiece upper surface are obtained, and processing for obtaining the edge actual measurement positions a _{m to} f _m is added from these straight lines. After the edge actual measurement positions a _{m to} f _m are obtained, the same calculation as in the first embodiment is performed.

According to the present embodiment, since the three-dimensional position and orientation can be measured by one measurement from one direction, the measurement operation time can be shortened.
In the above embodiment, it is clear that even if the workpiece 105 is a cube, it can be implemented by assuming a long side and a short side.

In the first embodiment, the workpiece 105 has a rectangular parallelepiped shape (including a cube). In this embodiment, as shown in FIG. 6, a case where the workpiece 105 has a cylindrical shape will be described.
As shown in the first embodiment, among the four stages of the teaching work, the teaching of the rough position of A and the addition of the command of C are the same in this embodiment. In the workpiece shape input of B, the external dimensions (for example, diameter and height) of the cylinder are input. The procedure for generating the scanning operation for D includes the same procedure as in the first embodiment. The same procedure as that of the first embodiment is applied to the robot 101 from the position and posture of the gripping position stored in the memory 1023 by the motion pattern generation unit 1025 in the controller 102 and the shape dimension of the workpiece stored in the workpiece dimension input unit 1024. This is a procedure for automatically generating a command for moving the mounted position detection sensor 106. A procedure different from that of the first embodiment will be described below.

D-1. The position and posture of the workpiece in the robot coordinate system are obtained by converting the position and posture of the workpiece in the robot coordinate system by performing transformation such as calculation order transformation. From the position and orientation of the gripper 104 in the robot coordinates at the grip center position HC and the workpiece shape dimensions, the calculation position and orientation of the workpiece 105 can be expressed in robot coordinates. The calculation position and orientation of the workpiece 105 are represented by two points in the robot coordinate system (circle centers on the upper surface and the lower surface of the workpiece 105).
The calculation position and posture of the workpiece 105 are deviated from the actual position and posture of the workpiece. Therefore, the two circular center points representing the calculation position and posture of the workpiece 105 are the approximate positions of the two circular center points of the actual workpiece 105.
Note that the position and orientation of the workpiece 105 may be expressed by a method other than the two center points of the circle of the workpiece 105.

D-2. Teaching of Edge Position Detection Operation on Circumference FIG. 7 shows a view of the workpiece 105 as viewed from above. The XYZ axes represent the work coordinate system. The upper surface of the workpiece 105 is a measurement surface of the position detection sensor 106. The thin line circle is in the ideal position. The bold circle is the position where the workpiece is actually placed. For the circle at the ideal position, divide the radius of the workpiece 105 into two equal parts, draw two straight lines along the Y axis at a position ½ in the X axis direction from the circle center WC, and each position intersects the circumference. , A, b, c, d. Let e and f be positions where a straight line is drawn from the circle center WC to the X axis to intersect the circumference. The positions where a straight line is drawn from the circle center WC to the Y axis and intersect with the circumference are g and h. However, as long as different workpiece cross-sectional positions that are equidistantly spaced in the opposite direction from the circle center WC can be obtained, the position is not limited to a ½ position, and may be an arbitrary position.
A point on a circle (thick circle) corresponding to a, b, c, d, e, f, g, h on the circle representing the ideal position and indicating the position where the workpiece is actually placed is indicated by aa, bb, cc, dd, ee, ff, gg, hh.
A virtual point located outside the position deviation allowable error ΔY of the workpiece 105 from these positions a to d is defined as a ₁ , b ₁ , c ₁ , d ₁ (b ₁ , d ₁ are not shown). . Points that are positioned above the height adjustment amount ΔZ so that these positions a _{1 to} d ₁ fall within the measurement distance of the position detection sensor 106 are defined as a ₂ , b ₂ , c ₂ , and d ₂ .
As shown in the cross section A, when the position detection sensor 106 is a two-dimensional sensor, the sensor scanning direction is taken in the Y-axis direction, and the sensor distance measuring direction is taken in the Z-axis direction. Next, when performing a scanning operation in the Y-axis direction, a ₁ and c ₁ , b ₁ and d ₁ can be measured, respectively (a ₁ and c _1, etc. enter the dotted line within the measurement range in one scan. And so on, the sensor measurement ranges of the position detection sensor 106 are respectively appropriate positions a ₃ and b ₃ (b ₃ is not shown, but in the cross section B between a ₂ and c ₂ and b ₂ and d ₂ , a ₃ is a position corresponding to. become) position and determining the height. The two points a ₃ and c ₃ are stored in the memory 1023 as teaching points serving as a reference for the scanning operation.
Here, the points a to d of the scanning operation are determined by dividing the radius into two, but if the distance from e or f is larger than the positional deviation allowable error ΔX, it is determined according to a preset fixed value. Also good.
When the position detection sensor 106 is a one-dimensional sensor, the scanning operation is a straight line that passes through a ₂ and c ₂ , and b ₂ and d ₂ , and therefore the sensor measurement start and end positions a ₂ to a _{2 of the} position detection sensor 106. the d ₂ as teaching points, and stores in the memory 1023.
As described above, the teaching position for the scanning operation can be automatically generated.

  The procedure during the regeneration operation can be divided into three stages (E to G) as shown in the first embodiment. The correction operation during the regeneration operation of G is the same as that in the first embodiment.

E. Scan operation during regenerative operation During the regenerative operation, the work program is executed, and the scan operation command is processed immediately before gripping the work 105 with the gripper 104, so that the position detection sensor 106 measures the work 105 twice. To perform the operation. Specifically, when the position detection sensor 106 is a two-dimensional sensor, scanning is performed once at each of the positions a ₃ and b ₃ . When the position detection sensor 106 is a one-dimensional sensor, measurement is performed by performing a linear operation from a ₂ to c ₂ and from b ₂ to d ₂ .

F. Correction amount calculation By measuring twice by the position detection sensor 106, the edge actual measurement positions aa to dd with respect to the edge positions a to d in the calculation of the circumference of the workpiece 105 in the robot coordinates are obtained. The difference between the edge calculation positions a to d and the actual measurement positions aa to dd is the amount of workpiece displacement.
A specific calculation method is as follows. Of the four positions, three are used (except for d here), and the circle center position is obtained from these positions. Here, the coordinates of these three points are a (X1, Y1), b (X2, Y2), and c (X3, Y3). The midpoint coordinates (X4, Y4) of the line segment ab are obtained.
X4 = (X1 + X2) / 2 Formula (7)
Y4 = (Y1 + Y2) / 2 formula (8)
A = tan ⁻¹ {(Y2−Y1) ÷ (X2−X1)} + 90 Equation (9) where A is an angle with respect to the X axis of a line that passes through the midpoint coordinates (X4, Y4) and is orthogonal to the line segment ab )
It becomes.
Similarly, the midpoint coordinates (X5, Y5) of the line segment bc are obtained.
X5 = (X2 + X3) / 2 Formula (10)
Y5 = (Y2 + Y3) / 2 Formula (11)
B = tan ⁻¹ {(Y3−Y2) ÷ (X3−X2)} + 90 Equation (12) where B is an angle with respect to the X axis of a line that passes through the midpoint coordinates (X5, Y5) and is orthogonal to the line segment bc. )
It becomes.
Here, an equation of a straight line with an angle A passing through the midpoint coordinates (X4, Y4) of the line segment ab and an equation of a straight line with an angle B passing through the midpoint coordinates (X5, Y5) are created, and the simultaneous equations are solved. Circle center coordinates (XC, YC) can be obtained.
XC = X4 + {(Y5-Y4)-(X5-X4) tanB}
÷ (tanA-tanB) Equation (13)
YC = Y4 + (XC-X4) tanB Formula (14)
Similarly, by obtaining the circle center coordinates (XC1, YC1) for the actual measurement positions aa to dd, a difference from the circle center coordinates (XC, YC) of the ideal position is obtained and used as a position correction amount.

  According to the present embodiment, one point of the workpiece teaching position of the cylindrical workpiece is stored, and the dimension data (radius, height, etc.) of the workpiece is input. From the workpiece teaching position and workpiece dimension data, an operation pattern of the robot is generated so that the position detection sensor arranged in the gripper 104 can perform a specific measurement operation. By operating the robot in accordance with this operation pattern, the position of the workpiece can be acquired by the position detection sensor, and the deviation amount of the position and orientation of the workpiece can be calculated and corrected. Therefore, if the worker roughly teaches the workpiece position, it is possible to automatically acquire and correct the workpiece position automatically.

In the third embodiment, the displacement amount of the position and orientation of the workpiece 105 is obtained by scanning the one-dimensional or two-dimensional position detection sensor twice at a certain distance from the center of the circle. In this embodiment, a case where a two-dimensional position detection sensor is used for dynamic measurement using a three-dimensional sensor will be described.
As shown in FIG. 7, when the position detection sensor 106 is a two-dimensional sensor, the position detection sensor 106 is moved in the X-axis direction of the workpiece 105, the entire workpiece is continuously measured, and the three-dimensional shape of the workpiece 105 is measured. Ask for. Thereby, the correction amount of the position and orientation can be obtained. The start position and end position in this case are defined as follows. On the straight line passing through the points e and f on the circumference obtained in the third embodiment, the virtual points e ₁ and f ₁ located outside are set by the positional deviation allowable error ΔX of the workpiece 105. The points positioned above the height adjustment amount ΔZ so that these positions e ₁ and f ₁ fall within the measurement distance of the position detection sensor 106 are defined as a start position e ₂ and an end position f ₂ , respectively.
As shown in the cross section C, when the position detection sensor 106 is a two-dimensional sensor, the sensor scanning direction is taken in the Y-axis direction, the sensor distance measuring direction is taken in the Z-axis direction, and e ₂ and f ₂ are used as a reference for scanning operation Is stored in the memory 1023 as a teaching point.
During the reproduction operation, the position detection sensor 106 continuously scans the positions e ₂ to f ₂ to acquire three-dimensional data.
Processing for obtaining the center position of the circle on the upper surface of the workpiece is added from the measurement data of the three-dimensional shape of the workpiece 105. The circle center is obtained after obtaining a position (3 points or 4 points) on the circumference similar to that in the third embodiment.

  According to the present embodiment, since the three-dimensional position and posture of the cylindrical workpiece can be measured by one measurement from one direction, it is possible to shorten the time for the measurement operation.

In the third embodiment, the position deviation amount is obtained by comparing the position of the circle at the ideal position with the position of the actually measured circle. In this embodiment, a method for preventing a detection failure or the like by measuring again when there is an error of a certain amount or more in the positional deviation amount will be described.
This will be described with reference to the flowchart of FIG. The workpiece 105 is a rectangular parallelepiped.
As the teaching work, the same four stages (A to D) as in the first embodiment are performed, and the three stages (E to G) as in the first embodiment are also performed during the regeneration operation. The difference from the first embodiment is that, when the correction amount calculation of F is finished, if the deviation amount between the position and the posture exceeds a preset threshold value, the scan operation during the regeneration operation of E is started again. That is.
When the position detection sensor is a laser sensor, the irregular reflection of the laser increases at the edge portion of the workpiece 105, and the detection error of the workpiece edge position may increase. However, the edge position can be accurately obtained by repeating the measurement of the scanning operation. Also, stable measurement is possible by averaging the measurement results of a plurality of scan operations and obtaining the edge position from the average value.

  According to this method, the position of the circle at the ideal position and the position of the actually measured circle are compared to determine the amount of misalignment. Detection failure and the like can be prevented. Further, even when irregular reflection occurs at the edge portion of the workpiece 105, it is possible to detect a stable edge position by averaging a plurality of measurement data.

In the fifth embodiment, the detection error is reduced by comparing the position of the circle at the ideal position with the position of the actually measured circle and re-measurement if there is an error of a predetermined amount or more. I was letting. In the present embodiment, a case will be described in which detection failure is prevented by changing the measurement direction when there are many detection failures.
This will be described with reference to the flowchart of FIG. Similar to the fifth embodiment, the workpiece 105 is a rectangular parallelepiped.
As in the fifth embodiment, the teaching work is performed in four stages (A to D), and the work during the regeneration operation is performed in three stages (E to G). The difference from the fifth embodiment is that when the correction amount calculation of F is finished, if the deviation amount between the position and the posture exceeds a preset threshold value, the scan operation during the regeneration operation of E is started again. However, when the number of redoes reaches a predetermined number, the process returns to the generation of the D scan operation.
In the generation of the scanning operation of D, for example, in the first and fifth embodiments, the edge position detection operation on the long side is a scanning operation in the direction from the edge position a to c in the case of the cross section A, and the positional deviation amount is a certain amount or more. If the number of measurement restarts reaches a predetermined number, the scan operation direction is changed. Specifically, the edge position detection operation on the long side is a scan operation in the direction from the edge position c to a for the cross section A. Further, for example, in the third embodiment, as shown in FIG. 7, a scanning operation is performed in a Y-axis direction with a cylindrical workpiece, and there is an error that is greater than a certain amount of misalignment. The scanning operation is performed with the operation direction as the X-axis direction.
In addition, you may change to the position of predetermined distance from both ends, without changing long direction W1 and W2 of the workpiece | work 105 into 4 equal parts, without changing an operation direction. A method of changing the posture measured by the robot without changing the movement direction may be used. In FIG. 3, the position detection sensor may be rotated around the X axis to change the laser irradiation angle to the edge.

  According to this embodiment, the position of the circle at the ideal position is compared with the position of the actually measured circle to obtain the amount of misalignment. When there are many failures, changing the measurement direction can prevent detection failure. In addition, by changing the posture measured by the robot without changing the operation direction, it is possible to change the laser irradiation angle to the edge and prevent failure of detection.

The present invention is a teaching reproduction operation of a robot for use in welding, painting, assembly, etc.
The position roughly taught by the operator can be corrected by acquiring a detailed three-dimensional position and posture of the work by using the position detection sensor, the work teaching position and the work dimension data.

Claims (20)

A memory for storing a first position and posture immediately before the end effector grips the workpiece;
A workpiece dimension input unit for inputting dimension data of the workpiece;
Based on the first position and orientation and the input dimension data of the workpiece, an operation pattern generation for generating an operation command for a position detection sensor provided in an end effector to measure the position and orientation of the workpiece And
The robot is operated based on the operation command generated by the operation pattern generator, the position and orientation of the workpiece are measured, and the first position and orientation, and the shape dimension data of the workpiece determined. Control means for obtaining a deviation amount between the calculated position and orientation of the workpiece and the actually measured position and orientation, and correcting the position and orientation of the end effector based on the deviation amount. Robot control device. The robot control apparatus according to claim 1, wherein the position detection sensor is a one-dimensional position detection sensor. The robot control apparatus according to claim 1, wherein the position detection sensor is a two-dimensional position detection sensor. 2. The robot control apparatus according to claim 1, wherein when the work teaching position is stored, one point of the same position and posture as the end effector acting on the work is registered in actual work. The robot control apparatus according to claim 1, wherein the workpiece dimension data is input from a teaching apparatus. The robot control apparatus according to claim 1, wherein the workpiece dimension data is data based on CAD data. The workpiece has a rectangular parallelepiped shape,
The operation command is such that the position detection sensor includes a position on four edges on a long side that is a predetermined distance away from each of four vertices of a rectangle that is a measurement target surface of the workpiece, and two short sides of the rectangle. 2. The robot control device according to claim 1, wherein the robot control device is an operation command for measuring a position on two edges, each being a middle point. The workpiece has a rectangular parallelepiped shape,
The operation command is an operation command for the two-dimensional position detection sensor to measure a three-dimensional shape of the workpiece while moving in a short side direction of a rectangle that is a measurement target surface of the workpiece. The robot control device according to claim 3. The workpiece has a cylindrical shape with a measurement target surface as an upper surface,
The operation command is an operation command for measuring positions on four edges where parallel straight lines that are equidistantly spaced in the opposite direction from the center of the circle of the upper surface and the upper surface intersect each other. 4. The robot control device according to 3. The workpiece has a cylindrical shape,
The robot control apparatus according to claim 3, wherein the operation command is an operation command in which the two-dimensional position detection sensor scans an upper surface that is a measurement target surface of the workpiece in a certain direction. When the amount of positional deviation of the workpiece exceeds a preset threshold, the robot is operated again based on the operation command to obtain a position and orientation by actual measurement of the workpiece, and the first position and orientation 2. A control means for re-determining a deviation amount between the calculated position and orientation of the workpiece obtained based on the workpiece shape dimension data and the position and orientation of the second actual measurement. The robot control device described in 1. When the amount of positional deviation of the workpiece exceeds a preset threshold, the movement pattern generator changes the movement direction to the reverse or orthogonal direction, and the position detection sensor detects the actual measurement position of the workpiece. The operation command to be generated is regenerated, the robot is operated based on the regenerated operation command, the position and posture of the workpiece are measured, the first position and posture, and the shape dimension data of the workpiece 2. The robot control according to claim 1, further comprising a control unit that recalculates a deviation amount between the calculated position and orientation of the workpiece obtained based on the above and the position and orientation obtained by the re-measurement. apparatus. A robot provided with an end effector for gripping a workpiece;
A control device for controlling the robot,
The control device includes a position detection sensor that measures the position and posture of the workpiece;
A memory for storing a first position and posture immediately before the end effector grips the workpiece;
A workpiece dimension input unit for inputting dimension data of the workpiece;
An operation pattern generator for generating an operation command for a position detection sensor provided in an end effector to measure the position of the workpiece based on the first position and orientation and the input dimension data of the workpiece. When,
A command generator that generates a motion command for the robot based on the motion pattern generated by the motion pattern generator;
The amount of deviation between the position and posture of the workpiece obtained by calculation from the first position and posture and the shape dimension data of the workpiece and the position and posture actually measured by the position detection sensor is obtained, and the deviation A robot system that corrects a position and a posture of the end effector based on a quantity. The robot system according to claim 13, wherein the position detection sensor is provided in the end effector. The robot system according to claim 14, wherein the position detection sensor is a one-dimensional position detection sensor. The robot system according to claim 14, wherein the position detection sensor is a two-dimensional position detection sensor. The workpiece has a rectangular parallelepiped shape,
The operation command is such that the position detection sensor includes a position on four edges on a long side that is a predetermined distance away from each of four vertices of a rectangle that is a measurement target surface of the workpiece, and two short sides of the rectangle. The robot system according to claim 13, wherein the robot system is an operation command for measuring a position on two edges, each being a middle point. The workpiece has a cylindrical shape with a measurement target surface as an upper surface,
The operation command is an operation command for measuring positions on four edges where parallel straight lines that are equidistantly spaced in the opposite direction from the center of the circle of the upper surface and the upper surface intersect each other. 14. The robot system according to 13. A robot control method characterized in that a robot including an end effector for gripping a workpiece and a position sensor for detecting the position of the workpiece is taught and reproduced in the following steps:
(1) A first position and posture immediately before the end effector grips the workpiece are roughly taught.
(2) The calculation position and orientation of the workpiece are obtained based on the first position and orientation and the shape data of the workpiece.
(3) Based on the position and posture of the workpiece, an operation command of the robot for actually measuring the workpiece position is generated.
(4) The robot is operated based on the operation command, and the position of the workpiece is measured by the position detection sensor.
(5) The amount of deviation between the calculated position and orientation of the workpiece and the position and orientation of the workpiece measured by the position detection sensor is obtained.
(6) The position and posture of the end effector at the gripping position are corrected based on the amount of deviation. 14. The robot control method according to claim 13, wherein the position detection sensor is provided in the end effector.

Images