JP5606424B2 - Component extraction method and component extraction system - Google Patents

Description

  The present invention relates to a component extraction method and a component extraction system.

  In recent production forms, there is a demand for production equipment that takes into account the variety / variation form that can cope with the rapid change of the production amount and the change of the model from the conventional high-mix / small-volume production. In particular, there is a demand for the development of highly versatile equipment with a wide target product range from equipment specialized for dedicated functions for individual products. Among them, robots, which are versatile nuclear products, are frequently used in recent years. The part picking method when gripping a part with this robot has been conventionally performed by using a dedicated device such as a parts feeder. However, in recent high-mix low-volume production forms, a single part is in a state of being stacked in multiple parts. It is necessary to directly grip and supply to a predetermined apparatus such as an automatic assembly machine.

  In Patent Document 1, in a component pickup device, a group of bolts in a messy overlapping state on a tray is imaged with a CCD camera, and an isolated bolt whose contour is isolated from other bolts on the captured image is searched. It is described that when an isolated bolt is found, the robot picks up the isolated bolt, and when the isolated bolt cannot be found, the vibration device is operated to search for the isolated bolt again. Thus, according to Patent Document 1, since the isolated bolts are picked up one by one from the bolt group on the tray, it is not necessary to align and position the parts required before starting the grip pickup operation by the conventional robot. Therefore, it is said that it is possible to omit manual and dedicated mechanisms for aligning and positioning parts.

  Patent Document 2 discloses that in a component supply apparatus, a robot can grab and disperse a plurality of components from a mountain of components, eliminate the overlap between components, photograph each dispersed component with a camera, and supply from the image It describes that a component in a posture is identified, grasped by a robot, and supplied to a component supply location. Thus, according to Patent Document 2, since the parts are not overlapped by dispersing the parts, it is not necessary to separate the parts from the background, and the image recognition apparatus for recognizing the orientation of the parts is a simple two-dimensional image recognition. It is said that things will improve.

Japanese Patent Laid-Open No. 11-300670 Japanese Patent Laid-Open No. 06-127698

  In the technique described in Patent Document 1, it is assumed that the number of bolt groups in a messy overlapping state is at most about 2 to 5, and a plurality of parts fill the bottom surface of the tray. However, it is not assumed that they are piled up on the tray. Patent Document 1 does not describe at all how to improve the efficiency of the process of taking out one component from a tray in which a plurality of components are stacked and supplying them to a predetermined apparatus.

  In the technique described in Patent Document 1, even if an attempt is made to remove a bolt from a tray in which a plurality of bolts are piled up and vibration is applied to the tray by an external force, the peak of the bolt only collapses slightly. It is difficult to take out the bolt (part) because the bolt is not in an isolated state (isolated bolt) and cannot be found by image recognition.

  In the technique described in Patent Document 2, since it is necessary for the robot to grasp a plurality of parts from a pile of parts and distribute the parts one by one so that they do not overlap, it is necessary to control the operation of a very complicated robot hand. The amount of calculation tends to increase. That is, in the technique described in Patent Document 2, the efficiency of processing for taking out one component from a tray in which a plurality of components are stacked and supplying the same to a predetermined apparatus is likely to be reduced.

  If a robot grabs multiple parts from a pile of parts and performs processing with simple robot hand motion control, there is a lack of certainty depending on the probability that duplication will not occur. Duplicate parts are likely to occur between dispersed parts, and it is necessary to repeat the dispersion process. Therefore, the efficiency of the process of taking out one part from a tray in which a plurality of parts are stacked and supplying it to a predetermined apparatus is improved. It tends to decline.

  The present invention has been made in view of the above, and provides a component extraction method and a component extraction system that can improve the efficiency of the process of taking out one component from a tray in which a plurality of components are stacked and supplying them to a predetermined apparatus. The purpose is to obtain.

  In order to solve the above-described problems and achieve the object, a part picking method according to one aspect of the present invention includes a first robot having a three-dimensional image sensor and a first robot hand, and a plurality of parts stacked. The three-dimensional image sensor recognizes the three-dimensional position and state of one part to be taken out from the tray, and the first part is taken out from the tray by the first robot hand according to the recognition. A temporary placement step in which the first robot temporarily places the one part taken out in the first take-out step on a stage with the first robot hand; and a swing mechanism, A swinging step of swinging the stage so as to guide the posture of the one part temporarily placed on the stage to a specific posture among a plurality of postures; A second robot having a two-dimensional image sensor and a second robot hand recognizes the position of the one component in the specific posture on the stage by the two-dimensional image sensor, and according to the recognition, And a second picking-up step for picking up one part with the second robot hand.

  According to the present invention, since the posture of the part on the stage is guided to a specific posture, registration data at the time of posture pattern comparison by the two-dimensional image sensor mounted on the second robot can be greatly reduced. The content of image processing by the image sensor can be simplified. As a result, the load of image processing by the two-dimensional image sensor can be reduced, the processing speed of image processing by the two-dimensional image sensor can be improved, and productivity can be increased. In other words, it is possible to improve the efficiency of the process of taking out one component from a tray in which a plurality of components are stacked and supplying them to a predetermined apparatus.

FIG. 1 is a diagram illustrating a configuration of a component pickup system according to an embodiment. FIG. 2 is a diagram illustrating a configuration of a stage and a swing mechanism in the embodiment. FIG. 3 is a sequence diagram illustrating the operation of the component pickup system according to the embodiment. FIG. 4 is a diagram illustrating an operation of the component pickup system according to the embodiment. FIG. 5 is a diagram showing the shape of a component in the embodiment. FIG. 6 is a diagram illustrating the posture of a component in the embodiment. FIG. 7 is a sequence diagram illustrating the operation of the component removal system according to the modification of the embodiment.

  Embodiments of a component picking method and a component picking system according to the present invention will be described below in detail with reference to the drawings. Note that the present invention is not limited to the embodiments.

Embodiment.
A component picking system 100 according to an embodiment will be described with reference to FIGS. 1 and 2. FIG. 1 is a diagram showing a configuration of a component picking system 100. FIG. 2 is a diagram illustrating the configuration of the stage 40 and the swing mechanism 50.

  The parts removal system 100 is a system for taking out one part from the tray 30 in which a plurality of parts are stacked and supplying the same to a predetermined device (not shown). The predetermined apparatus is, for example, an automatic machine that performs a predetermined assembly operation using supplied parts.

  Specifically, the component extraction system 100 includes a robot (first robot) 10, a robot (second robot) 20, a tray 30, a stage 40, a swing mechanism 50, and a control unit 60.

  The robot 10 is, for example, a multi-joint robot with 6 degrees of freedom, and includes a plurality of robot arms 13-1, 13-2, a plurality of joint portions 14-1 to 14-3, a base portion 15, a flange 16, and a three-dimensional image sensor. 11 and a robot hand (first robot hand) 12. The plurality of robot arms 13-1 and 13-2 are connected to the base portion 15 via the plurality of joint portions 14-1 to 14-3. A flange 16 is provided at the tip of the robot arm 13-1. For example, a robot hand 12 or the like is attached to the flange 16 as a working tool. The three-dimensional image sensor 11 is attached to the flange 16 along with a tool such as the robot hand 12. The robot controller 61 controls the positions and postures of the robot hand 12 and the flange 16. As a result, the robot 10 performs a predetermined operation on a predetermined object.

  The three-dimensional image sensor 11 includes an imaging unit 11a and an image processing unit 11b. The imaging unit 11a acquires a three-dimensional image of the object scene. For example, the imaging unit 11a acquires a three-dimensional image of a plurality of components WK-1 to WK-N existing inside the tray 30. The image processing unit 11b performs predetermined image processing on the three-dimensional image acquired by the imaging unit 11a. The image processing unit 11b transmits the three-dimensional image data subjected to the image processing to the control unit 60. Note that the image processing unit 11b may be arranged in a housing in which the imaging unit 11a is built, or may be arranged outside the housing in which the imaging unit 11a is built in.

  The robot 20 is, for example, a 6-DOF multi-joint robot, and includes a plurality of robot arms 23-1, 23-2, a plurality of joint portions 24-1 to 24-3, a base portion 25, a flange 26, and a two-dimensional image sensor. 21 and a robot hand (second robot hand) 22. The plurality of robot arms 23-1 and 23-2 are connected to the base portion 25 via a plurality of joint portions 24-1 to 24-3. A flange 26 is provided at the tip of the robot arm 23-1. For example, a robot hand 22 or the like is attached to the flange 26 as a working tool. A two-dimensional image sensor 21 is attached to the flange 26 along with a tool such as the robot hand 22. The robot controller 62 controls the positions and postures of the robot hand 22 and the like and the flange 26. Thereby, the robot 20 performs a predetermined operation on the predetermined object.

  The two-dimensional image sensor 21 includes an imaging unit 21a and an image processing unit 21b. The imaging unit 21a acquires a two-dimensional image of the object scene. For example, the imaging unit 21a acquires a two-dimensional image of one component WK present on the stage 40. The image processing unit 21b performs predetermined image processing on the two-dimensional image acquired by the imaging unit 21a. The image processing unit 21b transmits the two-dimensional image data subjected to the image processing to the control unit 60. Note that the image processing unit 21b may be arranged in a housing in which the imaging unit 21a is built, or may be arranged outside the housing in which the imaging unit 21a is built in.

  On the tray 30, a plurality of parts WK-1 to WK-N are piled up and piled up. In the present specification, the “stack” indicates a state in which the plurality of parts WK-1 to WK-N are stacked so as to fill the bottom surface of the tray 30.

  The stage 40 is a stage on which one component WK among the plurality of components WK-1 to WK-N stacked on the tray 30 is to be temporarily placed. For example, the stage 40 includes a bottom surface portion 41 and four side surface portions 42 to 45 as shown in FIG. The bottom surface portion 41 forms a bottom surface extending in the X direction and the Y direction. Each of the side surface portions 42 to 45 extends from the outer edge of the bottom surface portion 41 so as to rise to the + Z side. Accordingly, it is possible to suppress one component WK from jumping out of the stage 40 when one component WK is placed on the bottom surface portion 41 and the stage 40 is swung in, for example, the X direction and the Y direction.

  The swing mechanism 50 swings the stage 40 in, for example, the X direction and the Y direction. For example, the swing mechanism 50 includes an X-axis drive source 51, a Y-axis drive source 52, an X-axis drive mechanism 53, and a Y-axis drive mechanism 54 as shown in FIG. The X-axis drive source 51 swings the stage 40 in the X direction via the X-axis drive mechanism 53 under the control of the swing controller 63. The Y-axis drive source 52 swings the stage 40 in the Y direction via the Y-axis drive mechanism 54 under the control of the swing controller 63.

  The control unit 60 controls the respective units in the component picking system 100 as a whole. The control unit 60 includes a robot controller 61, a robot controller 62, and a swing controller 63. The robot controller 61 controls the robot 10 according to the robot program and transmits / receives predetermined control signals to / from the robot controller 62 and the swing controller 63. The robot controller 62 controls the robot 20 according to the robot program and transmits / receives predetermined control signals to / from the robot controller 61 and the swing controller 63. The swing controller 63 controls the swing mechanism 50 in accordance with, for example, a control signal received from the robot controller 61 or the robot controller 62.

  For example, the control unit 60 controls the robot 10 so that the three-dimensional image sensor 11 recognizes the three-dimensional position and state of one component WK to be removed from the tray 30 in the first period. In the first period, the control unit 60 controls the robot 10 to take out one component WK with the robot hand 12 according to the recognition result. In the second period, the control unit 60 controls the robot 10 so that the picked-up component is temporarily placed on the stage 40 by the robot hand 12. The second period is a period following the first period. The controller 60 controls the swing mechanism 50 so as to swing the stage 40 in the third period. Thereby, the posture of one component WK temporarily placed on the stage 40 can be changed to a specific posture among a plurality of postures. The third period is a period following the second period. The control unit 60 controls the robot 20 so that the two-dimensional image sensor 21 recognizes the position of one component WK in a specific posture on the stage 40 in the fourth period. The fourth period is a period following the third period. In the fourth period, the control unit 60 controls the robot 20 so that one part WK is taken out by the robot hand 22 and supplied to a predetermined device (for example, an automatic machine) according to the recognition result.

  Next, operation | movement of the components extraction system 100 is demonstrated using FIGS. FIG. 3 is a sequence diagram illustrating the operation of the component picking system 100. 4A to 4D are diagrams illustrating a procedure for taking out one component WK from the tray 30. FIG. FIGS. 5A, 5B, and 5C are a plan view, a side view, and a front view showing the shape of one component WK, respectively. 6A to 6E are diagrams illustrating a procedure for guiding the posture of one component WK to a specific posture.

  In step S1, the robot controller 61 controls the robot 10 to move the three-dimensional image sensor 11 to a position where the inside of the tray 30 can be imaged (see FIG. 4A). The imaging unit 11a (see FIG. 1) of the three-dimensional image sensor 11 images a plurality of components WK-1 to WK-N stacked in the tray 30 three-dimensionally. That is, the imaging unit 11a acquires three-dimensional images of the plurality of components WK-1 to WK-N stacked in the tray 30 and passes them to the image processing unit 11b.

  Then, the robot controller 61 controls the robot 10 to recognize the three-dimensional position and state of the component WK to be gripped by the robot hand 12 among the plurality of components WK-1 to WK-N according to the captured image. To do. That is, the image processing unit 11b of the three-dimensional image sensor 11 performs predetermined image processing on the three-dimensional image acquired by the imaging unit 11a. The image processing unit 11b transmits the three-dimensional image data subjected to the image processing to the robot controller 61. The robot controller 61 uses the three-dimensional image data to identify one component WK to be gripped by the robot hand 12 among the plurality of components WK-1 to WK-N, and the three-dimensional position of the identified one component WK. And the state of the component WK (for example, the posture of the component WK).

  In step S <b> 2, the robot controller 61 controls the robot 10 and takes out one component WK from the tray 30 with the robot hand 12 according to the result recognized in step S <b> 1. That is, the robot controller 61 moves the tip of the robot hand 12 to the vicinity of the three-dimensional position of the component WK to be gripped, and the orientation of the tip of the robot hand 12 according to the state of the component WK (for example, the posture of the component WK). And the posture is controlled, and the component WK specified in step S1 is gripped (see FIG. 4B). At this time, the tip claw that grips the parts of the robot hand 12 has little influence on the stacked state of the parts WK when gripping, and has a thin tip shape so as not to grip parts other than the specified part WK at the same time. Nail with is desirable.

  In step S <b> 3, the robot controller 61 controls the robot 10 to temporarily place the one component WK taken out in step S <b> 2 on the stage 40 with the robot hand 12. That is, the robot controller 61 moves the robot hand 12 holding the specified one component WK upward above the tray 30 and further moves above the stage 40 (see FIG. 4C). Then, the robot controller 61 opens the robot hand 12 above the stage 40 and temporarily places the gripped component WK on the stage 40 (see FIG. 4D).

  Then, the stage 40 is swung so as to narrow down (guide) the posture of the component to a specific posture. Specifically, the processes in steps S4 to S7 are performed.

  In step S <b> 4, the robot controller 61 transmits a swing command for instructing to swing the stage 40 to the swing controller 63.

  When the swing controller 63 receives the swing command, the swing controller 63 controls the processing in steps S5 and S6 to be performed.

  In step S5, the swing controller 63 determines a swing pattern that is a driving pattern for swinging the stage 40 as a predetermined swing pattern. In the predetermined swing pattern, the posture of one component WK placed on the stage 40 is a specific posture (see FIG. 6E) among a plurality of postures (see FIGS. 6A to 6E). It is a pattern acquired experimentally in advance so as to lead to The predetermined swing pattern includes a swing width, a swing cycle, the number of swings (or a swing time), and the like.

  In step S6, the swing controller 63 controls the swing mechanism 50 to swing the stage 40 according to the swing pattern determined in step S5.

  For example, when the shape of the component WK on the stage 40 is a shape as shown in FIGS. 5A to 5C, the initial state posture of the component WK on the stage 40 is indefinite, for example, FIG. There are four possible postures shown in (d). That is, the X-axis drive source 51 and / or the Y-axis drive source 52 are used when the first vertical posture, the second vertical posture, the first horizontal posture, or the second horizontal posture is in a state. The stage 40 is vibrated by driving the X-axis drive mechanism 53 or the Y-axis drive mechanism 54 or both in the left-right direction, the front-rear direction, or both directions, thereby changing to a stable posture shown in FIG. .

  At this time, it is desirable to increase the change in posture by attaching a rubber sheet, for example, to the bottom surface of the stage 40 (the surface of the bottom surface portion 41) and increasing the frictional force between the component WK and the bottom surface. As the drive source for generating vibration, it is preferable to use an air cylinder that can be easily connected and controlled. However, depending on the part that narrows down the posture, a drive source such as a servo motor can be used to increase the vibration. It becomes possible to change the speed to a stable posture more reliably by changing the speed during deceleration and vibration.

  When the swing of the stage according to a predetermined swing pattern is completed, the swing controller 63 transmits a swing completion notification to the robot controller 62.

  In step S7, if the robot controller 62 has not received the swing completion notification, it is determined that the swing of the stage 40 has not been completed (No in step S7), and the process proceeds to step S7. When the notification is received, it is determined that the swing of the stage 40 has been completed (Yes in step S7), and the process proceeds to step S8.

  In step S <b> 8, the robot controller 62 controls the robot 20 and moves the two-dimensional image sensor 21 to a position where the top of the stage 40 can be imaged. The imaging unit 21a (see FIG. 1) of the two-dimensional image sensor 21 images the component WK placed on the stage 40 two-dimensionally. That is, the imaging unit 21a acquires a two-dimensional image of the component WK placed on the stage 40 and passes it to the image processing unit 21b.

  Then, the robot controller 62 controls the robot 20 to recognize the two-dimensional position of the component WK to be gripped by the robot hand 22 according to the captured image. That is, the image processing unit 21b of the two-dimensional image sensor 21 performs predetermined image processing on the two-dimensional image acquired by the imaging unit 21a.

  For example, the image processing unit 21b extracts a component outline pattern from a two-dimensional image by edge detection or the like, compares the extracted component outline pattern with a previously registered component outline pattern, and performs image processing on the position of the component. Ask for.

  Here, if the position of the component WK is not narrowed down (steps S5 and S6), the component external patterns registered in advance for comparison have the shapes shown in FIGS. 5 (a) to 5 (c). Four patterns corresponding to FIGS. 6A to 6D are required for the component WK.

  On the other hand, in the present embodiment, since the orientation of the component WK is narrowed down (steps S5 and S6), the component outline patterns registered in advance for comparison are shown in FIGS. 6 can be reduced to a single pattern corresponding to FIG. 6E. As a result, the number of patterns registered in the image processing unit 21b can be reduced, and the content of the image processing by the image processing unit 21b can be simplified. Can be planned.

  The image processing unit 21b transmits the two-dimensional image data subjected to the image processing to the robot controller 62. The robot controller 62 recognizes the position of one component WK to be gripped by the robot hand 22 on the stage 40 using the two-dimensional image data. That is, the robot controller 62 recognizes the two-dimensional position of one component WK to be gripped by the robot hand 22.

  In step S9, the robot controller 62 controls the robot 20, takes out one component WK from the stage 40 with the robot hand 22 according to the result recognized in step S8, and supplies it to a predetermined device (for example, an automatic machine). To do. That is, the robot controller 62 moves the tip of the robot hand 22 to the vicinity of the two-dimensional position of the component WK to be gripped, and grips the component WK on the stage 40. Then, the robot controller 62 raises the robot hand 22 holding one component WK above the tray 30 and further moves to a predetermined device.

  As described above, in this embodiment, the three-dimensional position and state of the target part to be grasped by the three-dimensional image sensor mounted on the first robot from the tray in which a large number of parts are stacked are recognized and one piece. By taking out these parts with a robot, it is not necessary to mount a large number of dedicated part feeders for each type of parts for a wide variety of parts, and the cost of the part picking system can be reduced. In addition, since the posture of the component picked up by the first robot is indefinite, a large number of component postures are narrowed down to specific postures by temporarily placing them on the stage and vibrating the stage with an external force. This greatly reduces the registration data when comparing the posture pattern by the image processing unit of the two-dimensional image sensor mounted on the second robot, and simplifies the contents of image processing by the image processing unit of the two-dimensional image sensor. it can. As a result, the load of image processing by the image processing unit of the two-dimensional image sensor can be reduced, the processing speed of image processing by the image processing unit of the two-dimensional image sensor can be improved, and productivity can be increased. In other words, it is possible to improve the efficiency of the process of taking out one component from a tray in which a plurality of components are stacked and supplying them to a predetermined apparatus.

  The removal of a part by the robot 20 for one part (step 8, step S9) and the removal of a part by the robot 10 for a part to be taken out after that one part (step 1, step S2) are: It may be performed in parallel. In this case, the process of taking out one part from a tray in which a plurality of parts are piled up and supplying them to a predetermined apparatus can be made efficient even when viewed with a plurality of parts taken out sequentially.

  Further, the component pick-up system 100 prepares a plurality of swing patterns, detects the posture of the component on the stage 40 before swinging the stage 40, and responds to the detected posture and the specific posture to be guided. The actually used swing pattern may be selected from a plurality of swing patterns.

  For example, the component extraction system 100 may perform steps S11 to S16 shown in FIG. 7 instead of the processes of steps S3 to S6 shown in FIG.

  In step S11, the robot controller 61 controls the robot 10 to detect the posture of the component WK on the stage 40. Specifically, the three-dimensional image sensor 11 is moved to a position where the top 40 can be imaged. The imaging unit 11a (see FIG. 1) of the three-dimensional image sensor 11 images the component WK on the stage 40 three-dimensionally. That is, the imaging unit 11a acquires a three-dimensional image of the component WK on the stage 40 and passes it to the image processing unit 11b. Then, the robot controller 61 controls the robot 10 to detect the posture of the component WK by recognizing the posture of the component WK on the stage 40 according to the captured image. The robot controller 61 transmits the detected attitude information of the component WK to the swing controller 63.

  When the swing controller 63 receives the information on the posture of the component WK, the swing controller 63 performs control so that the processes in steps S12 and S13 are performed.

  In step S12, the swing controller 63 determines a swing pattern for swinging the stage 40 from among a plurality of swing candidate patterns according to the posture detected in step S11 and the specific posture to be guided. .

  For example, with respect to the shape of the part WK shown in FIGS. 5A to 5C, the first vertical posture (see FIG. 6A) is guided to the first horizontal posture (see FIG. 6C). If you want to do this, for example, if you want to increase the swing width in the Y direction, decrease the swing width in the X direction, increase the swing speed in the -Y direction, and decrease the swing speed in the + Y direction, Select from candidate patterns.

  For example, with respect to the shape of the part WK shown in FIGS. 5A to 5C, the second vertical posture (see FIG. 6B) is guided to the first horizontal posture (see FIG. 6C). For example, a swing pattern is selected from a plurality of swing candidate patterns so as to increase the swing width in the X direction, decrease the swing width in the Y direction, and increase the swing width in the Z direction. At this time, the swing mechanism 50 further includes a Z-axis drive source (not shown) and a Z-axis drive mechanism (not shown). The Z-axis drive source is controlled by the swing controller 63. Originally, the stage 40 is swung in the Z direction via the Z-axis drive mechanism.

  In step S13, the swing controller 63 controls the swing mechanism 50 to swing the stage 40 according to the swing pattern determined in step S12. The swing controller 63 transmits a swing completion notice to the robot controller 61 when the swing of the stage according to a predetermined swing pattern is completed.

  In step S <b> 14, when the robot controller 61 receives the swing completion notification, the robot controller 61 controls the robot 10 to detect again the posture of the component WK on the stage 40. Specifically, the three-dimensional image sensor 11 is moved to a position where the top 40 can be imaged. The imaging unit 11a (see FIG. 1) of the three-dimensional image sensor 11 images the component WK on the stage 40 three-dimensionally. That is, the imaging unit 11a acquires a three-dimensional image of the component WK on the stage 40 and passes it to the image processing unit 11b. Then, the robot controller 61 controls the robot 10 to detect the posture of the component WK by recognizing the posture of the component WK on the stage 40 according to the captured image.

  In step S15, the robot controller 61 determines whether or not the posture of the component WK on the stage 40 is guided to a specific posture. If the posture of the component WK on the stage 40 is not guided to a specific posture (No in step S15), the robot controller 61 returns the process to step S11, and the posture of the component WK on the stage 40 is changed to a specific posture. When it is guided (Yes in step S15), the process proceeds to step S16.

  The loop processing from step S11 to step S15 is repeated until the posture of the component WK on the stage 40 is guided to a specific posture.

  In step S <b> 16, the robot controller 61 transmits a swing completion notification indicating that the swing process as the loop process of steps S <b> 11 to S <b> 15 has been completed to the robot controller 62.

  As described above, a plurality of swing patterns are prepared, and the postures of the parts on the stage 40 are detected before the stage 40 is swung. A plurality of swing patterns are detected according to the detected posture and the specific posture to be guided. By selecting a swing pattern to be actually used from the moving pattern, it is possible to guide to a posture other than the stable posture. Thereby, it is easy to make the specific posture to be guided easier to reduce the load of image processing by the image processing unit of the two-dimensional image sensor.

  As described above, the component extraction system according to the present invention is useful for component extraction by a robot.

DESCRIPTION OF SYMBOLS 10 Robot 20 Robot 30 Tray 40 Stage 50 Oscillation mechanism 60 Control part 100 Component pick-up system

Claims (3)

A first robot having a three-dimensional image sensor and a first robot hand recognizes the three-dimensional position and state of one part to be taken out from a tray on which a plurality of parts are stacked, and recognizes the three-dimensional image sensor. A first step of removing the one component from the tray with the first robot hand according to
A temporary placing step in which the first robot temporarily places the one part taken out in the first taking out step on a stage with the first robot hand;
A swinging step for swinging the stage so that the swinging mechanism guides the posture of the one part temporarily placed on the stage to a specific posture among a plurality of postures;
A second robot having a two-dimensional image sensor and a second robot hand recognizes the position of the one component in the specific posture on the stage by the two-dimensional image sensor, and according to the recognition A second step of taking out the one part from the stage with the second robot hand;
With
The swing step includes
A detection step in which the three-dimensional image sensor detects the posture of the one part temporarily placed on the stage;
A determining step for determining a swing pattern for swinging the stage among a plurality of swing candidate patterns according to the posture detected in the detection step;
A swing execution step in which the swing mechanism swings the stage according to the swing pattern determined in the determination step;
A method for picking up parts. The second taking-out step for the one part and the first taking-out step for a part to be taken out next to the one part among the plurality of parts are performed in parallel. The part taking-out method according to claim 1 . A first robot having a three-dimensional image sensor and a first robot hand;
A second robot having a two-dimensional image sensor and a second robot hand;
A tray with a stack of multiple parts,
A stage on which one of the plurality of parts is to be temporarily placed;
A swing mechanism for swinging the stage;
Recognizes the three-dimensional position and status of one component to be taken out from the front Symbol tray by the 3-dimensional image sensor, the one component and Eject by the first robot hand in accordance with the recognition, the pre Symbol removed the one component is temporarily placed on the stage in the first robot hand has to guide the one part of the attitude that is temporarily placed on the front Symbol stage to a specific position of the plurality of postures the swing mechanism to swing the said stage, recognizing the position of said one component in a specific orientation in the top of a previous SL stage the two-dimensional image sensor as the in response to the recognition 1 A control unit for controlling the first robot, the second robot, and the swing mechanism so as to take out one part with the second robot hand ;
With
The controller is configured to detect the posture of the one component temporarily placed on the stage by the three-dimensional image sensor when guiding to the specific posture, and to control the plurality of pieces according to the detected posture. A swing pattern for swinging the stage is determined, and the swing mechanism is controlled to swing the stage according to the determined swing pattern. Parts picking system.

Images