JP3405045B2 - Component supply method and device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a component supply method and device for recognizing the position and orientation of a component conveyed on a conveyor by a visual recognition device and handling the component by a robot.

[0002]

2. Description of the Related Art A conventional part supply method is a method of recognizing the type, position and orientation of parts on a parts feeder and taking them out by a robot (Japanese Patent Laid-Open No. 64-5742), or from the parts feeder. There is a method (Japanese Unexamined Patent Publication No. 2-201111) in which it is transferred onto a conveyor, the recognition is performed on the conveyor, and when the component moves to the supply position, the conveyor is stopped and handling is performed.

[0003]

However, in the above-mentioned conventional example, since it is temporarily stopped to perform visual recognition or handling of parts, it takes a long time to stop the parts feeder and the conveyor, and the speed is increased. It was difficult to do. In addition, parts that cannot be handled by the robot and parts that cannot be visually recognized require a troublesome work of flowing them on the conveyor as they are, accumulating them once, and then flowing them again.

Further, since the moving speed of the conveyor or the capturing of the image by the visual recognition device is not controlled based on the component information such as the number of the recognized components, the position or the posture thereof, so as to be suitable for the removal of the components, It was difficult to take out parts efficiently.

That is, when the conveyor is moving at a constant speed, the waiting time of the robot increases if there are few parts on the conveyor, and conversely, if there are too many parts, some parts cannot be taken out or parts are recirculated. There is a problem of being hurt. Also, regarding image capture, for example, if a certain control is performed to capture an image when the conveyor moves for a certain distance, parts hanging in a certain area of the image cannot be recognized, If there are no parts in the range, there is a problem of unnecessary image capture.

The present invention has been made in view of the above problems, and an object of the present invention is to perform visual recognition or handling of components continuously without stopping the transportation of the components, and to perform visual recognition. By controlling the moving speed of the conveyor based on the component information from the device or the control of the image capture to the visual recognition device, the components can be reliably taken out at high speed and fed to the next process. It is an object of the present invention to provide a component supply method and a device having such features that components can be efficiently taken out, components can be taken out corresponding to various postures, and the amount of components supplied by a parts feeder can be automatically set.

[0007]

According to a first aspect of the invention, there is provided a method for supplying a plurality of components from a parts feeder onto a conveyor, and a robot removes the conveyed components. In the parts supply method of supplying to the next step, returning some unremovable parts to the parts feeder, a visual recognition device having a visual area having a certain visual width on the conveyor is sequentially captured as an image and recognized, Recognize the number, position or posture of parts in the visual area as part information, determine the parts that can be taken out based on these parts information, make the moving speed of the conveyor suitable for taking out the robot, and The position or figure that can be taken out by the robot while controlling to take in efficiently It constitutes a feature to retrieve the parts from the conveyor.

According to such a component supplying method, components are loaded from the parts feeder onto the conveyor, and an image of the loaded components on the conveyor is displayed on the conveyor by the visual recognition device in a visual area having a certain visual width. The images are sequentially captured. Then, based on this image, the number, position or orientation of the parts in the visual area is recognized as the part information. Based on these pieces of component information, a component in a retrievable state is determined, and the robot retrieves the component in the retrievable position or posture. In addition, parts that cannot be taken out are returned to the upstream side of the conveyor again and circulated.

In the above-described component supplying method, since the moving speed of the conveyor or the capturing of the image is controlled so as to be suitable for extracting the component, the component can be reliably extracted. In addition, the loss time such as the waiting time of the robot is small, and the parts can be taken out efficiently.

According to the first aspect of the invention, the parts to be taken out are determined based on the part information related to the posture, the inter-part distance is calculated based on the part information related to the position, and each part is calculated based on the inter-part distance. Are classified into separate parts, adjacent parts or overlapping parts, and the separating parts are first taken out, then the adjacent parts are taken out, and the overlapping parts are not taken out and are sent off.

In such a component supplying method, when taking out a neighboring component, the taking out of one component may change the state of the other component. Therefore, after taking out one component, how to take out the other component is decided. Need to change. For this reason, since it takes time to take out the adjacent parts, it is efficient to take out the separated parts first, and then take out the adjacent parts by a predetermined method, in which the taking out does not affect other parts. Parts are taken out.

According to a second aspect of the invention, in the first aspect of the invention, when taking out the adjacent parts, after taking out one of the parts, the position of the other part is checked in the image after taking out, and before the taking out. If the position is displaced as compared with the image, the extraction of the other part is postponed.

In such a component supplying method, when the adjacent components are taken out, there is a possibility that the position of the image of the other component is shifted after taking out one of the components. However, when the position is deviated, the removal of this component is postponed, so that the attempt to remove this component does not fail, the reliable removal of the component is performed, and the loss time does not occur.

According to a third aspect of the present invention, in the first aspect of the invention, the amount of parts input from the parts feeder is determined in proportion to the ratio of the removable parts determined based on the parts information. It is configured as a feature.

In such a component supply method, the amount of components input from the part feeder is determined so as to maximize the number of retrievable components, so that efficient component ejection is performed.

According to a fourth aspect of the present invention, there is provided a component supply device, which is a loop-shaped conveyor that continuously moves and circulates the components, a part feeder that loads the components onto the conveyor, and a conveyor movement amount measuring device. The parts are dispersed on the conveyor, the position or orientation of this part is updated, and the separation promoting device that makes it a removable part and the visual area with a certain visual width on the conveyor are sequentially captured as an image, A visual recognition device that recognizes the number, position, or orientation as part information, and calculates the inter-part distance based on this part information, and calculates each part based on this inter-part distance.
Classify different parts as separate parts, adjacent parts or overlapping parts
Then, the discriminating device that discriminates the parts that can be taken out and the order of taking them out, the conveyor control device that similarly controls the drive of the conveyor based on the part information, and the timing of capturing the image to the visual recognition device are moved to the conveyor. and an image capture control device for controlling based on the presence or absence of components on the quantity and conveyor, minute based on the discrimination result of the discriminating device
Remove the separated parts first, then the adjacent parts and
The present invention is characterized by comprising a robot and its hand for picking up parts without picking them up and feeding them out to the next process.

In such a component feeder, the components can be circulated on a loop-shaped conveyor, the components can be put in from a parts feeder, and the components dispersed on the conveyor can be taken out. In addition, the parts that could not be taken out have their dispersed state updated by the separation promoting device,
The position and posture are changed to a new state, and the sheet is circulated on the conveyor again.

For the parts in a dispersed state, the visual recognition device sequentially takes in a visual area having a certain visual width on the conveyor as an image, and the number, position or orientation of the parts is recognized as part information, and the discrimination device Based on this component information, the components that can be taken out and the order of taking them out are determined. Further, the drive of the conveyor is controlled by the conveyor control device based on this component information. Then, based on this determination result, the robot and its hand take out the component and supply it to the next process.

Further , the image capturing control device controls the timing of capturing an image into the visual recognition device based on the amount of movement of the conveyor and the presence / absence of parts on the conveyor.

The above-mentioned movement of the conveyor is continuously performed, and visual recognition or handling of the parts is performed without stopping the conveyance of the parts. Then, since the moving speed of the conveyor based on the component information from the visual recognition device or the control of capturing the image to the visual recognition device is controlled, the component can be reliably taken out at high speed.

Furthermore, determination device obtains component information from the vision recognition device, to determine the overlapping parts, adjacent parts,
Since the optimum take-out procedure is determined according to the state of the parts, the parts can be taken out efficiently.

According to a fifth aspect of the present invention, in the first aspect of the invention, the time required for obtaining the component information from the captured image is the image processing time, and the average image of a plurality of continuously captured images. The processing time is calculated, the average image processing time is compared with the supply determination threshold value, and the parts are supplied from the hopper to the parts feeder.

In such a component supply method, the amount of components at positions to be circulated on the conveyor and to be taken out is estimated from the average image processing time, and the amount of components corresponding to this component amount is supplied from the hopper. be able to.

According to a sixth aspect of the present invention, in the first aspect of the present invention, a component sensor for determining the presence or absence of a component is provided on the downstream side of the component take-out position on the path along which the component circulates, and the component sensor is used to determine a constant value. The component existence probability within the determination time of is checked, the component existence probability is compared with the supply determination threshold value, and the component is supplied from the hopper to the parts feeder.

In such a component supply method, the amount of components at the position after being circulated on the conveyor and taken out is estimated from the component existence probability, and the amount of components corresponding to this component amount is supplied from the hopper. You can

According to a seventh aspect of the present invention, in the fifth aspect of the present invention, a conveyor is formed in a loop shape, and a portion into which parts are loaded from a hopper of the conveyor is used as a return parts feeder, and the parts are attached to the return parts feeder. While providing a component sensor to determine the presence or absence of, by checking the component presence probability within a fixed determination time by this component sensor, comparing this component presence probability and the average image processing time with each supply determination threshold, The feature is that parts are supplied from the hopper.

In such a component supplying method, the component amount before taking out is estimated from the average image processing time, and the component amount after taking out is estimated from the component existence probability ratio, and the amount of the components corresponding to these component amounts is estimated from the hopper. Can be supplied.

The invention according to claim 8 is the invention according to claim 5 or 7.
In the invention described, in the state of continuing the component supply from the hopper, when the state in which no removable component is recognized in the image is continuously recognized, when the average image processing time is smaller than the notification determination threshold value In addition, it is configured to notify this result.

In such a component supply method, it is determined and notified that the hopper is empty and no component is actually supplied.

The invention of claim 9, wherein the claim 5 or 7
In the invention described, in the state where the parts supply from the hopper is stopped, the parts are recognized in the image, but there is a continuous absence of parts that can be taken out, and the average image processing time is clogged with parts. When it is larger than the judgment threshold value, the operation of the entire apparatus performing this component supply method is stopped.

In such a component supplying method, it is determined that the component is clogged on the upstream side of the component take-out position, and in such a state, the operation of the entire apparatus is stopped.

[0032]

DETAILED DESCRIPTION OF THE INVENTION One embodiment of the present invention will be described below with reference to the accompanying drawings.

With reference to FIGS. 1 to 3, the basic parts of the component supply method and apparatus according to this embodiment will be described below.

FIG . 1 is a perspective view showing a component supply apparatus for implementing the component supply method of this embodiment, and FIG. 2 is a control block diagram showing control of each component of the component supply apparatus, and FIG. FIG. 3 is a perspective view schematically showing constituent elements of the component supply device.

As shown in FIG . 1, the component supply apparatus of this embodiment has a vibration feeder 31, 32 having different moving directions as a component feeder 66 for charging the component 50 on the conveyor 1.
The vibration feeder 31 and the vibration feeder 32 are stepped.
The component 50 that moves by is moved to the vibration feeder 31. Further, the conveyor 1 has a first conveyor 34, a reversing mechanism 37, and a second conveyor 38. The component 50 can circulate in the loop formed by the conveyor 1 and the vibrating feeders 31, 32.

In this component feeder, the vibration type hopper is used.
The component 50 supplied from 33 is carried by the vibration feeders 31 and 32,
It can be dispersed and transferred on the first conveyor 34. In addition, as shown in FIG. 2 or 3, the first conveyor
The image of the component 50 is captured from the camera 35 to the visual recognition device 2 on the 34, and the visual recognition processing is performed to recognize the number, position or posture of the component 50 as the component information 3. Then, the robot 36 is moved based on this part information 3,
The part 50 can be taken out and supplied to the pallet 40 by 36 hands.

The difficult part is taken out by the robot 36 50
Is transferred to the second conveyor 38 by the reversing mechanism 37 mounted on the first conveyor 34, and further returned to the vibrating feeder 32 for circulation. At this time,
The vibrating feeder 32 functions as a separation promoting device 25 that disperses the component 50 on the conveyor 1 and updates the position or posture of the component 50 to make the component 50 removable.

Reference numeral 39 in this figure denotes a control device, which accommodates each device having the following functions. That is, the visual recognition device 2 including the camera 35 that captures the visual area 20 having a certain visual width H on the conveyor 1 as an image, and the component 50 that can be taken out based on the component information 3 from the visual recognition device 2. The determination device 4 for determining the take-out order, the conveyor control device 14 for controlling the drive of the conveyor 1, particularly the first conveyor 34 based on the component information 3, and the timing for capturing the image to the visual recognition device 2 are also set. The image capturing control device 5 is controlled based on the movement amount of the conveyor and the presence or absence of the component 50 on the conveyor 1.

[0039] Based on FIGS. 2 and 3, illustrating the manner of control of each element of the component supplying apparatus described above to below.

As shown in these figures, the component information 3 captured by the camera 35 and processed by the visual recognition device 2
Is sent to the discriminating device 4, and the discriminating device 4 discriminates whether or not the robot 36 can take out the parts. The order of taking out the respective parts 50 is determined. From the take-out order data and the part information 3, the taken-out part data is taken out. 13 is created. Note that H indicates a constant visual width and 20 indicates a visual area, and an image in such a limited range is captured from the camera 35.

The component information 3 is also sent to the image capturing control device 5, and the sensor signal 7 from the component arrival detection sensor 6 which detects the arrival of the component 50 and the movement amount measuring device 10 of the conveyor 1 are transmitted. The optimum image capture timing is determined from the conveyor movement amount data 11.
Then, the timing signal 12 is sent to the visual recognition device 2 to capture the next image. The moving amount of the conveyor 1 is measured by the encoder 9 attached to the conveyor 1.

The taken-out parts data 13 created by the discriminating device 4 is sent to the conveyor control device 14 and is determined as the optimum conveyor moving speed at which the robot 36 can take out the parts 50 without loss time. Such moving speed data is sent to the conveyor drive device 15 as a speed command 16, and the conveyor motor 17 is controlled by this speed command 16.

Further , the taken-out parts data 13 is sent to the robot controller 18 and converted into robot coordinates together with the conveyor movement amount data 11, and the robot 36 and the hand.
The part 50 is taken out by 36a. Further, the hand 36a has a component gripping sensor, and determines whether or not the component 50 is reliably taken out.

As described above, in this embodiment, the conveyor 1 is continuously moved, and the visual recognition or the component 50 is performed.
Since the handling is performed without stopping the conveyance of the component 50, the component 50 can be efficiently taken out. Then, based on the component information 3 from the visual recognition device 2, when the number of the components 50 to be taken out is large, the conveyor 1
Since the robot 36 has a low waiting time by slowing down the moving speed of the robot, and the control of capturing an image into the visual recognition device 2 is performed, extra visual recognition is not performed, and therefore high speed is achieved. The component 50 can be reliably taken out.

[0045] below with reference to FIGS. 4 to 5, component recognition information 3 according to the visual recognition device 2, and conveyor 1
The control of the moving speed will be described more specifically.

FIG . 4 is a schematic perspective view showing the positional relationship between the conveyor 1 and the visual area 20, and FIG. 5 is a flowchart showing the control of the moving speed of the conveyor 1.

In FIG . 4, section k-1 to section k + 4 are the conveyors 1 divided into visual areas 20 and numbered. The image is captured by the visual recognition device 2 when the visual area 20 of the section k + 3 enters the visual area 20 below the camera 35. And the discrimination device 4
The parts 50 that can be taken out are discriminated by the following, and a taken-out parts table which is taken-out parts data 13 is created as shown in Table 1 below.

[0048]

[Table 1]

The extraction of the part 50 by the robot 36, the visual area 20 of the section k is performed when entering the removable region of the lower robot 36, the robot 36 performs the extraction of the part 50 in synchronization with the conveyor 1 There is. Further, in the parts take-out table shown in Table 1, the robot 36 erases all the parts 50 at the time when they are taken out and writes new data, and the memory is effectively used.

[0050] Hereinafter, description will be given control of the movement speed of the conveyor 1 with reference to FIG. After the parts 50 in the area of the section k are taken out, the number of the parts 50 in the section k + 1 is read from the taken-out parts table of Table 1 (step 1), and the number of parts is obtained from the speed reference table shown in Table 2 below. Read the moving speed. (Step 2)

[0051]

[Table 2]

[0052] Then, to monitor the amount of movement of the conveyor 1 (step 3), the left border of the interval k + 1 shown in FIG. 4, the change rate begins upon reaching the speed change start position before removable areas (Step 5), section k +
The speed change is completed when 1 enters the area that can be taken out by the robot 36.

In order to stop the conveyor 1, after the conveyor 1 is stopped (step 6), the robot 36 takes out the parts 50 in the section, and the last part 50 is removed.
When one of them is left, a conveyor movement start signal is issued from the robot controller 18 to start the movement of the conveyor 1 (step 8). The last part 50 is taken out by the robot 36 synchronized with the conveyor 1 which started moving.

Based on FIGS. 6-8 , a component 50 is described below.
The recognition of the component information 3 relating to the posture will be specifically described.

FIG . 6 is an explanatory diagram showing the stable postures of the cylindrical component A having a terminal pin at one end and the disc-shaped component B, and registration patterns corresponding to the stable postures. FIG. 7 is an explanatory diagram showing the posture data of the part A. FIG. 8 is an explanatory diagram of the recognition image.

As shown in FIG . 6, three kinds of stable postures can be considered for the part A. That is, as shown in column 1, the terminal pin is directed upward, as shown in column 2, it is tilted and stabilized, and as shown in column 3, the terminal pin is stable toward the conveyor 1 side. The registered patterns for these three types of stable postures are registered respectively. Further, in the case of the part B, two types of stable postures, that is, a separated state and a state in which they are not separated but overlapped with each other, are possible. For these stable postures, two types of registered patterns shown in FIG. Has been done.

[0057] in association with these registered pattern, performs matching including the rotation is to recognize each part 50. As the recognized part information 3, the product type, position (x, y, z) and attitude (Θ, α, Γ) are output.

As shown in FIG . 7, as attitude data, Θ is a rotation angle when the component spindle is vertical, α is a rotation angle in a vertical plane of the component spindle, that is, a component tilt angle, and Γ is a horizontal plane of the component spindle. Indicates the rotation angle at, that is, the direction in which the part fell.

As shown in FIG . 8, the posture data is recognized from the recognized image as follows. In other words, the part A2 shown in this figure matches the registered pattern of the part A that has rotated γ0 and is tilted and stable as shown in the second column, and x and y are obtained from the coordinate values at this time. The value of is γ0. The height z and the tilt angle α are values determined from the shape of the part A. The value of Θ can be obtained from the positional relationship of the terminal pins, which is the characteristic point of the part A.
In addition, the component information values in the other two stable states are also
The same can be obtained.

Further , the component information values of the components B1 and B2 can be similarly obtained. It should be noted that the parts B1 may overlap with each other and thus have a low degree of coincidence with the registered pattern, but by setting the level of the degree of coincidence at the time of matching to be low, the recognition is not hindered.

[0061] Based on FIGS. 9 to 11, will be specifically described the processing in the case of detecting a component 50 on the boundary line of the visual area 20 below.

As shown in FIG . 9, in the image of the visual area 20, there are four parts 50, two of which are at the boundary on the conveyor movement start side. An image of the component 50 is searched in the x-axis direction for each y-axis coordinate from the conveyor movement start side of such an image together with its length. The graph of FIG. 10 shows such a search result, and at the positions where the images are present, mountains corresponding to the respective images are formed. Based on this result, the position of the center of gravity and the maximum length of each mountain are output as a search result as shown in Table 3 below.

[0063]

[Table 3]

Next, based on the search result, in accordance with the flowchart shown in FIG. 11, is to control the image capture. That is, the arrival of the component 50 within the visual field of the camera 35, that is, within the visual area 20 is detected by the component arrival detection sensor 6 (step 1), and a timing signal 12 for capturing an image is output to the visual recognition device 2 ( Step 2). Further, the amount of movement of the conveyor at this time is read (step 3). Further, the component information 3 of the visual recognition device 2
The recognition result data is obtained (step 4), the number of recognized parts is checked, and if the number of parts is 0, the process returns to step 1. Next, from the search result of the image in the visual area 20,
It is checked whether or not there is a part 50 which cannot be recognized due to the boundary (step 6). If there is no part 50 on the boundary, the next image is captured when the visual width H of the visual area 20 is moved. In some cases, the capture position of the next image is reduced by the maximum length of the image of the component 50 at this boundary so that the image of this component 50 will be included in the next image.

[0065] This control eliminates the component 50 that does not work in not recognize, because removable part 50 is increased, with extraction of efficient component 50 is made, by reducing the component 50 to be circulated, Parts by circulation
50 damage can be prevented.

The determination of the removable components 50 and the determination of the removal order will be specifically described below with reference to FIGS. 12 to 14.

FIG . 12 is a flow chart showing the control of the discriminating apparatus 4, FIG. 13 is an explanatory view showing the interference distance D of the components 50 in proximity, and FIG. 14 is an explanatory diagram showing a concrete example of the proximity state. , (A), (B) and (C) show different proximity components in different proximity states.

As shown in FIG . 12, in the discriminating device 4, each component 50 is detected from the position data of the component 50 by the visual recognition device 2.
Calculate the distance between parts. Then, the component 50 sufficiently separated from the other component 50, the adjacent component 50, or the overlapping component 50 by the distance between the components is classified into three types of a separation component, a proximity component, and an overlapping component, respectively. , A classification table shown in Table 4 below is created.

[0069]

[Table 4]

[0070] Then, the separation part in the classification table examine its attitude, the only parts 50 of the posture which is the extraction target is written into the extraction component data table shown in Table 1 above. Next, regarding the proximity component in the classification table, the proximity component extraction method reference table as shown in Table 5 below is referred to, the extraction method of the component 50 is checked, and the component information 3 is written in the extraction component data table.

[0071]

[Table 5]

In the above procedure, when the respective parts 50 are classified according to the distance between the parts, as shown in FIG.
Depending on whether or not there is one part 50 taken out the other part
The interference distance D, which is the limit of whether or not 50 is affected, will be different. Therefore, an inter-part distance reference table as shown in Table 6 below is created in advance for classification.

[0073]

[Table 6]

[0074] For example, if the inter-component distance between parts 50a and the part 50b of FIG. 13 is a case where L, the and the direction of each part 50b with respect to the main axis of the parts 50a, was 30 degrees, The interference distance D between the component 50a and the component 50b can be read from Table 6 as LL0 and LLl, respectively, and L- (LL0 + LLl) is calculated. If this value is almost 0, it is determined that they are close to each other, and If they are separated, it is judged that they are overlapped if they are-. FIG. 14 shows a specific example of the proximity component. As shown in FIG. 14A, when the components are stable in the same direction, it does not matter which one is taken out. But,
In the case where the two are leaning against each other and stable as shown in (B), it is highly possible that the other is moved when one is taken out. Therefore, in consideration of the loss time for picking up the moving part 50, it is preferable to take out one of them and not take out the other. Also, like (C),
In the case where the parts 50b are upside down, if the part 50b is taken out first, the protrusion below the part 50b may come into contact with the protrusion above the part 50a and move the part 50a. Therefore, if the component 50a is taken out first, the component 50 can be taken out efficiently and reliably.

Since the efficient take-out order differs depending on the postures of the parts 50 that are close to each other, a close-up part take-out method reference table as shown in Table 5 is created to determine the take-out method of the parts 50. Is particularly preferable. In addition, in this Table 5, the method of taking out the component 50 is determined only by the difference in the posture, but when different types or three or more components 50 are close to each other, the taking-out method should be decided by the same idea. You can

As described above, by taking out from the separated parts and then taking out the adjacent parts by a predetermined method, it is possible to take out the other parts 50 with less influence and to improve the taking out efficiency.

A method of detecting the positional deviation of the other component 50 caused by the removal of the one component 50 in the above-described method of extracting the adjacent component will be specifically described with reference to FIGS. 15 to 16.

FIG . 15 is an explanatory diagram of the above method, in which the camera
The arrangement of 35a, 35b, the robot 36 and the conveyor 1 is shown. FIG. 16 is an explanatory diagram showing image processing,
(A), (B), and (C) show images in respective processing states.

As shown in FIG . 15, two cameras 35a and 35b are used, and the camera on the front side in the conveyor moving direction is used.
The component information 3 for taking out the component 50 is obtained at 35a. Then, the camera 35b captures the recognition area 20 having the same recognition width H as that of the camera 35a as an image in the parts extraction area of the robot 36.

FIG . 16A shows a recognition image of the camera 35a, and FIG . 16B shows a recognition image of the camera 35b after the one of the adjacent parts 50 is taken out. (C) shows a difference image between the recognition images of (A) and (B).

When one of the parts 50 is taken out and the position of the other part 50 is not displaced, the area in the difference image is equal to the area of the taken out part 50, and the position of the other part 50 is displaced. In this case, since the area is larger than the area of the taken-out component 50, the positional deviation can be detected. As described above, since the positional deviation can be detected easily and surely, the efficiency of taking out the adjacent parts can be easily realized.

Based on FIG . 17, the parts in various postures will be described below.
A method for taking out 50 will be specifically described. This figure is a view showing the hand 36a of the robot 36. (A) is a schematic front view of the hand 36a, (B) is a perspective view of one rotating body 36a1 of the hand 36a, and (C) is the same. FIG. 9 is a perspective view of the other rotating body 36a2 of the hand 36a.

The hand 36a has two jaws 36 on the chuck 36a3.
The claws 36a4 are provided with the a4 facing each other, and a rotating body 36a1 and a rotating body 36a2, which are driven by a motor 36a5 to rotate, are attached to the respective claws 36a4. One rotating body 36a1
Is designed so that the part 50, which has a recessed tip end, is held in the center. The other rotary body 36a2 has a guide groove.

In the robot controller 18, the visual recognition device 2
According to α of component information 3 of (representing the tilt angle of component 50),
It controls the motor 36a5 and inserts the guide groove of the rotating body 36a2 into the part 50.
Adjust to the posture of. Then, the robot 36 is controlled according to the values of x, y, z and Γ, and the component 50 is taken out. Next,
The motor 36a5 is controlled again, the component 50 is adjusted to the posture at the time of supply, and further, the component 50 is rotated according to the value of Θ and supplied to the next process.

As described above, it is possible to take out even the parts 50 which have fallen down or whose front and back are reversed and supply them in a normal state, and the ratio of the parts 50 that can be taken out is increased. Can be taken out.

A method of setting the component loading amount in the parts feeder 66 will be specifically described below with reference to FIGS. 18 to 19.

FIG . 18 is a perspective view showing the periphery of the parts feeder 66 including the conveyor 1 and the camera 35, and FIG. 19 is a flowchart showing the control of the parts feeder 66.

As shown in these figures, prior to the removal of the component 50 by the robot 36, the component 50 is moved to the conveyor 1
It circulates above and examines the average number of parts that can be taken out. The average value is examined for each case of the data table as shown in Table 7 below while changing the amplitude of the parts feeder 66 and the amount of mounted parts.

[0089]

[Table 7]

From this result, the amplitude value and the component mounting amount that maximize the average value of the number of parts that can be taken out are obtained,
The robot 36 is actually operated to supply the parts. Then, while supplying the quantity of the parts taken out by the robot 36 from the hopper 65, the parts are continuously supplied.

Since the parts feeder 66 is automatically adjusted as described above, labor can be saved, and the parts 50 can be circulated in the state in which the number of taken-out parts is the largest.
It is possible to improve the extraction efficiency of the component 50.

[0092] Hereinafter, with reference to FIGS. 20 to 28, a description will be given component supply device having a structure different from the above description.

[0093] FIGS. 20 through 28 are explanatory views schematically showing these parts feeder, Fig. 20, 21, 22, 23, 24 and 25 or 27, respectively different separation promoting device 25,
, And FIG. 28 shows the conveyor 1 having different structures.

[0094] Incidentally, the function of the separation promoting unit 25 to be described below, in apparatus described in FIGS. 1 to 19 described above, by the vibration of the vibrating feeder 31, change the position or posture of the component 50, again It is distributed.

As shown in FIG . 20, this component feeder is
Looped conveyor 1 and parts 50 on this conveyor 1
The apparatus is different from the apparatus described in FIGS. 1 to 19 in that it has a hopper 65 for charging and a separation promoting apparatus 25 which is a rotating brush. This separation promoting device 25
Is provided on the downstream side of the hopper 65, and the parts 50 not taken out by the robot 36 are the loop-shaped conveyor 1
What is redispersed by this brush and is ready to be taken out is taken out by the robot 36.

In addition, the visual area 20 having a constant visual width H is sequentially captured as an image on the conveyor 1,
The visual recognition device 2 for recognizing the number, position or orientation of the parts 50 as the part information 3, the judgment device 4 for judging the parts 50 that can be taken out based on the part information 3 and the order of taking them out, and similarly the part information. The conveyor control device 15 for controlling the drive of the conveyor 1 based on 3 and the image acquisition control device for controlling the image acquisition timing to the visual recognition device 2 based on the conveyor movement amount and the presence or absence of the component 50 on the conveyor 1. 5 and the robot 36 that takes out the component 50 based on the discrimination result of the discrimination device 4 and supplies it to the next process
1 and FIG.
It is exactly the same as the device described in FIG. 19, and is exactly the same in the subsequent figures.

The separation promoting device 25 shown in FIG . 21 is a swinging device of the conveyor 1, and by swinging the conveyor 1 by this swinging device, the parts 50 are redispersed as in the case of FIG. .

The separation promoting apparatus 25 shown in FIG . 22 is configured such that the conveyor 1 is a mesh conveyor and air is blown from an air blowing device 25b below the mesh conveyor.

The separation promoting device 25 shown in FIG . 23 is composed of a separating means based on the characteristic values of the parts such as the weight or shape of the parts 50, and is compared by the centrifugal force generated by the rotation of the looped conveyor 1. The heavy component 50 is separated to the outside. Therefore, the parts 50 are sorted on the conveyor 1 according to the weight, so that the parts can be taken out smoothly.

The separation promoting device 25 shown in FIGS. 24 to 25 is
It is configured by a separating means based on the characteristic value of the component such as weight or shape of the component 50, and is divided on the conveyor 1 according to the height of the component 50 by the guide bars 25b, 25b having different heights. is there. That is, guide bars 25b, 25b having different heights are provided on the conveyor 1, so that the lower guide 25a is the upstream side and one end side of the conveyor 1, and the higher guide 25b is the other end side. In order to guide the component 50 from one end to the other end, the guides 25a and 25b are provided so as to be inclined with respect to the conveying direction of the conveyor 1 to form a separation promoting device.

FIG . 25A is a front view showing the guide bars 25a and 25b from the front side in the traveling direction of the conveyor 1, and FIG . 25B is a top view. As shown in these drawings, the guide bar 25a is disposed almost on the surface of the conveyor 1. Therefore, as shown in (B), the component 50 flowing downward in this figure is moved to the left side at one end, and then the higher guide bar 25b causes the lower component 50a indicated by ○ to be directly attached to this guide bar 25b. Pass through
The higher part 50b indicated by □ is again moved to the opposite side and sorted on the conveyor 1.

When the parts 50 having different heights are mixed, the removal of the part 50a having a low height has a high possibility of affecting the removal of the part 50b having a high height. By separating the parts 50 in accordance with the above, the taking out can be surely performed.

In the separation promoting apparatus 25 shown in FIG . 26, separation rods having large to small intervals are arranged in a line on the conveyor 1 in a line, and the rows of the separation rods are arranged in the conveying direction of the conveyor 1. The separation promoting device is configured to be inclined.

[0104] Thus, with this configuration, small parts 50 than the distance of the separation rod passes between the separating bar, a large part 50 is moved along the row of the separation bar 50 that the inclined, conveyor It is possible to divide the parts 50 on the 1 according to the size. Therefore, the position and orientation are updated, and the removable parts 50 are obtained, and the parts 50 are divided, which facilitates the removal.

In the separation promoting device 25 shown in FIG . 27, the vibrating device attached to the lower part of the conveyor 1 vibrates the parts 50 on the conveyor 1 to update the dispersed state such as the position or inclination, and the separation promoting device 25 can be taken out. The part 50 is obtained.

Further, in the component supplying apparatus shown in FIG. 28, the conveyor 1 is vertically arranged in multiple stages to effectively use the space, and in this figure, particularly two stages are shown.

[0107] That is, upper and lower stages of the conveyor 1a, has a 1b, part 50 supplied to the upper conveyor 1a one end,
It is configured so that it falls from the other end to the lower conveyor 1b, returns to the hopper 65, and is supplied again from the hopper 65 onto the conveyor 1a.

Further, the component supply device shown in FIG.
Similar to the 28, it has upper and lower two-stage conveyors 1a and 1b, but is different in the following points. This figure is a schematic side view of the component supply device.

[0109] That is, the visual area 20 by the camera and the position of the lower feed conveyors 1b front, derived is a component 50 in the rear end of the feed conveyor 1b. Then, the component 50 that could not be taken out is sent to the upper return conveyor 1a. This upper return conveyor 1a
The rear end is connected to the return parts feeder 66a,
A straight parts feeder 66b is arranged on the lower side of the return parts feeder 66a. The straight parts feeder 66b is connected to the lower conveyor 1b. Also, the return parts feeder 66a
A hopper 66 for replenishing and supplying the component 50 as needed is arranged above.

Further , in such a component supply apparatus, the time required to obtain the component information such as the position and orientation of the component 50 from the image captured by the camera is defined as the image processing time, and a plurality of consecutively captured components are acquired. Obtain the average image processing time of the image, compare the average image processing time and the supply determination threshold value, from the hopper 65 to the parts feeder 66a,
Parts are supplied to 66b.

[0111] Based on FIGS. 30 and 31, an example of the component feed more specifically. FIG. 30 is an explanatory diagram showing how to supply the above components, and FIG. 31 is a graph showing the relationship between the number of circulating components and the image processing time.

[0112] That is, first, as shown in the upper frame of FIG. 30,
Parts for each of the latest 1-10 consecutive images
50 recognition numbers N1 to N10 and corresponding image processing time T1 to
I am trying to get the data of T10. Then, as shown in the lower frame, the total of T1 to T10 is divided by 10 to obtain the average image processing time. In addition, the correspondence relationship between the number of circulating parts 50 and the average image processing time is obtained in advance as a circulating number table in the graph shown in FIG. Then, using this circulation number table, the circulation number of the corresponding component 50 is obtained and compared with the supply determination threshold value to determine whether or not the component 50 can be supplied. In other words, if the number of circulating parts is greater than the supply determination threshold value,
Since there are a sufficient number of 50 and they are efficiently taken out, the parts 50 are not supplied from the hopper 66. On the contrary, when the number is small, unnecessary image processing is performed, and the waiting time of the robot for taking out is increased, so that the component 50 is supplied. For example, in the case shown in FIG. 31, the average image processing time is about 2.9 seconds, which is greater than the supply determination threshold value of 130, and the number of parts 50 circulating on the conveyor 1 is sufficient. Therefore, the component 50 is controlled so as not to be supplied from the hopper 66.

[0113] Incidentally, the smaller the average image processing time from the supply determination threshold, it is also preferable to increase the supply amount of the component 50.

In FIG . 29, reference numeral 70 denotes a component sensor provided on the return parts feeder 66a, and this component sensor 70 determines whether or not the component 50 exists below the component sensor 70. Regardless of the average image processing time, the component sensor 70 checks the component existence probability within a certain determination time, compares the component existence probability with the supply determination threshold value, and returns from the hopper 65 to the return part. It is also possible to supply the components on the feeder 66a.

[0115] Incidentally, the parts sensor 70, if the downstream side of the component pickup position of a path component 50 circulates,
It may be provided at any position.

[0116] illustrating a specific example and operation of such components sensor 70, in the following with reference to FIG. 32 through FIG. 35. 32 is a perspective view showing the component sensor 70, FIG. 33 is an explanatory view showing an operation of the component sensor 70, and FIG. 34 is an explanatory view showing a set distance of the component sensor 70. FIG. 35 is an explanatory diagram showing an example of measuring the component existence probability, where (A) shows the case where the component is supplied and (B) shows the case where the component is not supplied.

As shown in these figures, the component sensor 70 is a general-purpose beam sensor and is of a twin-lens range-finding type limited distance reflection type. As shown in FIG. 33, when the target object is irradiated with a beam and the target object is at a position closer than the set distance, the component sensor 70 is in the ON state, and when it is at a far position, it is in the OFF state. Therefore, as shown in FIG. 34, if the distance is set to the substantially central position of the component 50, as shown by an imaginary line, the component 50 is turned on when the component 50 exists below the component sensor 70, and the component 50 is turned on. You can check for

[0118] Therefore, in the actual operating state, as shown in (A) of FIG. 35, small parts 50 on the return parts feeder 66a, part sensor 70 the time within a predetermined judgment time that is in the OFF state The component 50 is supplied when the ratio becomes equal to or higher than the supply determination threshold value α%. On the contrary, as shown in (B) of FIG. 35, when there are many parts 50 and the ratio of the time when the part sensor 70 is in the OFF state is equal to or less than the supply determination threshold value α%, the parts 50 are It is not supplied.

In such a component feeding method, the method based on the average image processing time described above is based on the amount of components before being taken out, whereas the method based on the amount of components after being taken out from the hopper 65 is used. The difference is that it is trying to determine the component supply amount.

[0120] Also, so as to take the respective determination results and the by the component sensors 70 and components existence probability average image processing time, it is also preferable to perform the more accurately the component supply.

Further , in the above-described component supply apparatus, when the components are continuously supplied from the hopper 65, the removable parts 50 are not recognized in the image of the recognition area 20 continuously. When the average image processing time is smaller than the notification determination threshold value, the result is notified by a lamp or sound. This notification determination threshold value is
The value is set so that the average image processing time is too short.

FIG . 36 shows an apparatus in an operating state in which a notification is made by such a notification determination threshold value, in which the hopper 65 is empty and the parts 50 are not actually supplied. . Therefore, the worker can take a procedure such as supplying the component 50 to the hopper 65 based on the notification result.

Further , in the above component supplying apparatus, the component 50 is recognized in the image in the recognition area 20 while the component supply from the hopper 65 is stopped, but there is no removable component 50. When the states are continuous, and when the average image processing time is larger than the component clogging determination threshold value, the operation of the entire apparatus performing this component supply method is stopped. The component clogging determination threshold value is set to a value such that the average image processing time is too long even if the component 50 is taken out normally.

FIG . 37 shows an apparatus in an operating state in which the operation of the entire apparatus is stopped by such a component clogging determination threshold value. As described above, the parts are clogged near the take-out position of the parts 50, and in this state, the image processing time is extremely long even if the parts supply is stopped.

Further , FIGS. 36A and 36B show a state in which the recognized component 50 is entangled in the image of the recognition area 20, and FIG. 36B is a partially enlarged view of FIG. . In such an entangled state, the entangled component 50 may be attached and taken out in the entangled state with the component 50 to be taken out. Therefore, the robot that takes out the parts 50 cannot supply the taken-out parts 50 one by one to the predetermined position, which is unsatisfactory.
Is determined not to take out.

[0126] determination of such entangled state scans the periphery of the part 50 is performed by checking the number of picture elements and scanning lines and parts 50 of the peripheral overlap. However, if the entangled portion is outside the recognition area 20, the above check cannot be performed.

[0127] Thus, the component 50 partially from the image are recognized protrudes, by determining a non-removable component 50 by the robot, so that to avoid a problem as described above.

[0128]

According to the component supply method of the present invention, the moving speed of the conveyer or the image capture is controlled so as to be suitable for the component removal, so that the components are reliably removed. In addition, the loss time such as the waiting time of the robot is small, and the parts can be taken out efficiently. Therefore, the method of the present invention is a component supply method capable of extracting a component at high speed and reliably and supplying this component to the next step. Since the component extraction efficiency is improved, it is less likely to circulate without extracting the component. , The possibility of damaging the parts is also reduced.

According to the first aspect of the present invention, by taking out the adjacent component next to the separated component, it is possible to efficiently take out the component.

According to the second aspect of the present invention, when taking out the adjacent parts, if one part is taken out and then the other part is displaced, the taking out of this part is postponed.
Therefore, it is possible to take out the component efficiently without fail and to take out the component reliably, without causing loss time, and to take out the component efficiently.

According to the third aspect of the present invention, the amount of parts to be fed from the parts feeder is determined so as to maximize the number of parts that can be taken out, so that the parts can be taken out efficiently.

In the invention according to claim 4 , since the conveyor is continuously moved and the visual recognition or the handling of the parts is performed without stopping the transportation of the parts, the parts can be efficiently taken out. You can Then, since the moving speed of the conveyor or the capturing of the image into the visual recognition device is controlled based on the component information from the visual recognition device, the component can be reliably taken out at high speed.

[0133] Further, determination device obtains component information from the vision recognition unit so to determine the optimum extraction procedure corresponds to the state of each part, can be efficiently extracted components.

According to the fifth aspect of the present invention, it is possible to estimate the amount of parts before taking out on the conveyor from the average image processing time, and supply the amount of parts corresponding to this amount of parts from the hopper. Therefore, it is possible to circulate on the conveyor the amount of the components that can be most efficiently taken out, and the efficiency of taking out the components is improved.

According to the sixth aspect of the present invention, it is possible to estimate the amount of parts that circulates without being taken out on the conveyor from the part existence probability, and supply the amount of parts corresponding to this part amount from the hopper. Therefore, it is possible to circulate the parts on the conveyer in an amount such that the parts are most efficiently taken out, and the parts taking-out efficiency is improved.

According to the seventh aspect of the invention, since it is based on both the estimated values of the amount of parts before taking out and the amount of parts after taking out, the amount of parts supplied from the hopper can be finely determined, and the parts taking efficiency can be improved. Will be improved.

According to the eighth aspect of the present invention, since the hopper is emptied and it is determined that the parts are not supplied, the hopper is supplied with the parts based on the notification result. Can respond to empty conditions.

According to the ninth aspect of the invention, it is determined that the parts are clogged on the upstream side of the parts take-out position. Then, since the operation of the entire apparatus is stopped in such a state, it is easy to perform the work of eliminating the component clogging and restoring it without continuing to supply the components to further worsen the component clogging. .

[Brief description of drawings]

FIG. 1 is a perspective view showing a component supply device according to an embodiment of the present invention.

FIG. 2 is a control block diagram showing control of each component of the component supply device.

FIG. 3 is a perspective view schematically showing each component of the above component supply device.

FIG. 4 is a schematic perspective view showing a positional relationship between a conveyor and a visual area of the component supply device.

FIG. 5 is a flowchart showing control of a conveyor moving speed of the above component supplying apparatus.

FIG. 6 is an explanatory diagram showing a stable posture of a component and its registration pattern in the visual recognition device of the component supply device.

FIG. 7 is an explanatory diagram showing posture data of parts in the same visual recognition device.

FIG. 8 is an explanatory diagram of a recognition image regarding a posture in the visual recognition device of the above.

FIG. 9 is a plan view showing a component image in a visual area in the visual recognition device of the above.

FIG. 10 is a graph showing a search result of an image in a visual area in the visual recognition device of the above.

FIG. 11 is a flowchart showing the control of the image capturing control device of the component supply device.

FIG. 12 is a flowchart showing the control of the discriminating device of the component supplying device.

FIG. 13 is an explanatory diagram showing an interference distance between adjacent components in the visual recognition device of the component supply device.

FIG. 14 is an explanatory diagram showing a specific example of a proximity state of a proximity component in the visual recognition device of the component supply device,
(A), (B), and (C) show different proximity states.

FIG. 15 is an explanatory diagram of a method of detecting a positional deviation of a component in a method of taking out a neighboring component of the component supply device, showing the arrangement of a camera, a robot and a conveyor.

FIG. 16 is an explanatory diagram showing image processing in the above method, in which (A), (B), and (C) show images in different processing states.

17A and 17B are views showing a robot hand of the above-mentioned component supply device, wherein FIG. 17A is a schematic front view of the hand, FIG. 17B is a perspective view of one rotating body of the hand, and FIG. It is a perspective view of the other rotary body of a hand.

FIG. 18 is a perspective view showing the periphery of a parts feeder including a conveyor and a camera of the above parts feeder.

FIG. 19 is a flowchart showing the control of the parts feeder of the above parts feeder.

FIG. 20 is an explanatory view schematically showing a component supply device according to another embodiment of the present invention.

FIG. 21 is an explanatory view schematically showing a component supply device according to a different embodiment.

FIG. 22 is an explanatory view schematically showing a component supply device according to a different embodiment.

FIG. 23 is an explanatory view schematically showing a component supply device according to a different embodiment.

FIG. 24 is an explanatory view schematically showing a component supply device according to a different embodiment.

FIG. 25 is another explanatory view showing the separation promoting device in the above-mentioned component supply device, (A) showing a front view and (B) showing a top view.

FIG. 26 is an explanatory view schematically showing a component supply device according to a different embodiment.

FIG. 27 is an explanatory view schematically showing a component supply device according to a different embodiment.

FIG. 28 is an explanatory view schematically showing a component supply device according to yet another embodiment.

FIG. 29 is a schematic side view of a component supply device according to yet another embodiment.

FIG. 30 is an explanatory view showing a method of supplying a component from a hopper in the component supply device of the above.

FIG. 31 is a graph showing the relationship between the number of circulated components and the image processing time in the component supply device of the above.

FIG. 32 is a perspective view showing a component sensor in the component supply device of the above.

FIG. 33 is an explanatory diagram showing the operation of the component sensor in the component supply device of the above.

FIG. 34 is an explanatory diagram showing set distances of component sensors in the component supply device of the above.

FIG. 35 is an explanatory diagram showing an example of measuring the component existence probability by the component sensor of the component supply device of the above, (A) shows the case where the component is supplied, and (B) shows the case where the component is not supplied. .

FIG. 36 is a schematic side view showing an operating state in which the hopper of the component supply device of the above is empty.

FIG. 37 is a schematic side view showing an operating state in which a component jam has occurred in the component supply device of the above.

FIG. 38 shows a state in which a recognized component is entangled in an image in a recognition area of the component supply device according to any one of the embodiments of the present invention, and (B) is a partially enlarged view of (A).

[Explanation of symbols]

1 conveyor 2 Visual recognition device 3 parts information 4 Discriminator 5 Image capture controller 6 parts arrival sensor 7 sensor signal 8 conveyor 9 encoder 10 Moving amount measuring device 11 Conveyor movement amount data 12 Timing signal 13 Extracted parts data 14 Conveyor control device 15 Conveyor drive 16 Speed command 17 Conveyor motor 18 Robot controller 31 Vibration feeder 32 vibrating feeder 34 First conveyor 35 camera 36 robots 37 Reversing device 38 Second conveyor 39 Controller 40 pallets 50 parts 65 hopper 66 parts feeder 70 parts sensor 71 scan lines

─────────────────────────────────────────────────── ─── Continuation of the front page (56) Reference JP-A-6-238584 (JP, A) JP-A-3-106701 (JP, A) JP-A-7-48018 (JP, A) JP-A-3- 239487 (JP, A) JP-A-6-305551 (JP, A) JP-A-6-115667 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) B25J 13/08 B25J 13 / 00

Claims (9)

(57) [Claims] 1. A parts supply method in which a plurality of parts are loaded from a parts feeder onto a conveyor, the conveyed parts are taken out by a robot and supplied to the next process, and some parts that cannot be taken out are returned to the parts feeder. In, a visual recognition device that sequentially captures a visual area having a certain visual width on the conveyor as an image and recognizes the number, position or posture of the components in the visual area as the component information and recognizes the component information regarding the posture.
The parts to be removed are determined based on the
Calculate the inter-part distance based on the part information about
Separate each part based on the distance between the parts
Items or overlapping parts and separate parts first
Next, take out the adjacent parts, and do not take out the overlapping parts.
In addition to the above , the moving speed of the conveyor is controlled to be suitable for taking out the robot and the image is efficiently taken in, and the parts at positions or postures that can be taken out by the robot are taken out from the conveyor. Parts supply method. 2. When taking out adjacent parts, after taking out one of the parts, the position of the other part is checked in the image after taking out, and the position is shifted compared with the image before taking out. In the case, the removal of the other component is postponed, and the component supplying method according to claim 1 . 3. The amount of parts input from the parts feeder,
The component supply method according to claim 1, wherein the determination is performed in proportion to a ratio of the removable components, which is determined based on the component information. 4. A loop-shaped conveyor that continuously moves and circulates parts, a part feeder that inserts parts on the conveyor, a moving amount measuring device of the conveyor, and parts that are dispersed on the conveyor to form the parts. The position and orientation of the parts are updated and the separation facilitator that makes them removable parts and the visual area with a certain visual width on the conveyor are sequentially captured as an image, and the number, position or attitude of the parts is recognized as part information. Calculate the distance between parts based on the visual recognition device and this part information, and
Separate each part based on separate parts, adjacent parts or overlap
Classifying into parts, a distinguishing device that distinguishes the parts that can be taken out and the order in which they are taken out, a conveyor control device that similarly controls the drive of the conveyor based on the part information, and the timing of image capture to the visual recognition device. , The image capture control device that controls based on the amount of movement of the conveyor and the presence or absence of parts on the conveyor, and the separated parts first based on the judgment result of the judgment device , and then the adjacent parts.
A parts supply device comprising a robot and a hand for supplying parts to the next process by picking up parts without protruding and taking out overlapping parts . 5. An image processing time is defined as a time required to obtain component information from a captured image, an average image processing time of a plurality of continuously captured images is obtained, and this average image processing time is determined to be the supply. The component supplying method according to claim 1, wherein the component is supplied from the hopper to the part feeder by comparing with a threshold value. 6. A component sensor for determining the presence / absence of a component is provided on the downstream side of a component extraction position on a path along which a component circulates, and the component sensor checks the component existence probability within a fixed determination time to detect the presence of this component. The component supply method according to claim 1, wherein the probability is compared with a supply determination threshold value, and the component is supplied from the hopper to the part feeder. 7. A conveyor is formed in a loop shape, and a part into which parts are loaded from a hopper of the conveyor is used as a return parts feeder. The return parts feeder is provided with a parts sensor for judging the presence or absence of parts, and the parts sensor. look for the part presence probability in a predetermined determination time by claim which this component existence probability and the average image processing time and the compared to the respective supply determination threshold, and performing the component supply from the hopper 5
Part supply method described. 8. The average image processing time is smaller than the notification determination threshold value in a case where the components that can be taken out are not recognized in the image continuously while the components are continuously supplied from the hopper. 8. The component supply method according to claim 5 , wherein the result is notified at times. 9. The average image processing time is clogged when the parts are recognized in the image but there are no parts that can be taken out while the parts supply from the hopper is stopped. 8. The component supply method according to claim 5, wherein the operation of the entire apparatus that is performing the component supply method is stopped when the value is larger than the determination threshold value.