Conveying system

By using imaging devices and machine learning models to detect container posture on the container manufacturing line, and using a rejection device to remove containers that do not adopt the prescribed posture, the problem of container posture detection is solved, and the yield of container manufacturing is improved.

CN116685417BActive Publication Date: 2026-04-07TOYO SEIKAN KAISHA LTD
View PDF 4 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-02-16
Publication Date
2026-04-07

AI Technical Summary

Technical Problem

Existing technologies make it difficult to easily and accurately detect the posture of containers positioned on a wide conveyor surface in a container manufacturing line, especially containers that are not in a prescribed posture.

Method used

Images of containers are captured using an imaging device, and abnormal postures are detected using a machine learning model. Combined with a rejection device, containers that do not adopt the prescribed posture are removed from the conveying system.

Benefits of technology

It enables simple and accurate detection of container posture, improves the yield rate in the container manufacturing process, and reduces the generation of defective products.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116685417B_ABST
    Figure CN116685417B_ABST
Patent Text Reader

Abstract

The conveying system includes: a conveying device configured to convey the containers in a conveying direction while the containers are arranged in a plurality of bottomed cylindrical containers in the conveying direction and a width direction orthogonal to the conveying direction; an imaging device configured to acquire an image of the containers being conveyed by the conveying device; and a detection device configured to detect, using the image, a container that does not assume a prescribed posture among the plurality of containers being conveyed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to a conveying system. Background Technology

[0002] For example, in manufacturing lines for cylindrical containers that hold beverages, food, etc., these containers are typically conveyed in a prescribed posture, such as standing upright. However, sometimes, for various reasons, such containers may lie down or deviate from the prescribed posture. In such cases, the container that has deviated from the prescribed posture needs to be removed from the manufacturing line.

[0003] For example, Japanese Patent Application Publication No. 2019-73374 discloses a conveying device that conveys a cylindrical container by guiding it along a conveying path with the container upright, and the conveying device is capable of discharging overturned containers. The conveying device includes: a sensor for detecting overturned containers; an opening for discharging overturned containers from the conveying path; and a movable member for opening or closing the opening. When the sensor detects an overturned container, the opening opens via the movable member, and the overturned container is discharged from the conveying path. Summary of the Invention

[0004] In the conveying device disclosed in Japanese Patent Application Publication No. 2019-73374, containers are conveyed in a row, making it relatively easy to detect the container posture using sensors. In container-related manufacturing lines, containers are sometimes conveyed on a conveyor surface with a width, and in such cases, it is sometimes necessary to detect the container posture.

[0005] The purpose of this invention is to easily and accurately detect containers that are not in the prescribed posture while being transported in a conveying device.

[0006] According to one aspect of the present invention, a conveying system includes: a conveying device configured to convey the containers along the conveying direction in which a plurality of bottomed cylindrical containers are arranged in a conveying direction and a width direction orthogonal to the conveying direction; an imaging device for acquiring an image of the containers being conveyed by the conveying device; and a detection device for using the image to detect containers among the plurality of containers being conveyed that are not in a prescribed posture.

[0007] According to the present invention, containers that are not in the prescribed posture can be easily and accurately detected in a conveying device. Attached Figure Description

[0008] Figure 1A This is a schematic front view showing an example of the configuration of a conveying system according to one embodiment.

[0009] Figure 1B This is a schematic top view showing an example configuration of a conveying system according to one embodiment.

[0010] Figure 2 This is a schematic top view used to illustrate the partitions of a transfer device in one embodiment.

[0011] Figure 3 This is a schematic flowchart illustrating an example of the operation of a detection device according to one embodiment.

[0012] Figure 4 This is a schematic flowchart illustrating an example of the operation of a control device in one implementation.

[0013] Figure 5 This is a diagram illustrating an example of an image captured by an imaging device and a detection example of one embodiment. Detailed Implementation

[0014] One embodiment will be described with reference to the accompanying drawings. This embodiment relates to a conveying system having a conveying device for conveying a plurality of bottomed cylindrical containers. In this conveying system, the containers being conveyed are photographed, and the photographed images are used to detect containers among the plurality of containers being conveyed that are not in a prescribed posture. Furthermore, the detected containers that are not in a prescribed posture are removed from the conveying device.

[0015] [System Structure]

[0016] Figure 1A This is a schematic front view showing an example of the configuration of the conveying system 1 in this embodiment. Figure 1B This is a schematic top view showing an example configuration of the conveying system 1 of this embodiment. The example shown here is part of a can manufacturing line. Cans 90 in the process of manufacturing are conveyed along the conveying direction from left to right in the figure. Here, as an example, the cans 90 being conveyed are manufactured in the following manner: using a can body manufacturer, while using coolant, the can body is stretched thin by deep drawing; furthermore, after the bottom is formed, the useless parts are removed to adjust the height.

[0017] The conveying system 1 includes a conveying device 10. The conveying device 10 is, for example, a belt conveyor, a slat conveyor, or the like, and has the form of a platform that extends continuously in a belt-like shape along the conveying direction. The upper surface of this platform is a conveying surface 11, and cans 90 are placed on the conveying surface 11. The conveying surface 11 moves along the conveying direction, thereby the conveying device 10 conveys the cans 90 along the conveying direction. In this example, the cans 90 have a bottomed cylindrical shape. The cans 90 are, in principle, placed on the conveying surface 11 in an inverted position with their bottoms facing upwards, and the conveying device 10 is configured to convey the cans 90 in this position. While not limited to the width direction, which is orthogonal to the conveying direction, in this embodiment, multiple cans 90 are irregularly arranged on the conveying surface 11 in both the conveying and width directions.

[0018] A washer machine 20 is provided in the conveying device 10. The washer machine 20 is a device for cleaning and drying the coolant adhering to the tank 90. ​​The washer machine 20 is configured to perform a cleaning process including spraying cleaning fluid into the tank 90 being conveyed by the conveying device 10 to clean the tank 90. ​​Therefore, the washer machine 20 is a spraying device for spraying fluid.

[0019] The cans 90, which should be upright on the conveyor surface 11, may sometimes lie down or become inverted in the cleaning machine 20, for example, due to the spraying of cleaning fluid. The conveyor system 1 of this embodiment is equipped with a detection system 50, which detects cans on the conveyor surface 11 that are in abnormal postures, other than the normal cans 91 that are inverted, such as overturned cans 92 and upright inverted cans 93.

[0020] The detection system 50 detects overturned cans 92 or inverted cans 93 based on images of cans 90 being conveyed by the conveying device 10. Therefore, the detection system 50 includes an imaging device 52 located downstream of the washing machine 20. The imaging device 52 is configured to capture images of the entire width direction of the conveying surface 11. Alternatively, multiple imaging devices 52 can be used to capture images of the entire width direction. Using the imaging device 52, images of the cans 90 being conveyed by the conveying device 10 are acquired. The imaging device 52 is not limited to this; for example, it may be connected to a PoE (Power over Ethernet) hub 54, powered from the PoE hub 54, and the captured images may be transmitted via the PoE hub 54.

[0021] The detection system 50 includes a detection device 56. The detection device 56 is a computer comprising integrated circuits such as a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), or a Central Processing Unit (CPU); for example, it may also include a Graphics Processing Unit (GPU). The detection device 56 acquires images from the imaging device 52 via a PoE hub 54. The detection device 56 uses the acquired images to detect cans 90 that are not in a normal position, such as a lying can 92 or an inverted can 93.

[0022] As an example, the detection device 56 of this embodiment is equipped with a machine learning-completed model configured to detect cans 90 that are not in a normal posture. This model is not limited to this; for example, it could be a model generated through supervised learning based on deep learning. The teaching data used could be, for example, images of only a plurality of normal cans 91, or images of overturned cans 92 among the plurality of normal cans 91, the coordinates of the overturned cans 92, and information indicating that they are overturned cans 92, or images of inverted cans 93 among the plurality of normal cans 91, the coordinates of the inverted cans 93, and information indicating that they are inverted cans 93.

[0023] By importing the learned model generated using teaching data into a small computer equipped with an FPGA, a detection device 56 can be configured as an edge device in a factory equipped with a conveyor device 10, etc. It should be noted that the detection device 56 can be connected to a network, or it can be configured to allow for remote operation such as re-editing, re-learning, and re-installation of the model.

[0024] The detection device 56 detects, for example, a tilted can 92 based on images acquired from the imaging device 52, and outputs information about the coordinates of the detected tilted can 92 and information indicating that it is a tilted can 92. Furthermore, the detection device 56 detects, for example, an inverted can 93 based on images acquired from the imaging device 52, and outputs information about the coordinates of the detected inverted can 93 and information indicating that it is an inverted can 93. In other words, the detection device 56 performs the detection of abnormal cans 90, classifies the posture of the cans 90, determines the coordinates of the cans 90, and outputs this information.

[0025] The conveying system 1 includes, for example, a rejection device 30 downstream of the imaging device 52 for rejecting the tilted cans 92 and the inverted cans 93 from the conveying device 10. The rejection device 30 includes a transfer device 31. In the rejection device 30, a gap 32 is provided in the conveying device 10 to disconnect the conveying surface 11. The upstream side of the gap 32 in the conveying device 10 is referred to as the upstream side conveying device 16, and the downstream side of the gap 32 in the conveying device 10 is referred to as the downstream side conveying device 17. The transfer device 31 operates under the control of a control device 40. The control device 40 uses information output from the detection device 56 to control the transfer device 31.

[0026] The transfer device 31 holds the upright normal can 91 by its bottom being drawn from above at the upstream conveyor 16, and moves it along the conveying direction across the gap 32, placing it on the downstream conveyor 17 of the conveyor 10. On the other hand, under the control of the control device 40, the transfer device 31 does not draw in the reclining cans 92 and the inverted cans 93, as well as the cans surrounding them, and thus does not hold these cans. As a result, the cans not held by the transfer device 31, including the reclining cans 92 and the inverted cans 93, fall through the gap 32 and are thus rejected from the conveyor 10. Thus, the control device 40, based on the coordinates of the reclining cans 92 or the inverted cans 93 determined by the detection device 56, causes the rejection device 30 to reject at least one can 90, including those not in an inverted position.

[0027] Reference Figure 2 The rejection device 30 will be further explained. Figure 2 This is a schematic top view used to illustrate the partitioning of the transfer device 31. The transfer device 31 has multiple partitions in the width direction. Figure 2 In the example shown, there are twelve zones. The number of zones can also be arbitrary. The transfer device 31 is configured to switch between transfer with and without attraction, i.e., with and without the tank 90, for each of these zones.

[0028] exist Figure 2 In the example shown, the entire area of ​​the conveyor surface 11 in the width direction is captured by images based on the fields of view A to D of the four imaging devices 52, and these images are analyzed by the detection device 56. As a result, the detection device 56 determines the position in the width direction of the container in the presence of a tilted container 92 or an inverted container 93. Based on the coordinates determined by the detection device 56, the control device 40 controls which zones in zones 1 to 12 allow the transfer device 31 to transfer the container 90, and which zones do not allow the transfer device 31 to transfer the container 90.

[0029] If no overturned can 92 or inverted can 93 is detected, and only normal can 91 is present, the control device 40 causes the transfer device 31 to transfer can 90 in all zones 1 to 12, and the rejection device 30 does not reject any can 90, but transports all can 90. On the other hand, for example, if an overturned can 92 or inverted can 93 is detected at a position corresponding to zone 5 of the field of view B, the control device 40 causes the transfer device 31 to stop, for example, the transfer of can 90 in zones 3 to 7 at the time when the overturned can 92 or inverted can 93 passes, and continues the transfer of can 90 in zones 1 to 2 and zones 8 to 12.

[0030] For example, if the tilted can 92 is not facing upwards, it will not adhere to the transfer device 31 even if the transfer device 31 is activated, and will fall into the gap 32. Therefore, the control described above is unnecessary to remove only the tilted can 92. However, if a tilted can 92 is not in a normal state, the normal cans 91 around it may be contaminated. Therefore, in this embodiment, when a tilted can 92 is detected, the normal cans 91 around it are also removed.

[0031] Just as there is a difference between a lying can 92 and an inverted can 93, the range of cans 90 that may contaminate the surrounding cans 90 may vary depending on the orientation of the can 90. Therefore, the control device 40 may also be configured to adjust the range of cans 90 that the rejection device 30 rejects based on the orientation of the cans 90 output by the detection device 56.

[0032] Consider a configuration where the transfer device 31 is not divided into sections as in this example, but instead stops transferring across the entire width. In contrast, if a configuration allows control over whether or not transfer occurs by each section, as in this embodiment, a necessary and sufficient number of cans 90 that may be contaminated can be eliminated, resulting in an increased yield.

[0033] As described above, by means of the rejection device 30, cans 90 that are not in an inverted position, such as the lying can 92 and the upside-down can 93, as well as the cans 90 that may be contaminated around them, are rejected, and only the normal cans 91 are further conveyed downstream by the conveying device 10 to the next process.

[0034] [System Actions]

[0035] The operation of the conveying system 1 will be described. The conveying device 10 operates continuously in a conveying direction to transport the items placed on the conveying surface 11. Although the conveying device 10 has a gap 32 in the middle, the items are continuously transported from upstream to downstream by the conveying device 10 because the transfer device 31 moves the items across the gap 32. Upstream of the conveying device 10, cans 90 are successively supplied to the conveying surface 11 of the conveying device 10. The washing machine 20, located in the middle of the conveying device 10, also operates continuously. The washing machine 20 continuously washes the cans 90 being transported by the conveying device 10 through a prescribed operation.

[0036] The imaging device 52 of the detection system 50, located downstream of the washing machine 20, continuously or intermittently captures images of the conveying surface 11 of the conveying device 10 to photograph all the cans 90 being conveyed. The imaging device 52 transmits the captured images to the detection device 56 via the PoE hub 54. The detection device 56 detects cans 90 that are not in the prescribed posture, i.e., cans 90 that are not inverted with their bottoms facing upwards, based on the images received from the imaging device 52.

[0037] Reference Figure 3 The flowchart shown illustrates the operation of the detection device 56. In step S11, the detection device 56 acquires an image captured by the imaging device 52. In step S12, the detection device 56 inputs the acquired image into the learned model and obtains the analysis results derived from the model. In step S13, the detection device 56 determines whether an anomaly detection, such as detecting a can 90 not in the prescribed posture, has been performed. If no anomaly is detected, in step S14, the detection device 56 outputs a signal indicating normal operation. Afterward, the process returns to step S11, and the above process is repeated.

[0038] If an anomaly is detected in step S13, the process proceeds to step S15. In step S15, the detection device 56 outputs an anomaly signal indicating that an anomaly has been detected. This anomaly signal includes the type of anomaly detected, such as information indicating that a tilted tank 92 or an inverted tank 93 has been detected. Furthermore, the anomaly signal may include information related to the coordinates of the tilted tank 92 or the inverted tank 93. Afterward, the process returns to step S11, and the above process is repeated. As described above, the detection device 56 sequentially analyzes whether there is an anomaly in the tank 90 and outputs the analysis results.

[0039] The transfer device 31 of the rejection device 30 operates under the control of the control device 40 to transfer the can 90 across the gap 32 of the conveying device 10. (Refer to...) Figure 4 The flowchart shown illustrates the control actions of the control device 40.

[0040] In step S21, the control device 40 acquires a determination signal, such as a normal signal or an abnormal signal, output from the detection device 56. In step S22, the control device 40 determines whether an abnormality exists based on the determination signal. If no abnormality is detected, in step S23, the control device 40 causes the transfer device 31 to transfer all the tanks 90 in the partition. Afterward, the process returns to step S21, and the above process is repeated. That is, during the period when no abnormality is detected, the tanks 90 are not rejected by the rejection device 30, and all tanks 90 are transported.

[0041] If an anomaly is determined in step S22, the process proceeds to step S24. In step S24, the control device 40 determines the zone where the transfer device 31 stops the transfer of the can 90 based on the anomaly signal. For example, the control device 40 determines the zone that includes a predetermined range in the case of a reclining can 92 or an inverted can 93 as the transfer stop zone, centered on the coordinates of the detected reclining can 92 or in the case of an inverted can 93.

[0042] In step S25, the control device 40 causes the transfer device 31 to transfer cans 90 for zones other than the determined transfer stop zone, and stops the transfer of cans 90 for the transfer stop zone. As a result, in the rejection device 30, cans 90 in zones other than the transfer stop zone are transferred from the upstream conveyor 16 to the downstream conveyor 17 across the gap 32 and continue to be conveyed by the downstream conveyor 17. On the other hand, cans 90 in the transfer stop zone are rejected when they fall into the gap 32 and are not conveyed thereafter.

[0043] Through the actions described above, in cases where the washing machine 20 produces cans 90 that are not in the prescribed posture, such as overturned cans 92 or inverted cans 93, the rejection device 30 removes these cans 90 and any potentially contaminated cans 90 around them, ensuring that only the cans 90 in the prescribed posture, which are unlikely to be contaminated, are conveyed downstream. As a result, for example, ink rejection caused by contamination can be prevented during the subsequent printing process after washing, thus preventing the production of defective products.

[0044] Example

[0045] Figure 5 An example of an image 80 captured by the imaging device 52 and a detection example are shown. The seemingly circular object in image 80 is the bottom of a normal can 91. The overturned cans 92 and inverted cans 93, which are mixed in with these normal cans 91, are detected by the detection device 56 as shown by quadrilateral 82 representing an overturned can and quadrilateral 83 representing an inverted can, respectively.

[0046] Thus, it can be confirmed that by inputting the image captured by the imaging device 52 into the detection device 56 which has a learned model, anomalies can be appropriately detected. Furthermore, it can be confirmed that by using the detection results obtained by the detection device 56, appropriate rejection can be performed by the rejection device 30, and only qualified products can be manufactured efficiently.

[0047] In the conveying system 1 of this embodiment, abnormal tanks 90 are detected based on images captured by the imaging device 52. Alternatively, ultrasonic sensors, optical sensors, or other sensors could be used to detect overturned tanks 92. However, to detect multiple tanks 90 arranged in both the conveying and width directions, a large number of sensors are required. In particular, in the case where the tanks 90 are irregularly arranged in both the conveying and width directions, as in this embodiment, even more sensors are needed than in the case where the tanks 90 are regularly arranged. In contrast, as in this embodiment, image-based analysis allows for the acquisition of a wide range of information with a small number of cameras, resulting in high efficiency.

[0048] Furthermore, in detection using the sensor described above, since the height of the tilted tank 92 is different from the height of the normal tank 91, the detection of the tilted tank 92 is relatively easy. However, false detections are prone to occur, for example, in the detection of an inverted tank 93 that has water inside. In contrast, according to this embodiment, inverted tanks 93 and the like that have water inside can also be detected with high accuracy.

[0049] Furthermore, in detection using the sensors described above, for example, when switching to the manufacture of tanks 90 with different heights, the sensors need to be adjusted whenever the conditions of the tank 90 change. In particular, such adjustments are required for all sensors that need to be installed as described above. In contrast, according to this embodiment, it is easier to manufacture a model capable of detecting anomalies in tanks 90 with different heights, and even if the tank 90 to be manufactured is changed, it can be handled simply by changing the model imported into the detection device 56.

[0050] In the conveying system 1 of this embodiment, a rejection device 30 is provided that cooperates with the detection system 50. Therefore, cans 90 that are the cause of defects can be rejected without manual intervention, resulting in high efficiency. In particular, the rejection device 30 can reject cans 90 in each section along the width direction, thus rejecting a necessary and sufficient number of cans 90, thereby suppressing unnecessary rejection and achieving a high yield. Moreover, the range of cans 90 to be rejected can be adjusted according to the posture of cans 90 in abnormal positions, etc., so that cans 90 can be rejected more appropriately.

[0051] Furthermore, in the detection device 56 of this embodiment, a learned model is used to detect abnormal containers, thus making it easier to achieve high-precision detection related to a variety of states, which is difficult in conventional rule-based detection.

[0052] The preferred embodiments have been shown above to illustrate the present invention. However, the present invention is not limited to the embodiments described above, and various modifications can be made within the scope of the present invention.

[0053] For example, even if the transfer device 31 is configured to stop transferring across the entire width direction without partitioning, although the yield rate will deteriorate, it can selectively reject cans 90, including those lying down or upside down 92 or 93, thus rejecting cans 90 that may be contaminated. Furthermore, even if the operation of the rejection device 30 is not controlled by the output from the detection device 56, cans 92, etc., will still be rejected by the rejection device 30. In this case, it is also possible to remove cans 90, etc., around the detected lying down 92 manually, based on the notification from the detection device 56. Furthermore, the rejection device is not limited to a structure that causes the cans 90 to be rejected to fall through the gap 32; any rejection device from various designs can be used. Furthermore, the detection device 56 is not limited to using a learning completion model; it can operate through various programs capable of detecting cans 90 that have not adopted the prescribed posture. Furthermore, this conveying system 1 can be introduced to any location in the can-making process. In processes not limited to the cleaning machine 20, but including processes involving the spraying of gases or liquids, the posture of the can 90 can easily change; therefore, the imaging device 52 can also be located downstream of this process. Furthermore, the imaging device 52 can be located anywhere without being restricted by such processes. For example, this technology can also be used in processes where the cans 90 are regularly arranged on the conveyor surface 11. Moreover, this conveying system 1 is not limited to the manufacture of cans; for example, it can also be applied to the manufacture of resin containers such as plastic bottles, and other containers.

[0054] The entire contents of the documents described in this specification and the contents of the Japanese application specification, which forms the basis of the Paris Convention priority claim of this application, are hereby incorporated.

Claims

1. A conveying system, the conveying system comprising: The conveying device is configured to convey the containers along the conveying direction, wherein a plurality of bottomed cylindrical containers are arranged in the conveying direction and in the width direction orthogonal to the conveying direction; The imaging device acquires an image of the container being conveyed by the conveying device; The detection device uses the image to detect a container that is not in a prescribed posture among a plurality of containers being transported, and determines the coordinates of the container in the width direction; The rejection device is configured to selectively reject the container from the conveying device according to each partition in the width direction; and The control device controls the operation of the rejection device. The detection device is equipped with a machine learning model that determines the posture of the container, wherein the prescribed posture is an inverted posture with the bottom facing upwards. The model is generated through supervised learning using images of the container in the inverted posture, the container in an upright posture with the bottom facing down, and the container in a lying posture as teaching data. The model is configured to distinguish at least the inverted, upright, and lying postures of the container. The control device, based on the coordinates determined by the detection device, causes the rejection device to reject at least containers in upright and lying positions, as well as containers surrounding the upright and lying positions.

2. The conveying system according to claim 1, wherein, The conveying device is configured to convey the containers along the conveying direction when the containers are irregularly arranged in the conveying direction and the width direction.

3. The conveying system according to claim 1 or 2, wherein, The conveying device has the form of a platform that extends continuously along the conveying direction, and is configured to hold the container and convey the container along the conveying direction. The rejection device is configured such that, at the gap of the platform provided in the conveying device, it attracts the container that is not to be rejected from above and moves it along the conveying direction in such a way that it crosses the gap, so that the container to be rejected falls into the gap.

4. The conveying system according to claim 1 or 2, wherein, The conveying system also includes a spraying device, which is located on the conveying device and sprays fluid into the container being conveyed. The shooting device is located downstream of the spraying device.

Citation Information

Patent Citations

  • Transport device

    JP2019073374A

  • Can washing apparatus and process for washing can

    JP2010036089A

  • Waste sorting device

    JP2020199444A

  • Method and apparatus for detecting faults during object transport

    US20200189855A1