JP7206451B2 - Object sorting system, object sorting program, information processing device, object sorting method and sorting device - Google Patents

Description

The present invention relates to an object sorting system, an object sorting program, an information processing device, an object sorting method, and a sorting device.

In the sorting of wastes carried out in waste treatment plants, etc., first, various physical and chemical principles are followed according to the differences in the weight, magnetism, volume, shape, optical properties, hardness and buoyancy of the waste. A sorting process is performed, followed by further sorting, which is manual dependent. In this case, for example, an operator stands by a belt conveyor (hereinafter also referred to as a conveying device), visually confirms the waste conveyed by the belt conveyor, and manually sorts each type of waste. (Patent Documents 1 to 8).

Japanese Utility Model Laid-Open No. 5-005632 Japanese Utility Model Laid-Open No. 5-080571 JP-A-6-106091 JP-A-7-256231 JP-A-9-262556 JP-A-2000-005626 JP 2008-068197 A JP 2011-050959 A

However, in the above-mentioned waste treatment plants, etc., there are cases where it is not possible to secure the required number of workers due to factors such as the working environment in which the waste sorting work is carried out. Therefore, in a waste treatment plant or the like, for example, the operating rate of waste sorting work cannot be increased, and a sufficient amount of waste cannot be sorted in some cases. Moreover, the waste sorting accuracy depends on the skill and judgment of the operator. Therefore, it may not be possible to ensure stable sorting accuracy in the above-described waste treatment plant or the like.

Accordingly, an object of the present invention is to provide an object sorting system, an object sorting program, an information processing device, an object sorting method, and a sorting device that can improve the sorting accuracy and operating rate of waste sorting work. That's what it is.

An object sorting system according to the present invention for achieving the above object is an object sorting system for sorting objects conveyed by a conveying device by type, comprising: an imaging device for imaging the object; An information processing device that determines the type of an object included in image data captured by an apparatus, and a selection device that selects the type of an object included in the image data captured by the imaging device according to the determination result of the information processing device. and wherein the information processing device uses a learning model learned by associating image data including a learning target with a type of the learning target, thereby capturing an image captured by the imaging device It is characterized by determining the type of object included in the image data.

Further, in one aspect of the object sorting system of the present invention for achieving the above object, the sorting device divides the objects imaged by the imaging device into one or more images corresponding to each type of object. It is characterized by moving to a region corresponding to the type determined by the information processing device among the regions.

Further, in one aspect of the object sorting system of the present invention for achieving the above object, the sorting device has one or more sorting arms, and each of the one or more sorting arms has the transport device and a shaft portion provided at each base end portion is pivotally supported on the transport route, and each of the one or more sorting arms rotates the tip portion around the shaft portion. By moving, the object conveyed by the conveying device is brought into contact with the tip portion and guided to the region corresponding to the type determined by the information processing device among the one or more regions. and

In one aspect of the object sorting system of the present invention for achieving the above object, the one or more sorting arms are spaced apart from the upstream side toward the downstream side in the transport direction of the transport device. It is characterized by constituting a group of sorting arms arranged in a plurality of stages.

Further, in one aspect of the object sorting system of the present invention for achieving the above object, the sorting device is arranged such that the positions of the objects conveyed by the conveying device in the width direction of the conveying route are equal to the conveying route. and a guide wall that guides the objects conveyed by the conveying device so as to be within the range of rotation of the sorting arm in the width direction.

Further, in one aspect of the object selection system of the present invention for achieving the above object, the information processing device uses the learning model to select a plurality of types of objects captured by the imaging device. It is characterized by calculating probabilities of being each of the type candidates, and specifying the type candidate corresponding to the maximum probability among the calculated probabilities as the type of the object imaged by the imaging device.

Further, in one aspect of the object sorting system of the present invention for achieving the above object, at least a part of the transport route is a circular route, and the sorting device moves over a predetermined position on the transport route. It is characterized by moving the detected object to a region other than the one or more regions when detecting the object that has been conveyed more than a predetermined number of times.

Further, in one aspect of the object sorting system of the present invention for achieving the above object, the sorting device has one or more compressed air generators, and each of the one or more compressed air generators , arranged on a conveying path of the conveying device, each of the one or more compressed air generators being conveyed by the conveying device by delivering compressed air to an object conveyed by the conveying device; It is characterized by guiding the target object to the area corresponding to the type determined by the information processing device among the one or more areas.

In one aspect of the object sorting system of the present invention for achieving the above object, the determination result includes the probability that the type of the object imaged by the imaging device is each of a plurality of type candidates. , the information processing device uses the learning model to identify the maximum probability among the probabilities included in the determination result, and the object imaged by the imaging device is a type candidate corresponding to the maximum probability. and the sorting device identifies the object imaged by the imaging device as one or more corresponding to a combination of one of the plurality of candidate types and one of a plurality of ranges of probability. , to a region corresponding to the combination of the type of the object identified by the information processing device and the maximum probability.

Further, in one aspect of the object selection system of the present invention for achieving the above object, the information processing device classifies image data including the object for learning and the type of the object for learning. The learning model is generated by learning the learning data included.

Further, in one aspect of the object sorting system of the present invention for achieving the above object, the conveying device has an average value of the density of the objects on the conveying path of the conveying device that is greater than a first threshold value. case, the conveying speed of the conveying device is reduced, and when the average value of the densities is smaller than a second threshold value, the conveying speed is increased.

Further, an object selection program according to the present invention for achieving the above object is a learning model generated by learning image data including an object for learning and learning data including the type of the object for learning. is used to determine the type of object included in the image data captured by the image capturing device, and the image capturing device for the sorting device that selects the type of target included in the image data captured by the image capturing device. is characterized by causing the computer to execute a process of transmitting the determination result of the type of object included in the image data captured by the.

Further, an information processing apparatus according to the present invention for achieving the above object generates a learning model generated by learning image data including a learning target and learning data including the type of the learning target. By using the and a transmission unit that transmits a determination result of a type of an object included in image data captured by the imaging device.

Further, in the object selection method of the present invention for achieving the above object, there is provided a learning model generated by learning image data containing a learning object and learning data containing the type of the learning object. is used to determine the type of object included in the image data captured by the image capturing device, and the image capturing device for the sorting device that selects the type of target included in the image data captured by the image capturing device. is characterized by causing the computer to execute a process of transmitting the determination result of the type of object included in the image data captured by the.

Further, an object sorting method according to the present invention for achieving the above object is a method for sorting objects conveyed by a conveying device by type, wherein an imaging device picks up an image of the object. an information processing device that determines a type of an object included in image data captured by the imaging device; and a sorting device that selects a target included in the image data captured by the imaging device according to the determination result of the information processing device. The type of object is selected, and the information processing device uses a learning model learned by associating image data including a learning target with the type of the learning target, so that the imaging device captures an image. and determining the type of object included in the obtained image data.

Further, the sorting apparatus of the present invention for achieving the above object comprises a conveying apparatus for conveying an object on a conveying path; The object conveyed by the conveying device is pivotally supported on the path and the tip portion is rotated about the shaft portion so that the object conveyed by the conveying device is brought into contact with the tip portion. One or more sorting arms for guiding to a region corresponding to the type of the object to be transported by the transport device, among the one or more regions corresponding to each type of object, and the object to be transported by the transport device a guide wall that guides the objects conveyed by the conveying device so that the position in the width direction of the conveying path is within the rotation range of the sorting arm in the width direction of the conveying path. Characterized by

Further, in one aspect of the sorting device of the present invention for achieving the above object, the one or more sorting arms are arranged at intervals from the upstream side to the downstream side in the conveying direction of the conveying device. It is characterized by constituting a multi-stage sorting arm group.

Further, in one aspect of the sorting device of the present invention for achieving the above object, each of the one or more selection arms selects an object to be transported by the transporting device. guidance to a region corresponding to a type determined by an information processing device that determines the type of an object included in image data captured by an imaging device, from among one or more regions corresponding to each type of object, A processing device uses a learning model learned by associating image data including a learning target with a type of the learning target, so that the object included in the image data captured by the imaging device It is characterized by determining the type.

In one aspect of the sorting apparatus of the present invention for achieving the above object, at least a part of the conveying route is a circular route.

Further, the sorting apparatus of the present invention for achieving the above object comprises a conveying device for conveying an object on a conveying route, and a conveying device arranged on the conveying route for compressing the object conveyed by the conveying device. By sending air, the object to be conveyed by the conveying device is moved to the object to be conveyed by the conveying device in one or more areas corresponding to each type of the object to be conveyed by the conveying device. one or more compressed air generators guiding to areas corresponding to the types of

Further, the sorting apparatus of the present invention for achieving the above object comprises a conveying apparatus for conveying an object on the conveying path, a conveying apparatus arranged on the conveying path, a shaft portion pivotally supported on the conveying path, and the One or more partition walls extending radially outward so as to be rotatable about the shaft, each of the one or more partition walls having a radially outer end between adjacent partition walls in the circumferential direction. has a sorter that forms one or more open areas, and the sorter rotates the one or more partition walls to cause the one or more areas to be transported by the transport device regions corresponding to each of the types are opposed to the upstream side of the conveying direction of the conveying device, and the object conveyed by the conveying device is guided to the region corresponding to the type of the object conveyed by the conveying device characterized in that

According to the object sorting system, the object sorting program, the information processing device, the object sorting method, and the sorting device according to the present invention, it is possible to improve the sorting accuracy and operating rate of the waste sorting work.

FIG. 1 is a diagram illustrating the configuration of an object sorting system 10. As shown in FIG. FIG. 2 is a diagram for explaining the configuration of the sorting device 2. As shown in FIG. FIG. 3 is a diagram illustrating another configuration of the sorting device 2. As shown in FIG. FIG. 4 is a diagram for explaining the configuration of the information processing device 1. As shown in FIG. FIG. 5 is a diagram for explaining an outline of the sorting process executed by the information processing device 1. As shown in FIG. FIG. 6 is a diagram illustrating details of the sorting process executed by the information processing apparatus 1. As shown in FIG. FIG. 7 is a diagram for explaining the details of the sorting process executed by the information processing device 1. As shown in FIG. FIG. 8 is a diagram for explaining the details of the sorting process executed by the information processing device 1. As shown in FIG. FIG. 9 is a diagram illustrating a specific example of the configuration when using a circular route. FIG. 10 is a diagram for explaining a specific example of the configuration when the sorting board 41 is used. FIG. 11 is a diagram for explaining a specific example of the configuration when the sorting board 41 is used. FIG. 12 is a diagram illustrating a specific example of the configuration when the sorting board 41 is used. FIG. 13 is a diagram illustrating a specific example of the configuration when the compressed air generator 25 is used. FIG. 14 is a diagram illustrating a specific example of the configuration when the compressed air generator 25 is used.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described with reference to the drawings. However, such an embodiment does not limit the technical scope of the present invention.

[Configuration of object sorting system]
FIG. 1 is a diagram illustrating the configuration of an object sorting system 10. As shown in FIG. An object sorting system 10 shown in FIG. 1 has an information processing device 1 , a sorting device 2 , and an operation terminal 5 .

The information processing device 1 is a computer device, for example, a general-purpose PC (Personal Computer). Then, the information processing device 1 determines the type of waste to be conveyed on the belt conveyor included in the sorting device 2, for example, by using a learning model for determining the type of waste. Note that the information processing apparatus 1 may be of a stationary type, a notebook type, a tablet type, or the like.

The sorting device 2 sorts the wastes conveyed on the belt conveyor according to the content of the operation via the operation terminal 5 by the operator OP. Further, the sorting device 2 sorts the wastes conveyed on the belt conveyor by type based on the determination result by the information processing device 1 .

[Configuration of sorting device (1)]
Next, the configuration of the sorting device 2 will be described. FIG. 2 is a diagram for explaining the configuration of the sorting device 2. As shown in FIG.

The sorting device 2 shown in FIG. 2 has a belt conveyor 21 . The sorting device 2 shown in FIG. 2 also has alignment mechanisms 22a and 22b. The sorting device 2 shown in FIG. 2 also has a sorting arm 23a. Further, cameras 24a, 24c and 24e for taking images on the belt conveyor 21 are arranged near the sorting device 2 shown in FIG.

The belt conveyor 21 conveys each waste OB in the conveying direction (arrow direction in FIG. 2). In the following description, it is assumed that two types of waste OB (waste OB1 and OB3) are conveyed on the belt conveyor 21. FIG.

The alignment mechanisms 22a and 22b guide the waste OB transported from the upstream side in the transport direction (the right side in FIG. 2) to the area sandwiched between the alignment mechanisms 22a and 22b in the width direction of the belt conveyor 21. .

The sorting arm 23a transports from the upstream side in the transport direction (right side in FIG. 2) based on the content of the operation by the operator OP via the operation terminal 5 and the determination result of the type of waste OB by the information processing device 1. Each of the collected waste OB is sorted by type. Specifically, the sorting arm 23a has a shaft portion 23a1 provided at the base end portion (left end portion in FIG. 2) which is pivotally supported on the belt conveyor 21. 2), each waste OB on the belt conveyor 21 is brought into contact with the leading end and guided to an area corresponding to the type of each waste OB.

In addition, the sorting arm 23a is supported by a support mechanism (not shown) provided above the belt conveyor 21 (perpendicular to the conveying surface of the belt conveyor 21). may be arranged so as not to come into direct contact with the

More specifically, when the waste OB being conveyed on the belt conveyor 21 is the waste OB1, the sorting arm 23a shown in FIG. area), and if the waste OB being conveyed on the belt conveyor 21 is the waste OB3, the area corresponding to the waste OB3 (the area below the sorting arm 23a in FIG. 2) ) to move the waste OB.

The sorting arm 23a shown in FIG. 2 is positioned near the center between the alignment mechanism 22a and the alignment mechanism 22b in the width direction of the belt conveyor 21. As shown in FIG. That is, in the example shown in FIG. 2, the position of the sorting arm 23a in the width direction of the belt conveyor 21 is included in the range where each waste OB is guided by the alignment mechanisms 22a and 22b. This enables the sorting arm 23a to sort each waste OB into two types.

The aligning mechanism 22a and the aligning mechanism 22b can separate the waste OBs by guiding the waste OBs. This enables the sorting arm 23a to sort each waste OB with high accuracy.

After that, each of the wastes OB1 and OB3 is stored in each of stockyards SY1 and SY3 (hereinafter collectively referred to as stockyard SY).

Further, in the example shown in FIG. 2, the camera 24a photographs the vicinity of the alignment mechanisms 22a, 22b and the sorting arm 23a. In addition, the cameras 24c and 24e respectively photograph areas corresponding to the stockyards SY1 and SY3, respectively.

In addition, each of the cameras 24a, 24c and 24e may be, for example, a visible light camera or an infrared camera. It may be used in conjunction with an ultrasonic generator (not shown) that generates ultrasonic waves for recognition.

Also, in the example shown in FIG. 2, the case where the alignment mechanisms 22a and 22b are provided on the belt conveyor 21 has been described, but the present invention is not limited to this. Specifically, for example, only one of the alignment mechanisms 22 a and 22 b may be provided on the belt conveyor 21 .

Further, in the example shown in FIG. 2, when the alignment mechanisms 22a and 22 are provided so that their positions in the conveying direction match (the alignment mechanisms 22a and 22b are arranged in parallel with respect to the width direction of the sorting device 2). provided) has been described, but the present invention is not limited to this. Specifically, the alignment mechanisms 22a and 22 may have different positions in the transport direction.

Also, the angles of the alignment mechanisms 22a and 22b with respect to the conveying direction may be changed by, for example, a motor (not shown).

[Configuration of sorting device (2)]
Next, another configuration of the sorting device 2 will be described. FIG. 3 is a diagram illustrating another configuration of the sorting device 2. As shown in FIG.

The sorting device 2 shown in FIG. 3 has a belt conveyor 21 . The sorting device 2 shown in FIG. 3 also has alignment mechanisms 22a, 22b, 22c, 22d, 22e and 22f (hereinafter collectively referred to as alignment mechanism 22 or guide wall 22). The sorting device 2 shown in FIG. 3 also has sorting arms 23a, 23b and 23c (hereinafter collectively referred to as sorting arm 23). Further, cameras 24a, 24b, 24c, 24d, 24e, and 24f (hereinafter collectively referred to as cameras 24) are arranged near the sorting device 2 shown in FIG. ing. That is, the sorting device 2 shown in FIG. 3 has a configuration including a first stage sorting arm group consisting of the sorting arm 23a and a second stage sorting arm group consisting of the sorting arms 23b and 23c.

The belt conveyor 21 conveys each waste OB in the conveying direction (arrow direction in FIG. 3). In the following description, it is assumed that four types of waste OB (waste OB1, OB2, OB3 and OB4) are conveyed on the belt conveyor 21. FIG.

The alignment mechanisms 22a and 22b guide the waste OB transported from the upstream side in the transport direction (the right side in FIG. 3) to the area sandwiched between the alignment mechanisms 22a and 22b in the width direction of the belt conveyor 21. .

The sorting arm 23a transports from the upstream side in the transport direction (right side in FIG. 3) based on the content of the operation by the operator OP via the operation terminal 5 and the determination result of the type of waste OB by the information processing device 1. Each of the collected waste OB is sorted by type. Specifically, the sorting arm 23a has a shaft portion 23a1 provided at the base end portion (the left end portion in FIG. 3) that is pivotally supported on the belt conveyor 21, and the tip portion (the 3), each waste OB on the belt conveyor 21 is brought into contact with the leading end and guided to an area corresponding to the type of each waste OB.

In addition, the sorting arm 23a is supported by a support mechanism (not shown) provided above the belt conveyor 21 (perpendicular to the conveying surface of the belt conveyor 21). may be arranged so as not to come into direct contact with the

More specifically, when the waste OB being conveyed on the belt conveyor 21 is the waste OB1 or OB2, the sorting arm 23a shown in FIG. When the waste OB being conveyed on the belt conveyor 21 is the waste OB3 or OB4, the area corresponding to the waste OB3 and OB4 (the area above the arm 23a in FIG. the area below the sorting arm 23a).

The sorting arm 23a shown in FIG. 3 is positioned near the center between the alignment mechanism 22a and the alignment mechanism 22b in the width direction of the belt conveyor 21. As shown in FIG. That is, in the example shown in FIG. 3, the position of the sorting arm 23a in the width direction of the belt conveyor 21 is included in the range where each waste OB is guided by the alignment mechanisms 22a and 22b. This enables the sorting arm 23a to sort each waste OB into two types.

The alignment mechanisms 22c and 22d align the waste OB transported from the upstream side in the transport direction (the waste OB sorted by the sorting arm 23a) with the alignment mechanisms 22c and 22d in the width direction of the belt conveyor 21. guides to the area between

Similar to the sorting arm 23a, the sorting arm 23b sorts each of the waste OBs transported from the upstream side in the transport direction into two types. Specifically, when the waste OB being conveyed on the belt conveyor 21 is the waste OB1, the sorting arm 23b shown in FIG. area), and when the waste OB being conveyed on the belt conveyor 21 is the waste OB2, the area corresponding to the waste OB2 (the area below the sorting arm 23b in FIG. 3) to move the waste OB to

The sorting arm 23b shown in FIG. 3 is positioned near the center between the alignment mechanism 22c and the alignment mechanism 22d in the width direction of the belt conveyor 21. As shown in FIG. That is, in the example shown in FIG. 3, the position of the sorting arm 23b in the width direction of the belt conveyor 21 is included in the range where each waste OB is guided by the alignment mechanisms 22c and 22d. As a result, the sorting arm 23b can further sort each waste OB conveyed from the upstream side into two types.

The alignment mechanisms 22e and 22f align the waste OB transported from the upstream side in the transport direction (the waste OB sorted by the sorting arm 23a) with the alignment mechanisms 22e and 22f in the width direction of the belt conveyor 21. guides to the area between

Similar to the sorting arm 23a, the sorting arm 23c sorts each of the wastes OB transported from the upstream side in the transport direction into two types. Specifically, when the waste OB being conveyed on the belt conveyor 21 is the waste OB3, the sorting arm 23c shown in FIG. area), and when the waste OB being conveyed on the belt conveyor 21 is the waste OB4, the area corresponding to the waste OB4 (the area below the sorting arm 23c in FIG. 3) to move the waste OB to

The sorting arm 23c shown in FIG. 3 is positioned near the center between the alignment mechanism 22e and the alignment mechanism 22f in the width direction of the belt conveyor 21. As shown in FIG. That is, in the example shown in FIG. 3, the position of the sorting arm 23c in the width direction of the belt conveyor 21 is included in the range where each waste OB is guided by the alignment mechanisms 22e and 22f. This allows the sorting arm 23c to further sort each waste OB transported from the upstream side into two types.

After that, each of the wastes OB1, OB2, OB3 and OB4 is stored in each of the stockyards SY1, SY2, SY3 and SY4 (hereinafter also collectively referred to as the stockyard SY).

Furthermore, in the example shown in FIG. 3, the camera 24a photographs the vicinity of the alignment mechanisms 22a and 22b and the selection arm 23a, and the camera 24b captures the vicinity of the alignment mechanisms 22c and 22d and the selection arm 23b, the alignment mechanism 22d, and the alignment mechanism 22d. 22e and the vicinity of the sorting arm 23c are photographed. In addition, the cameras 24c, 24d, 24e and 24f respectively photograph areas corresponding to the stockyards SY1, SY2, SY3 and SY4.

The sorting device 2 is arranged so that the waste OB2 guided by the sorting arm 23b is not stored in the stock yard SY3 and the waste OB3 guided by the sorting arm 23c is not stored in the stock yard SY2. A shielding wall (not shown) may be provided at the center position of the sorting arms 23b and 23c in the width direction of the conveyor 21 .

Also, in the example shown in FIG. 3, the case where the sorting device 2 includes three sorting arms 23 (that is, a two-stage group of sorting arms) has been described, but the present invention is not limited to this. Specifically, the sorting device 2 may include, for example, three or more sorting arms 23 (that is, three or more stages of sorting arm groups), and sort four or more types of waste OB.

In the example shown in FIG. 3, the number of the sorting arms 23 (sorting arms 23b and 23c) included in the sorting arm group that is second closest from the upstream side in the transport direction is the sorting arm group that is closest to the upstream side in the transport direction. A case has been described where the number is double the number of sorting arms 23 (sorting arms 23a) included in (the group of sorting arms adjacent to the upstream side in the transport direction), but it is not limited to this. Specifically, each of the multiple stages of sorting arm groups may have one or more sorting arms.

In addition, in the example shown in FIG. 3, the case where the sorting arm 23 sorts the waste OB by type has been described. The sorting may be performed according to whether or not the waste is OB or according to the size of the waste OB. As a result, the sorting device 2 can accurately sort each type of waste OB even when it is necessary to sort many types of waste OB that differ greatly in size, properties, etc. be able to do well.

[Configuration of information processing device]
Next, the configuration (hardware configuration) of the information processing apparatus 1 will be described. FIG. 4 is a diagram for explaining the configuration of the information processing device 1. As shown in FIG.

The information processing device 1 has a hardware configuration of a general-purpose computer device, and includes, for example, a CPU 101 as a processor, a memory 102, a network interface 103, and a storage medium 104, as shown in FIG. Each unit is connected to each other via a bus 105 .

The storage medium 104 has, for example, a program storage area (not shown) for storing a program (not shown) for sorting waste OB by type (hereinafter also referred to as sorting process). The storage medium 104 also has a storage area 110 that stores information used when performing the sorting process, for example. Note that the storage medium 104 may be, for example, an HDD (Hard Disk Drive) or an SSD (Solid State Drive).

The CPU 101 executes a program loaded from the storage medium 104 (storage area 110) to the memory 102 to perform the sorting process.

Also, the network interface 103 enables remote control of the sorting device 2 via a network, for example, by communicating with the sorting device 2 .

[Overview of sorting process]
Next, an outline of the sorting process executed by the information processing device 1 will be described. FIG. 5 is a diagram for explaining an outline of the sorting process executed by the information processing device 1. As shown in FIG.

First, the learning model generation unit 12 of the information processing device 1 generates a learning model by learning the learning data generated by the learning data generation unit 11 of the information processing device 1 . Specifically, the learning model generation unit 12 generates a learning model by, for example, learning image data including learning waste OB and learning data including learning waste OB types. . That is, the learning model generation unit 12 generates a learning model by learning learning data generated based on the result of sorting the waste OB performed by the operator OP via the operation terminal 5, for example. . Note that, for example, when the camera 24 is an infrared camera, the learning data generation unit 11 may generate, as learning data, infrared data acquired by the infrared camera for the waste OB for learning.

Then, when the type determination unit 13 of the information processing device 1 acquires the image data of the waste OB conveyed on the belt conveyor 21 of the sorting device 2 (hereinafter also referred to as a new waste OB), the acquired image By inputting the data into the learning model, determination results for new types of waste OB are obtained.

After that, the sorting arm control unit 14 of the information processing device 1 operates the sorting arm 23 according to the determination result acquired by the type determination unit 13 . Specifically, the sorting arm control unit 14 instructs a control device (not shown) that controls the sorting arm 23, for example, so that the sorting arm 23 performs an operation corresponding to the content of the determination result.

Further, the transport control unit 15 of the information processing device 1 controls the transport speed of the belt conveyor 21 according to the density of the waste OB on the belt conveyor 21 .

That is, the object sorting system 10 according to the present embodiment sorts the waste OBs conveyed on the belt conveyor 21 by using a learning model obtained by learning the image data of the waste OBs and the types of wastes. do it automatically.

As a result, the object sorting system 10 can efficiently sort the waste OB, and can reduce the number of workers required for sorting the waste OB.

[Details of sorting process]
Next, details of the sorting process executed by the information processing apparatus 1 will be described. 6 to 8 are diagrams for explaining the details of the sorting process executed by the information processing apparatus 1. FIG.

[Learning process]
First, among the sorting processes, a process for generating a learning model (hereinafter also referred to as a learning process) will be described.

As shown in FIG. 6, the learning data generation unit 11 acquires image data included in moving image data of, for example, each area corresponding to each stockyard SY on the belt conveyor 21 (S11).

Specifically, as shown in FIG. 2, the operator OP sorts out the wastes OB conveyed on the belt conveyor 21 by, for example, operating the sorting arms 23 via the operation terminal 5. , wastes OB1, OB2, OB3 and OB4 are stored in stockyards SY1, SY2, SY3 and SY4, respectively. Each of the cameras 24c, 24d, 24e, and 24f, for example, while the sorting work is being performed by the operator OP, captures areas corresponding to the stockyard SY1, the stockyard SY2, the stockyard SY3, and the stockyard SY4, respectively. Take a picture of

Therefore, for example, the learning data generation unit 11 acquires image data included in the moving image data for each area captured by each camera 24 in parallel with capturing the area corresponding to each stockyard SY by each camera 24. do.

Then, the learning data generation unit 11 stores the waste OB included in the image data acquired in the process of S11 and the area corresponding to each stockyard SY on the belt conveyor 21 for each area corresponding to each stockyard SY. Learning data including the type of waste OB that is contained is generated (S12).

Specifically, the learning data generation unit 11 indicates, for example, image data including each waste OB and that the type of each waste OB is waste OB1 for each waste OB stored in the stockyard SY1. Generate training data containing information (labels). Similarly, the learning data generator 11 also generates learning data for each of the waste OBs stored in the stockyards SY2, SY3, and SY4.

As a result, the learning data generation unit 11 can automatically generate the number of learning data corresponding to the period during which the waste OB is sorted by the operation of the operator OP.

Note that the camera 24 may use a plurality of types of cameras (including an infrared camera other than a visible light camera) to capture image data of the same waste OB from a plurality of directions. In this case, the learning data generator 11 may generate a plurality of learning data for the same waste OB.

As a result, the learning data generator 11 can generate a learning model with higher determination accuracy for the type of waste OB.

After that, the learning model generation unit 12 generates a learning model by learning the learning data generated in the process of S12 (S13).

Note that the learning data generation unit 11 may generate new learning data as needed. Then, the learning model generation unit 12 may update the learning model generated in the process of S13 at any time by using new learning data.

[Main process of sorting process]
Next, the main processing of sorting processing will be described.

As shown in FIG. 7, the type determination unit 13 acquires image data included in moving image data obtained by photographing the vicinity of the sorting arm 23 (S21).

Specifically, the type determination unit 13, for example, collects moving image data near the sorting arm 23a photographed by the camera 24a described in FIG. Get the image data contained in . In addition, the type determination unit 13 may, for example, use video data of the vicinity of the sorting arms 23b and 23c photographed by the camera 24b described in FIG. data).

Then, the type determination unit 13 inputs the image data acquired in the process of S21 to the learning model generated in the process of S13, thereby determining the type of the object included in the image data acquired in S21. A result is acquired (S22).

Specifically, the type determination unit 13 receives, for example, image data included in moving image data of the vicinity of the sorting arm 23a photographed by the camera 24a, so that the waste OB being transported near the sorting arm 23a is detected. The determination result of the type of (waste OB to be sorted by the sorting arm 23a from now on) is obtained. Further, the type determination unit 13 receives, for example, image data included in moving image data of the vicinity of the sorting arms 23b and 23c photographed by the camera 24b, so that the waste conveyed near the sorting arms 23b or 23c is detected. A determination result is obtained for the type of object OB (waste OB to be sorted by the sorting arm 23b or 23c).

After that, the sorting arm control unit 14 operates the sorting arm 23 according to the determination result obtained in the process of S22 (S23).

Specifically, for example, in the example described with reference to FIG. The controller is instructed to turn the tip of 23a leftward (upward in FIG. 2). On the other hand, for example, when it is determined that the type of waste OB being transported in the vicinity of the sorting arm 23a is the waste OB4, the sorting arm control unit 14 moves the tip of the sorting arm 23a to the right (see FIG. 2). The controller is instructed to rotate downward).

Similarly, for example, when it is determined that the type of waste OB being transported in the vicinity of the sorting arm 23b is the waste OB1, the sorting arm control unit 14 sets the tip of the sorting arm 23b to the left (see FIG. 2). The controller is instructed to rotate in the upward direction). On the other hand, for example, when it is determined that the type of waste OB being transported in the vicinity of the sorting arm 23b is the waste OB2, the sorting arm control unit 14 sets the tip of the sorting arm 23b to the right (see FIG. 2). The controller is instructed to rotate downward).

As a result, the information processing apparatus 1 can sort the waste OB conveyed on the belt conveyor 21 by type by using the learning model.

In the example shown in FIG. 2, the alignment mechanisms 22a and 22b may each have a door portion (not shown) that can be opened and closed, for example. Specifically, the alignment mechanism 22a may have, for example, a door portion (hereinafter also referred to as a first door portion) whose position in the width direction of the belt conveyor 21 is the same as that of the sorting arm 23b. Also, the alignment mechanism 22b may have, for example, a door portion (hereinafter also referred to as a second door portion) whose position in the width direction of the belt conveyor 21 is the same as that of the sorting arm 23c.

Then, the information processing apparatus 1 determines the type of waste OB near the aligning mechanisms 22a and 22b photographed by a camera (not shown) arranged on the upstream side of the camera 24a, for example, based on the type determination result. is determined to be waste OB1 or waste OB2, the first door is opened at a necessary timing so that the waste OB determined to be waste OB1 or waste OB2 passes through the first door, and the type is waste OB3 or waste OB4. The second door may be opened at the necessary timing so that the waste OB determined to be present passes through the second door.

As a result, the information processing device 1 can further improve the sorting accuracy of each waste OB in the sorting device 2, for example.

[Conveyance control processing]
Next, among the sorting processes, the process of controlling the transport speed of the belt conveyor 21 (hereinafter also referred to as transport control process) will be described.

As shown in FIG. 8, the transport control unit 15 acquires image data included in moving image data of the vicinity of the sorting arm 23 (S31).

Specifically, the transport control unit 15 acquires, for example, image data at a predetermined timing included in moving image data of the vicinity of the sorting arm 23a photographed by the cameras 24a and 24b described with reference to FIG.

Then, the transport control unit 15, for example, calculates the density of the waste OB in the image data acquired in the process of S31 (S32). Further, the transport control unit 15 calculates the average value of the densities within a certain period of time (for example, 10 seconds) calculated in the process of S32 (S33).

Subsequently, the transport control unit 15 compares the average value calculated in the process of S33 with an upper limit threshold value (hereinafter also referred to as a first threshold value). Further, the transport control unit 15 compares the average value calculated in the process of S33 with a lower limit threshold (hereinafter also referred to as a second threshold) (S34).

Specifically, the transport control unit 15 determines whether or not the average value calculated in the process of S33 is greater than the first threshold. Further, the transport control unit 15 determines whether or not the average value calculated in the process of S33 is smaller than the second threshold.

After that, the transport control unit 15 controls the transport speed of the waste OB on the belt conveyor 21 based on the comparison result in the process of S34.

Specifically, when the transport control unit 15 determines that the average value calculated in the process of S33 is larger than the first threshold value, the transport control unit 15 reduces the transport speed of the waste OB on the belt conveyor 21 . That is, in this case, the transport control unit 15 uses the learning model to determine that there is a large amount of waste OB whose type needs to be determined, and reduces the transport speed of the waste OB on the belt conveyor 21. .

As a result, the transport control unit 15 can determine the type of each waste OB, even when the amount of waste OB requiring determination of the type is large, for example.

Further, when determining that the average value calculated in the process of S33 is smaller than the second threshold value, the transport control unit 15 increases the transport speed of the waste OB on the belt conveyor 21 . That is, in this case, the transport control unit 15 uses the learning model to determine that the amount of waste OB whose type needs to be determined is small, and increases the transport speed of the waste OB on the belt conveyor 21. .

As a result, the transport control unit 15 can shorten the time required to determine the type of each waste OB, for example, when the amount of waste OB whose type needs to be determined is small.

[Concrete example of configuration when circular path is used]
Next, a specific example of a configuration using a circular route will be described. FIG. 9 is a diagram illustrating a specific example of the configuration when using a circular route.

The sorting device 2 shown in FIG. 9 has a belt conveyor 31 containing circular paths at two locations. The sorting device 2 shown in FIG. 9 also has alignment mechanisms 32a, 32b, 32c, 32d, 32e, and 32f (hereinafter collectively referred to as alignment mechanisms 32). The sorting device 2 shown in FIG. 9 also has sorting arms 33a, 33b, and 33c (hereinafter collectively referred to as sorting arms 33). In addition, in the example shown in FIG. 9, the description of the camera for imaging the belt conveyor 31 is omitted.

The belt conveyor 31 conveys each waste OB in the conveying direction (arrow direction in FIG. 9).

The sorting arm 33a sorts, by type, each of the wastes OB transported from the upstream side (upper side in FIG. 9) in the transport direction. Specifically, when the waste OB being conveyed on the belt conveyor 31 is the waste OB1 or OB2, the sorting arm 33a shown in FIG. 33a), and the waste OB being conveyed on the belt conveyor 21 is the waste OB3 or OB4, the area corresponding to the waste OB3 and OB4 (the sorting area in FIG. 9). the area to the right of arm 33a).

Further, the sorting arm 33b sorts, by type, each of the wastes OB transported from the upstream side in the transport direction (the right side in FIG. 9). Specifically, when the waste OB being conveyed on the belt conveyor 31 is the waste OB1, the sorting arm 33b shown in FIG. 9 moves the waste OB to the stockyard SY11 corresponding to the waste OB1, When the waste OB being conveyed on the belt conveyor 21 is the waste OB2, the waste OB is moved to the stockyard SY12 corresponding to the waste OB2.

If the waste OB being conveyed on the belt conveyor 31 is not either the waste OB1 or OB2, the sorting arm 33b continues on the belt conveyor 31 without moving the waste OB to the stockyard SY11 or SY12. be transported.

That is, in the sorting device 2, for example, depending on the sorting accuracy of the waste OB in the information processing device 1, there is a possibility that the waste OB3 or OB4 is transported to the vicinity of the sorting arm 33b. Further, in the sorting device 2, for example, there is a possibility that the waste OB5, which is not supposed to be included in the waste OB, is transported to the vicinity of the sorting arm 33b.

Therefore, the sorting device 2, for example, does not move the waste OB that the information processing device 1 cannot determine to be either the waste OB1 or OB2 to either the stockyard SY11 or SY12, and Conveyance is continued on the conveyor 31 .

As a result, the sorting device 2 can cause the information processing device 1 to re-determine the type of the waste OB that the information processing device 1 could not determine to be either the waste OB1 or OB2. Therefore, the object sorting system 10 can improve the determination accuracy of the type of waste OB conveyed on the belt conveyor 31 .

Further, the sorting arm 33c sorts, by type, each of the wastes OB transported from the upstream side (left side in FIG. 9) in the transport direction. Specifically, when the waste OB being conveyed on the belt conveyor 31 is the waste OB3, the sorting arm 33c shown in FIG. 9 moves the waste OB to the stockyard SY13 corresponding to the waste OB3, When the waste OB being conveyed on the belt conveyor 21 is the waste OB4, the waste OB is moved to the stockyard SY14 corresponding to the waste OB2.

Then, if the waste OB being conveyed on the belt conveyor 31 is not either the waste OB3 or OB4, the sorting arm 33c continues belt movement without moving the waste OB to either the stock yard SY13 or SY14. It is conveyed on the conveyor 31.

For example, the information processing apparatus 1 may detect waste OB that has been conveyed at the same position on the belt conveyor 31 a predetermined number of times or more. Specifically, the information processing apparatus 1 detects waste OB that has been transported at the same place more than a predetermined number of times by analyzing moving image data captured by each camera on the belt conveyor 31, for example. It's okay. When a waste OB that has been transported at the same location a predetermined number of times or more is detected, the sorting device 2 operates, for example, a predetermined sorting arm (sorting arms other than the sorting arms 33a, 33b, and 33c) (not shown). Detected waste OB may be moved to a predetermined area (not shown) (an area other than the areas corresponding to stockyards SY11, SY12, SY13 and SY14).

As a result, the sorting device 2 can prevent, for example, the waste OB (waste OB5, etc.) not stored in any of the stockyards SY11, SY12, SY13, and SY14 from being continuously conveyed on the belt conveyor 31. become.

[Specific example of configuration when sorting board is used]
Next, a specific example of a configuration using a sorting board 41 (hereinafter also referred to as a sorting mechanism 41 or a sorting device 41) will be described. 10 to 12 are diagrams for explaining a specific example of the configuration when the sorting board 41 is used. Only the configuration different from the sorting device 2 described in FIGS. 2 and 3 will be described below.

The sorting device 2 shown in FIG. 10 has a sorting mechanism 41 including a plurality of areas capable of accommodating waste OB. The sorting mechanism 41 includes partition walls 42 , 43 , 44 , 45 , 46 , 47 , 48 and 49 that extend radially outward rotatably around a shaft (not shown) perpendicular to the conveying surface of the belt conveyor 21 . Prepare. Each of the partition walls 42, 43, 44, 45, 46, 47, 48, and 49 is arranged, for example, so as to equally divide the sorting mechanism 41, and is separated from other partition walls adjacent in the circumferential direction. Stock yards SY21, SY22, SY23, SY24, SY25, SY26, SY27, and SY28 are formed, each having an open radially outer end between them.

In the example shown in FIG. 10, the stockyard SY21 accommodates waste OB whose type has the highest probability of being waste OB3 and whose probability is equal to or greater than a first probability (for example, 90(%)). For example, the stockyard SY22 stores waste OB that has the highest probability of being waste OB3 and has a second probability (for example, 70(%)) or higher. For example, the stockyard SY23 accommodates waste OB that has the highest probability of being waste OB3 and has a third probability (e.g., 50 (%)) or higher. For example, the stockyard SY24 is an area that accommodates waste OB whose type has the highest probability of being waste OB3 and whose probability is less than the third probability.

The stockyard SY25 is, for example, an area that accommodates waste OB that has the highest probability of being the waste OB1 and whose probability is equal to or greater than the first probability. The stockyard SY26 is, for example, The waste OB having the highest probability of being the type of waste OB1 and having the probability of being the second probability or higher is stored in the stockyard SY27. This is the area where the waste OB whose probability is the highest and whose probability is equal to or higher than the third probability is stored. This is an area in which waste OB with a probability of less than 3 is stored.

Specifically, for example, the discrimination result of the type of waste OB in the vicinity of the sorting mechanism 41 indicates that the type of waste OB is the waste OB3 with the highest probability, and the probability is 90 (%) or more. When indicated, the information processing device 1 instructs the control device to rotate the sorting mechanism 41 so that the waste OB is stored in the stockyard SY21. That is, in this case, the information processing apparatus 1 rotates the sorting mechanism 41 so that the stockyard SY21 faces the upstream side in the transport direction, as shown in FIG.

As a result, the sorting device 2 can store the waste OB in the vicinity of the sorting mechanism 41 (the waste OB determined to correspond to the stock yard SY21) in the stock yard SY21, for example, as shown in FIG. become. Therefore, the sorting device 2 can sort, for example, the waste OB in the vicinity of the sorting mechanism 41 by type and by determination accuracy. Therefore, the operator OP can perform different post-processing, for example, for each piece of waste OB stored in a different area.

The stockyards SY21, SY22, SY23, SY24, SY25, SY26, SY27, and SY28 may accommodate, for example, different types of waste OB (8 types of waste OB).

Further, the sorting mechanism 41 may form a number of regions other than eight by having partition walls of a number other than eight, for example.

[Specific example of configuration when compressed air generator is used]
Next, a specific example of the configuration when using the compressed air generator 25 will be described. 13 and 14 are diagrams for explaining a specific example of the configuration when the compressed air generator 25 is used.

The sorting device 2 shown in FIG. 13 has compressed air generators 25a and 25b (hereinafter collectively referred to as compressed air generators 25) in place of the sorting arm 23 described in FIG. 2 and the like.

The compressed air generator 25a is arranged on the upstream side (the right side in FIG. 13) in the transport direction based on the content of the operation by the operator OP via the operation terminal 5 and the determination result of the type of the waste OB by the information processing device 1. Each of the waste OB transported from is sorted by type.

Specifically, as shown in FIG. 13, the compressed air generator 25a is arranged, for example, in an area on the transport path of the waste OB (an area above the belt conveyor 21 in FIG. 13). As in the case of using the sorting arm 23, the compressed air generator 25a sends out compressed air to the waste OB conveyed on the belt conveyor 21, so that each waste OB is It guides to the area corresponding to the type of OB.

More specifically, as shown in FIG. 14, for example, when the waste OB being conveyed on the belt conveyor 21 is the waste OB3, the compressed air generator 25a generates the waste OB (shown in FIG. 14). In the example, the waste OB3a) is moved to the area corresponding to the waste OB3 (the area below the belt conveyor 21 in FIG. 13).

In addition, the compressed air generator 25b is arranged on the upstream side in the transport direction ( right side) are sorted by type.

Specifically, as shown in FIG. 13, the compressed air generator 25b is arranged, for example, in an area on the transport path of the waste OB (an area below the belt conveyor 21 in FIG. 13). Then, the compressed air generator 25b sends out compressed air to the waste OB conveyed on the belt conveyor 21 in the same manner as when the sorting arm 23 is used, thereby separating each waste OB into each waste OB. It guides to the area corresponding to the type of OB.

More specifically, when the waste OB being conveyed on the belt conveyor 21 is the waste OB1, the compressed air generator 25b moves the waste OB to the region corresponding to the waste OB1 (the belt conveyor in FIG. 13). 21).

In the example shown in FIGS. 13 and 14, the case where the compressed air generators 25a and 25b are provided instead of the sorting arm 23 explained in FIG. . Specifically, the compressed air generators 25a and 25b may be provided together with the sorting arm 23 described in FIG. 2 and the like.

1: Information processing device 2: Sorting device 5: Operation terminal 11: Learning data generation unit 12: Learning model generation unit 13: Type determination unit 14: Sorting arm control unit 15: Conveyance control unit 21: Belt conveyor 22: Alignment mechanism 23 : Sorting arm 24: Camera 101: CPU
102: Memory 103: Network Interface 104: Storage Medium 105: Bus

Claims (17)

An object sorting system for sorting objects conveyed by a conveying device by type,
an imaging device that images the object;
an information processing device that determines the type of an object included in specific image data captured by the imaging device;
a selection device that selects a type of target object included in the specific image data according to a determination result by the information processing device;
The information processing device is
By using a learning model learned by associating image data including a learning object with the type of the learning object, the type of the object included in the specific image data is selected from a plurality of type candidates. Identify the probability of being each as the determination result,
identifying the maximum probability among the probabilities included in the determination result, and identifying a type candidate corresponding to the maximum probability as the type of the object included in the specific image data;
The sorting device selects an object included in the specific image data from one or more regions each corresponding to a combination of one of the plurality of type candidates and one of a plurality of ranges of probability. , moving to an area corresponding to the combination of the type of the object identified by the information processing device and the maximum probability;
An object sorting system characterized by: In claim 1,
The sorting device has one or more sorting arms,
Each of the one or more sorting arms is arranged on the transport path of the transport device, and a shaft portion provided at each base end is pivotally supported on the transport path,
Each of the one or more sorting arms rotates the tip portion about the shaft portion to bring the object conveyed by the conveying device into contact with the tip portion, and the one or more regions are selected. , guiding to an area corresponding to the combination of the type of the object identified by the information processing device and the maximum probability;
An object sorting system characterized by: In claim 2,
The one or more sorting arms constitute a multistage sorting arm group arranged at intervals from the upstream side to the downstream side in the transport direction of the transport device,
An object sorting system characterized by: In claim 2,
The sorting device transports the objects transported by the transport device such that the positions of the objects transported by the transport device in the width direction of the transport path are within the rotation range of the sorting arm in the width direction of the transport path. having a guide wall that guides the object to be
An object sorting system characterized by: In claim 2,
At least part of the conveying route is a circular route,
When the sorting device detects an object that has been transported over a predetermined position on the transport path a predetermined number of times, the sorting device moves the detected object to a region other than the one or more regions.
An object sorting system characterized by: In claim 1,
The sorting device has one or more compressed air generators,
Each of the one or more compressed air generators is arranged on a transport path of the transport device,
Each of the one or more compressed air generators sends compressed air to an object conveyed by the conveying device, thereby moving the object conveyed by the conveying device out of the one or more regions. , guiding to an area corresponding to the combination of the type of the object identified by the information processing device and the maximum probability;
An object sorting system characterized by: In claim 1,
The information processing device generates the learning model by learning image data including the learning target object and learning data including the type of the learning target object.
An object sorting system characterized by: In claim 1,
The conveying device is
when the average value of the density of the objects on the transport path of the transport device is greater than a first threshold, reducing the transport speed of the transport device;
if the average value of the densities is less than a second threshold, increasing the conveying speed;
An object sorting system characterized by: A target included in specific image data captured by an imaging device by using a learning model generated by learning image data including a learning target and learning data including the type of the learning target. identifying the probability that the type of thing is each of a plurality of candidate types;
identifying the maximum probability among the identified probabilities, and identifying a type candidate corresponding to the maximum probability as the type of object included in the specific image data;
The object included in the specific image data is identified from among one or more regions corresponding to combinations of one of the plurality of types candidates and one of a plurality of ranges of probability. transmitting the identified object type and the maximum probability to a sorting device that moves the object to a region corresponding to the combination of the object type and the maximum probability;
An object selection program characterized by causing a computer to execute processing. A target included in specific image data captured by an imaging device by using a learning model generated by learning image data including a learning target and learning data including the type of the learning target. identifying the probability that the type of an object is each of a plurality of type candidates; identifying the maximum probability among the identified probabilities; a determination unit that identifies the type of object;
The object included in the specific image data is identified from among one or more regions corresponding to combinations of one of the plurality of types candidates and one of a plurality of ranges of probability. a transmission unit that transmits the type of the identified object and the maximum probability to a sorting device that moves to an area corresponding to the combination of the type of object and the maximum probability;
An information processing device characterized by: A target included in specific image data captured by an imaging device by using a learning model generated by learning image data including a learning target and learning data including the type of the learning target. identifying the probability that the type of thing is each of a plurality of candidate types;
identifying the maximum probability among the identified probabilities, and identifying a type candidate corresponding to the maximum probability as the type of object included in the specific image data;
The object included in the specific image data is identified from among one or more regions corresponding to combinations of one of the plurality of types candidates and one of a plurality of ranges of probability. transmitting the identified object type and the maximum probability to a sorting device that moves the object to a region corresponding to the combination of the object type and the maximum probability;
A method of selecting an object, characterized in that the processing is executed by a computer. An object sorting method for sorting objects conveyed by a conveying device by type,
an imaging device imaging the object;
An information processing device determines the type of an object included in specific image data captured by the imaging device,
A sorting device sorts out the type of object included in the specific image data according to the determination result by the information processing device,
The information processing device is
By using a learning model learned by associating image data including a learning object with the type of the learning object, the type of the object included in the specific image data is selected from a plurality of type candidates. Identify the probability of being each as the determination result,
identifying the maximum probability among the probabilities included in the determination result, and identifying a type candidate corresponding to the maximum probability as the type of the object included in the specific image data;
The sorting device selects an object included in the specific image data from one or more regions each corresponding to a combination of one of the plurality of type candidates and one of a plurality of ranges of probability. , moving to an area corresponding to the combination of the type of the object identified by the information processing device and the maximum probability;
An object selection method characterized by: a conveying device that conveys an object on a conveying route;
Shafts arranged on the conveying path and provided at respective base ends are pivotally supported on the conveying path, and conveyed by the conveying device by rotating the tip portion around the shafts. the object to be conveyed by the conveying device is selected from among a plurality of candidate types of the object to be conveyed by the conveying device and a plurality of ranges of probability. one or more sorting arms for guiding to a region determined by an information processing device as a region corresponding to an object to be transported by the transport device, from among the one or more regions corresponding to a combination of any of the above;
Guide the objects conveyed by the conveying device so that the positions of the objects conveyed by the conveying device in the width direction of the conveying path are within the rotation range of the sorting arm in the width direction of the conveying path. a guide wall for
The information processing device is
By using a learning model learned by associating image data including a learning target with the type of the learning target, specific image data obtained by imaging the target transported by the transport device includes: identifying the probability that the type of the object to be received is each of the plurality of type candidates;
identifying the maximum probability among the identified probabilities, and identifying a type candidate corresponding to the identified maximum probability as the type of object included in the specific image data;
Each of the one or more selection arms selects an object included in the specific image data according to the combination of the type of the object specified by the information processing device and the maximum probability among the one or more regions. to guide you to the corresponding area,
A sorting device characterized by: In claim 13,
The one or more sorting arms constitute a multistage sorting arm group arranged at intervals from the upstream side to the downstream side in the transport direction of the transport device,
A sorting device characterized by: In claim 13,
At least part of the conveying route is a circular route,
A sorting device characterized by: a conveying device that conveys an object on a conveying route;
A plurality of objects to be transported by the transporting device are arranged on the transporting path and transported by the transporting device by sending compressed air to the objects to be transported by the transporting device. by the information processing device as a region corresponding to the object transported by the transport device, among one or more regions respectively corresponding to a combination of one of the type candidates and one of a plurality of ranges of probability one or more compressed air generators for guiding to the determined area;
The information processing device is
By using a learning model learned by associating image data including a learning target with the type of the learning target, specific image data obtained by imaging the target transported by the transport device includes: identifying the probability that the type of the object to be received is each of the plurality of type candidates;
identifying the maximum probability among the identified probabilities, and identifying a type candidate corresponding to the identified maximum probability as the type of object included in the specific image data;
Each of the one or more compressed air generators selects an object included in the specific image data as the type of the object identified by the information processing device from among the one or more regions and the maximum probability. to guide to the area corresponding to the combination,
A sorting device characterized by: a conveying device that conveys an object on a conveying route;
one or more partition walls arranged on the conveying path, a shaft supported on the conveying path, and extending radially outward rotatably around the shaft; each of which forms one or more regions with open radially outer ends between other partition walls adjacent in the circumferential direction;
By rotating the one or more partition walls, the sorter selects a combination of one of a plurality of candidate types of objects to be transported by the transport device and one of a plurality of ranges of probability. Of the one or more corresponding regions, the region determined by the information processing device as the region corresponding to the object to be transported by the transport device is opposed to the upstream side in the transport direction of the transport device, and the transport is performed. guiding an object conveyed by the device to the area determined by the information processing device;
The information processing device is
By using a learning model learned by associating image data including a learning target with the type of the learning target, specific image data obtained by imaging the target transported by the transport device includes: identifying the probability that the type of the object to be received is each of the plurality of type candidates;
identifying the maximum probability among the identified probabilities, and identifying a type candidate corresponding to the identified maximum probability as the type of object included in the specific image data;
The classifier guides the target object included in the specific image data to a region corresponding to the combination of the type of the target object specified by the information processing device and the maximum probability, among the one or more regions. do,
A sorting device characterized by:

Images