TWI888393B - Automatic tableware washing system, automatic tableware washing method, automatic tableware washing program, and storage medium - Google Patents

Abstract

The purpose of the present invention is to provide an automatic tableware washing system, which can identify the type and state of each tableware while the tableware is being transported on a conveyor at different three-dimensional angles, and the robot can safely pick up the tableware and store it in a specific storage location.

The automatic tableware washing system of the embodiment of the present invention is equipped with a tableware input part, a tableware washing part, a tableware conveying part, a tableware information acquisition part for identifying tableware and acquiring tableware information, a tableware placing part, and a tableware picking part. The characteristic of the system is that the information acquired by the tableware information acquisition part includes at least one of tableware type, tableware size, tableware shape, tableware color, tableware position, tableware posture, tableware orientation, tableware mark, or tableware code. If the tableware information acquisition part identifies tableware being conveyed by the tableware conveying part, the tableware picking part picks up the tableware according to the state of the tableware based on the information acquired by the tableware information acquisition part.

Description

Automatic tableware washing system, automatic tableware washing method, automatic tableware washing program, and storage medium

The present invention relates to a system for automatically washing tableware, etc.

In recent years, due to the shortage of restaurant operators, there is a demand to automate the steps of restaurant operations or services. There is also a demand for automation of dishwashing, and a technology such as Patent Document 1 has been proposed previously.

The tableware washing system described in the above-mentioned patent document 1 sets a mark or text that can be identified by humans in advance in the base of the ceramics in order to identify ceramics. The camera obtains the image on the washing conveyor, reads the mark or text set on the ceramics by template comparison from the obtained image, and thereby identifies the ceramics and selects the ceramics by the machine arm mechanism.

The tableware washing system described in the above-mentioned patent document 2 is a system in which the tableware base is marked with an identifier that can be identified by humans and used to determine the type of tableware (size, tea bowl, bowl, plate, etc.), and a symbol for specifying the position, range and direction of the identifier. The image on the washing conveyor is obtained by a camera, and the identifier set on the ceramic is identified from the obtained image by pattern matching, and the tableware is selected by a robot arm mechanism. In addition, in patent document 2, the identifier is a person-identifiable one, excluding barcodes, two-dimensional codes, magnetic codes, etc.

The tableware washing system described in the above-mentioned patent document 3 processes the images obtained by the camera in the picking area to identify the type of tableware and calculate the center of gravity position of the tableware taken out from the tableware washing machine and moved from the waiting area to the picking area and stationary, and selects the tableware by the machine arm mechanism and stacks the tableware in the release area according to the type. In patent document 3, pattern matching processing is performed as image processing to determine the type of tableware, the center of gravity position, and the direction of the long side of the plate when viewed from above (the direction of the plate on the rectangle).

The tableware washing system described in the above-mentioned patent document 4 is performed by a robot arm: moving the washing tray into the washing room, opening and closing the door of the tableware washing machine, pressing the washing switch, moving the washing tray from the washing room to the unloading table, and selecting tableware from the unloading table and transferring the tableware to the storage tray. When the washing tray is stationary, the selection camera takes an image of the selection area on the unloading table corresponding to the position of the washing tray, and the image is subjected to pattern matching processing to determine the type of tableware, the center of gravity of the tableware, and the direction of the long side when the plate is viewed from above (the direction of the plate on the rectangle).

[Prior Art Literature] [Patent Literature]

[Patent document 1] Japanese Patent Publication No. 2018-126272

[Patent Document 2] Japanese Patent Publication No. 2019-037349

[Patent Document 3] International Publication No. 2018/034251

[Patent Document 4] International Publication No. 2018/034252

In the tableware washing system described in the above-mentioned patent document 1, the marks or characters that can be identified by humans are placed in the feet of the ceramics. The ceramics are identified and selected by the machine arm mechanism by comparing the template with the image obtained by the camera. However, the premise is that all the ceramics must be arranged horizontally upside down relative to the conveyor (the feet of the ceramics are facing upwards), and there is no description for distinguishing the types of ceramics. When the ceramics are transported on the conveyor at different three-dimensional angles, the tableware washing system of patent document 1 cannot cope with it. In addition, the marks or characters placed in the feet of the ceramics are limited to those that can be identified by humans. Furthermore, it is not possible to automatically arrange the ceramics on the conveyor for washing.

In the tableware washing system described in the above-mentioned patent document 2, an identifier for determining the type of tableware and a symbol for specifying the position, range and direction of the identifier are set in the foot of the tableware, and the tableware washing system determines the type of tableware through pattern matching processing. However, the premise is that all the tableware must be used in a state where the tableware is horizontally arranged upside down relative to the conveyor (the foot of the tableware is facing directly upward). When the tableware is transported on the conveyor at different three-dimensional angles, the tableware washing system of patent document 2 cannot cope with it. In addition, the mark or text set in the foot of the ceramic ware is limited to those that can be identified by humans. Furthermore, it is not possible to automatically arrange the tableware on the conveyor for washing.

In the tableware washing system described in the above-mentioned patent document 3, the tableware stationary in the picking area is processed by the image obtained by the picking area camera and the tableware type is identified. Therefore, it is necessary to distinguish the picking conveyor and the washing conveyor in the picking area, and then set a relay conveyor in between, so that the structure of the conveyor becomes complicated. In addition, its premise is that all the tableware must be arranged horizontally with respect to the conveyor upside down (the foot of the tableware is facing upward). When the tableware is transported on the conveyor at different three-dimensional angles, the tableware washing system of patent document 3 cannot cope with it. Furthermore, it is not possible to automatically arrange the tableware on the washing conveyor.

In the tableware washing system described in the above-mentioned patent document 4, when the washing tray is stationary, a picking camera is used to capture an image of the picking area on the unloading table corresponding to the position of the washing tray, and the image is subjected to pattern matching processing to identify the type of tableware. Therefore, it cannot be used to pick up tableware moving on a conveyor. In addition, the premise is that all tableware must be arranged horizontally upside down relative to the conveyor (the feet of the utensils are facing directly upward). When the tableware is transported on the conveyor at different three-dimensional angles, the tableware washing system of patent document 4 cannot cope with it. Furthermore, it is not possible to automatically arrange the tableware on the washing tray.

In the above-mentioned patent documents 1 to 4, the situation of a robot picking up a plurality of tableware while the tableware is being transported on a conveyor at different three-dimensional angles is not considered. The inventor of the present invention aims to realize a fully automatic tableware washing system. Based on conditions such as automating the stage of arranging tableware before washing on the conveyor, arranging tableware at an angle that is not limited to the state where the tableware feet are facing upward, picking up tableware moving on the conveyor, and restricting the robot's posture, the inventor of the present invention has conducted static and dynamic analysis of tableware and robots, and has conducted in-depth research and development on a fully automatic tableware washing system, and finally completed the implementation form of the present invention. That is, the purpose of the present invention is to provide an automatic tableware washing system that can identify the type or state of each tableware while the tableware is being transported on a conveyor at different three-dimensional angles, and the robot can safely pick up the tableware and place it in a specific storage position.

The above-mentioned purpose of the present invention can be achieved by the following structure. That is, the first aspect of the present invention is an automatic tableware washing system comprising: a tableware washing section, which washes the tableware put in from the tableware putting section; a tableware conveying section, which conveys the tableware washed by the tableware washing section; a tableware information acquisition section, which identifies the tableware conveyed by the tableware conveying section and acquires the information of the tableware; a tableware placing section, which places the tableware conveyed by the tableware conveying section; and a tableware taking section, which takes the tableware from the tableware conveying section and gives the tableware to the tableware conveyor. The loading unit loads the tableware; and the feature is that the information acquired by the tableware information acquisition unit includes at least one of the tableware type, tableware size, tableware shape, tableware color, tableware position, tableware posture, tableware orientation, a mark attached to the tableware, or a code attached to the tableware; if the tableware information acquisition unit identifies the tableware transported by the tableware transport unit, the tableware picking unit picks up the tableware according to the state of the tableware based on the information acquired by the tableware information acquisition unit.

According to the implementation form of the present invention, an automatic tableware washing system can be provided, which can identify the type or state of each tableware while the tableware is being transported on a conveyor at different three-dimensional angles, and the robot can safely pick up the tableware and place it in a specific storage position.

1: Tableware feeding section

2: Dishwashing Department

3: Tableware removal section

10: Tableware transport department

11: Tableware

12: Multi-joint robot

13: Tableware placement part

14: Tableware information acquisition department

30: Tableware removal calculation unit

31: Tableware position/posture detection unit

32: Tableware removal sequence planning department

33: Tableware removal action command unit

35: Image acquisition unit

36: Tableware feeding section

37: Tableware transport department

38: Dishwashing Department

40:Manipulator

40A:Manipulator

40C: Robotic arm

41: Adsorption part

41A: 1st adsorption part

41B: Second adsorption part

42:AR marker

45: Extension

46: Robot installation department

47: Vacuum hose

48: Vacuum hose installation part on the robot side

49: Junction

50: Detection frame

51:Category display

55: Handover platform

56: Container storage robot

57:Container

60: Spring mechanism

61:Ball joint

62: Force sensor

70: Cross-transportation platform

80: Detector

S1~S5: Steps

Figure 1 is a step diagram of the automatic tableware washing system.

Figures 2A and 2B are the overall structure diagrams of the automatic tableware washing system.

Figure 3 is a block diagram of the tableware removal section.

Figure 4A of Figure 4 is a top view of the tableware removal section, and Figure 4B is a three-dimensional view of the tableware removal section.

Figure 5 is an illustration of an example of how to use the marking.

Figure 6 is a three-dimensional diagram of the robot.

Figure 7A of Figure 7 is a three-dimensional diagram of a variation of the robot, and Figure 7B is a side view of a variation of the robot.

Figure 8A of Figure 8 is a three-dimensional view of another variation of the robot, and Figure 8B is a side view of another variation of the robot.

Figure 9A of Figure 9 is an explanatory diagram of the upper stacking tableware placement section, and Figure 9B is an explanatory diagram of the lower stacking tableware placement section.

Figure 10A of Figure 10 is a three-dimensional diagram of a variation of the robot, and Figure 10B is a side view of a variation of the robot.

Figures 11A and 11B show a method of directly printing AR (Augmented Reality) marks on the back of tableware and then applying them.

Figure 12 is an illustration of image recognition of tableware without tableware marking.

Figure 13 is a top view of the tableware removal section of implementation form 4.

Figure 14A of Figure 14 is a top view of the automatic tableware washing system, and Figure 14B is a three-dimensional view of the automatic tableware washing system.

Figure 15 shows the periphery of the transfer station of the automatic tableware washing system.

The following is a description of the automatic tableware washing system of the embodiment of the present invention with reference to the drawings. However, the embodiments shown below are examples of automatic tableware washing systems that embody the technical ideas of the present invention, and the present invention is not limited to these embodiments, and can also be applied to other embodiments covered by the scope of the patent application.

[First implementation form]

The automatic tableware washing system of the first embodiment of the present invention is described with reference to Figures 1 to 9.

FIG1 is an example of a step diagram of the automatic tableware washing system of the present embodiment. The automatic tableware washing system of the present embodiment includes a tableware input part 1, a tableware washing part 2, and a tableware removal part 3. In addition, the automatic tableware washing system of the present embodiment includes a step S1 of removing residues or garbage from the tableware, a step S2 of arranging the tableware on a conveyor, a step S3 of washing the tableware with water flow and detergent, a step S4 of drying, and a step S5 of moving the tableware from the conveyor to a storage place. In the tableware feeding section 1, step S1 of removing residues or garbage from the tableware and step S2 of arranging the tableware and placing it on the conveyor are performed; in the tableware washing section 2, step S3 of washing the tableware with water flow and detergent and step S4 of drying are performed; in the tableware taking-out section 3, step S5 of moving the tableware from the conveyor to the storage place is performed.

FIG. 2 is an example of the overall configuration diagram of the automatic tableware washing system of this embodiment. Tableware 11 includes all utensils or appliances used for dining, such as Western tableware, Japanese tableware, shapes, etc., such as: plates, basins, bowls, bowls, bowls, etc. for holding food; cups with handles, cups without handles, glasses, tea bowls, wine vessels, etc. for holding drinks; chopsticks, forks, knives, spoons, stirring rods, spoons, etc. for handling food; cooking utensils or kitchen supplies, i.e. pots, frying pans or cutting boards, kitchen knives, chef knives, basins, measuring cups, containers, spatulas, scrapers, ladles, etc. Tableware 11 is transported from the tableware input part 1 to the tableware removal part 3 via the tableware washing part 2 by the tableware transporting part 10. The tableware conveying section 10 can be, for example, a single conveyor for conveying tableware from the tableware input section 1 through the tableware washing section 2 to the tableware taking-out section 3, but can also be divided into, for example, a conveyor for the tableware input section 1, a conveyor for the tableware washing section 2, and a conveyor for the tableware taking-out section 3 and composed of a plurality of conveyors, and can further be divided into a plurality of conveyors and provided in each section, or can be provided as a single conveyor instead of a plurality of conveyors.

Although the tableware conveying part 10 is not particularly limited, it is, for example, a conveyor that can convey tableware. In addition, as a type of conveyor, it can be adapted to various types of conveyors, for example: a) The conveyor surface is a flat surface, and the tableware is pre-inverted on the conveyor, that is, a flat conveyor; b) a space is provided under the conveyor surface for inserting tableware, which can be adapted to thinner tableware or even under the washing tray; an underflight conveyor; c) a flight conveyor that washes the tableware standing against a protrusion on the conveyor, which can be adapted to thinner tableware or even a washing tray; d) a tray conveyor used to wash the tableware stored in the washing tray, etc.

The tableware taking-out section 3 is provided with: a multi-joint robot 12 as a tableware taking-out section; a tableware placing section 13 for placing the tableware taken from the tableware conveying section 10 by the tableware taking-out section in a centralized manner; and a tableware information acquiring section 14 for identifying the tableware conveyed by the tableware conveying section 10 and acquiring tableware information. In the tableware taking-out section 3, the multi-joint robot 12 as the tableware taking-out section takes the tableware based on the tableware information from the tableware information acquiring section 14, and the taken-out tableware is placed in a centralized manner on the tableware placing section 13. Here, the multi-joint robot 12 is exemplified as the tableware taking-out section, but the present embodiment is not limited thereto, and includes all tableware taking-out mechanisms as long as they have a mechanism for taking tableware, such as industrial robots such as robots with freely deformable mechanical arms and lifting robots. In the tableware loading section 13, the tableware taken out by the tableware taking section is collectively loaded, for example, by type. As a method of collectively loading tableware in the tableware loading section 13, for example, the same type of tableware is stacked and loaded in the same direction in an inverted state. The tableware taking section 3 can grasp the type and posture of the tableware based on the information from the tableware information acquisition section 14, so that, for example, the tableware can be stacked and loaded on the tableware loading section 13 in a state where the long side and the short side of the tableware in a rectangular shape when viewed from above are in the same direction. The types of tableware include the size, shape, color, etc. of the tableware. In addition, the multi-joint robot 12 can also take tableware from the tableware conveying section 10 even when the tableware conveying section 10 is conveying tableware, that is, when the tableware is moving.

In this embodiment, the tableware automatic washing system is described from the perspective of fully automating it, but for example, part of the tableware feeding part 1 may be manually operated.

The tableware input section 1 includes a tableware immersion unit, a tableware extraction unit, and a cleaning pump. The tableware immersion unit has a tableware immersion tank and a nozzle, and the tableware extraction unit has a tableware extraction conveyor, a tableware alignment section, and a filter. Water is stored in the tableware immersion tank, and the water flow from the nozzle removes residues or garbage from the tableware. The water filtered by the filter is pressurized by the cleaning pump and sprayed from the nozzle. The tableware after the residues or garbage are removed by the water flow from the nozzle is transported to the tableware washing section 2 by the tableware extraction conveyor. In addition, the cleaning pump is depicted at the bottom of the tableware washing section, the filter is depicted at the bottom of the tableware extraction conveyor, and the nozzle is depicted at the front of the tableware extraction conveyor, but this is only an example, and the configuration of the cleaning pump, filter, and nozzle can be appropriately changed according to the shape or configuration of the tableware immersion unit, tableware retrieval unit, and tableware washing section.

The tableware alignment unit is a mechanism that aligns the types or postures of tableware conveyed to the tableware washing unit on the tableware extraction conveyor. For example, a multi-joint robot or other industrial robot can be used. The tableware to be picked up by the tableware picking conveyor may be overlapped or in a state where the tableware posture is not suitable for washing in the tableware washing unit 2. The tableware alignment unit adjusts the configuration or posture of the tableware to make it easier for the tableware washing unit 2 to wash the tableware or to remove the tableware in the tableware removal unit 3. If, for example, a large plate is conveyed in an upside-down state on the tableware extraction conveyor, other smaller tableware such as small plates or bowls may overlap under the large plate, and there is a risk of residue in the subsequent step, namely, the tableware washing section 2. In addition, if the tableware overlaps, the tableware information acquisition section 14 may not obtain the tableware information, and in this case, there is a possibility that the tableware is missed in the tableware extraction section 3. Therefore, if there is a large plate in an upside-down state on the tableware extraction conveyor, the large plate is adjusted to stand against the protrusion on the tableware retrieval conveyor to avoid, for example, the large plate overlapping other tableware.

Furthermore, if the tableware is arranged in a consistent posture relative to the conveying direction of the conveyor, for example, the tableware is arranged in a consistent manner with the back (bottom) facing the conveyor's travel direction, each tableware can be properly washed in the tableware washing section 2, and each tableware can be easily taken out in the tableware taking-out section 3. Furthermore, as described later, in the tableware feeding section 1, tableware is fed in according to the predetermined rules of tableware type, position and posture, thereby improving the image recognition accuracy of the tableware type, position and posture in the tableware taking-out section 3, reducing tableware picking failures, and increasing the speed of picking up tableware.

According to the type of the tableware washing section 2, that is, according to the configuration of the tableware input position and the tableware removal position, the configuration of the tableware input section 1 and the tableware removal section 3 is adjusted. Therefore, the overall configuration diagram of the tableware automatic washing system shown in Figure 2 is an example of this embodiment, and is for explaining the straight line type. This embodiment is not limited to the layout shown in the figure, and can be a layout corresponding to other types of tableware washing sections 2 such as corner types or floor types.

FIG3 is an example of a block diagram showing the tableware removal unit 3 of the present embodiment. The tableware removal unit 3 includes an image acquisition unit as the tableware information acquisition unit 14, a tableware removal calculation unit 30, and a multi-joint robot 12. The tableware removal calculation unit 30 is composed of, for example, a personal computer (hereinafter referred to as "PC (Personal Computer)") or a tablet terminal, and has a tableware position/posture detection unit 31, a tableware removal sequence planning unit 32, and a tableware removal action instruction unit 33. The tableware position/posture detection unit 31 obtains information related to the tableware position, posture, and type from the tableware information acquisition unit 14.

The tableware removal sequence planning unit 32 outputs the conveyor operation or stop command to the tableware transport unit 10, and outputs the conveyor operation or stop command corresponding to the conveyor operation to the tableware input unit 1.

In addition, the tableware position/posture detection unit 31 is connected to the tableware input unit 1, the tableware conveying unit 10, and the tableware washing unit 2 in addition to being connected to the tableware information acquisition unit 14. When it is possible to communicate with each unit, the tableware position/posture detection unit 31 can also use, for example, tableware position or posture information when tableware is input from the tableware input unit 1, tableware conveying speed information from the tableware conveying unit, and information related to the strength of the washing water flow in the tableware washing unit to obtain information related to the tableware position, posture, and type.

The tableware removal action instruction unit 33 outputs the tableware removal action instruction to the multi-joint robot 12. Based on the instruction, the multi-joint robot 12 takes the tableware from the tableware conveying unit 10 and places the tableware on the tableware loading unit.

FIG. 4 is an example of an overview diagram of the tableware taking-out section 3 of the present embodiment. FIG. 4A is a top view of the tableware taking-out section 3, and FIG. 4B is a three-dimensional view of the tableware taking-out section 3. In the tableware taking-out section 3, based on the tableware information from the tableware information acquisition section 14, the multi-joint robot 12 serving as the tableware taking section takes the tableware conveyed by the tableware conveying section 10, and collectively places the taken tableware on the tableware placing section 13. Here, in FIG. 4A and FIG. 4B, the configuration of the multi-joint robot 12, the tableware conveying section 10, and the tableware placing section 13 is an example of the present embodiment, and the tableware conveying section 10 may also be provided, for example, on the opposite side of the multi-joint robot 12 across the tableware conveying section 10. In this way, by setting the distance between the position where the multi-joint robot 12 picks up the tableware on the tableware conveying part 10 and the tableware placing part 13 as short as possible, the movement of the tableware picking part 12 can be accelerated.

At the end of the moving side of the tableware conveying section 10, a tableware receiving section (not shown) is provided for receiving tableware that the multi-joint robot 12 cannot pick up. Since each joint of the multi-joint robot 12 has a movable limit, when it moves from one posture to the next, there will be restrictions accompanying the movable limit. Depending on the posture of the tableware, it is possible that the multi-joint robot 12 cannot pick up the tableware due to the movable limit. In addition, if the tableware information acquisition section 14 cannot properly read the AR marker 42, for example, when the tableware overlaps each other, there is a possibility that the multi-joint robot 12 cannot pick up the tableware. Therefore, by pre-arranging a tableware receiving portion at the end of the moving side of the tableware conveying portion 10, the tableware can be safely received even when, for example, the multi-joint robot 12 is unable to pick up the tableware, so that the tableware automatic washing system can have redundancy.

FIG5 is an example of an explanatory diagram of a usage example of a marker of this embodiment. In this embodiment, the following example is described: a marker such as an AR marker 42 is set on the back (bottom) of tableware, so that the tableware detection camera of the tableware information acquisition unit 14 acquires an image of the tableware and performs image recognition, thereby identifying the tableware and acquiring information related to the position, posture, and type of the tableware.

As shown in FIG4 , the tableware detection camera serving as the tableware information acquisition unit 14 is disposed, for example, above the tableware conveying unit 10 and upstream of the tableware picking unit 12, and on both sides of the tableware conveying unit 10, facing the center of the width direction of the tableware conveying unit 10. Marks that vary depending on the type of tableware, such as an AR mark 42, are disposed on the back (bottom) of the tableware. In addition, the mark used in this embodiment is not limited to the AR mark 42, and any mark may be used as long as it can identify at least one of the type of tableware, the position of the tableware, or the posture of the tableware.

In addition, in FIG. 4, two tableware information acquisition units 14 are provided, but the number of tableware information acquisition units 14 is not limited to two, as long as it is one or more. For example, four tableware information acquisition units 14 may be provided. In the case of providing four tableware information acquisition units 14, the heights of the two tableware information acquisition units 14 shown in FIG. 4 are set to be the same, and the tableware information acquisition units 14 are provided at positions away from the conveying direction of the conveyor. For example, if the four tableware information acquisition units 14 are arranged at positions corresponding to the corners of the quadrilateral conveyor line, and the orientation is adjusted in such a way that the optical axis is focused on a specific area at the center of the width direction of the conveyor, the images of the tableware can be obtained from both the forward and reverse directions of the conveying direction by the four tableware information acquisition units 14.

That is, if four tableware information acquisition units 14 are provided, even if the tableware 11 does not conform to the rule of being arranged with the back side (bottom) of the tableware 11 facing the conveyor's traveling direction, the image of the back side (bottom) of the tableware 11 can be acquired by the tableware information acquisition unit 14 whose optical axis is tilted toward the conveyor's traveling direction.

Regarding the AR marker 42, "AR marker means a marker for specifying a location where additional information is displayed in an image recognition type AR (Augmented Reality) system, that is, a predetermined pattern image." (According to the website IT terminology dictionary e-Words). By using the AR marker 42, in the image detected by the tableware detection camera, although the image of the tableware changes due to the viewing angle or distance, as long as the AR marker 42 is used, the digital information attached to the AR marker 42 can be read, and the posture of the AR marker 42, that is, the posture of the tableware with the AR marker attached, can be detected. Furthermore, if the type of tableware is encoded in the AR marker 42 in advance, the position, posture, and type of the tableware can be detected as tableware information by photographing the AR marker with a tableware detection camera. In addition, the center position of the back side (bottom) of the tableware is presented as the position information of the tableware obtained by photographing the AR marker 42.

A robot arm 40 is installed on the front end side of the robot arm of the multi-joint robot 12, and a suction part 41 for sucking tableware is provided on the front end side of the robot arm 40.

FIG6 is an example of a three-dimensional diagram of a manipulator 40 of the present embodiment. The manipulator 40 has an extension portion 45, a manipulator mounting portion 46 on one end of the extension portion 45, a suction portion 41 on the other end of the extension portion 45, and a vacuum hose 47. The manipulator mounting portion 46 is mounted on the side of the manipulator arm of the multi-joint robot 12. The piping of the vacuum hose 47 extends through the extension portion 45 to the suction portion 41, and the suction portion 41 can suction the back side (bottom foot) of the tableware by vacuum. In addition, the other end of the vacuum hose is mounted on the robot side vacuum hose mounting portion (omitted from the figure) provided on the multi-joint robot 12, and the suction portion 41 is vacuum suctioned by the robot side vacuum hose mounting portion.

When the multi-joint robot 12 sucks and transports tableware, if larger tableware or heavier tableware is transported at high speed, the vibration becomes larger and the subsequent stacking may not be carried out smoothly. The reason for the vibration is that the suction part 41 of the robot 40 is deformed after sucking the tableware. By providing a ring-shaped component with higher rigidity than the robot in a manner of surrounding the suction part 41, and sucking the tableware with internal force during suction to suppress the deformation of the robot 40, the vibration of the tableware when the multi-joint robot 12 sucks and transports the tableware can be greatly reduced.

FIG7 is an example of a three-dimensional diagram showing another embodiment of the manipulator 40A of the present embodiment. The manipulator 40A has an extension portion 45, a manipulator mounting portion 46 on one end side of the extension portion 45, a coupling portion 49 on the other end side of the extension portion 45, a first suction portion 41A and a second suction portion 41B provided at the coupling portion, and a manipulator-side vacuum hose mounting portion 48. A vacuum hose not shown is mounted on the manipulator-side vacuum hose mounting portion 48, and the other end of the vacuum hose is mounted on the robot-side vacuum hose mounting portion (omitted from the diagram). The piping from the manipulator-side vacuum hose mounting portion extends through the extension portion 45 to the suction portions 41A and 41B, and the suction portions 41A and 41B can suction the back side (bottom) of the tableware by vacuum. The two suction parts 41A and 41B are arranged at a 90-degree interval on the joint part 49. Since the suction parts 41A and 41B are two parts, the multi-joint robot 12 can pick up the tableware without taking an unreasonable posture. In addition, the proportion of tableware removal failure can be reduced.

FIG8A of FIG8 is a three-dimensional view of another variation of the robot, and FIG8B is a side view of another variation of the robot. In FIG8, the same symbols are used for the same components as FIG7, and their descriptions are omitted.

The multi-joint robot 12 may perform a picking action at a position other than the tableware 11 due to an error in the recognition of the tableware 11 (the position and posture of the tableware 11). For example, the multi-joint robot 12 may perform a picking action at a position behind the tableware 11.

At this time, a part of the robot 40 such as the suction part 41 will contact the conveyor or other components of the tableware conveying part 10, and when the tableware 11 cannot be picked up, the robot 40 will be pushed toward the conveyor, which may cause damage to the conveyor, tableware 11 or the robot 40.

Previously, a force sensor (6-axis sensor) was installed on the robot 40 to detect the force carried by the robot 40. When the robot 40 is loaded with excessive external force, the multi-joint robot 12 will stop moving.

However, since force sensors are not waterproof, easily damaged by impact, and expensive, there is a demand to omit them.

The robot 40C has a plurality of spring mechanisms 60 and ball joints 61. In addition, the robot 40C also has a force sensor 62.

The plurality of spring mechanisms 60 and ball joints 61 are used to reduce the reaction force (external force) borne by the robot 40C from the pushing object of the conveyor, etc., as well as the pushing force borne by the conveyor and/or tableware 11 from the robot 40C.

A plurality of spring mechanisms 60 are disposed between the circular plate-shaped portion, i.e., the robot mounting portion 46, and the circular plate-shaped component disposed opposite to the robot mounting portion 46.

The spring mechanism 60 is a device that reduces the external force by elastically deforming the elastic part between the robot mounting portion 46 and the circular plate-shaped component when the robot 40C is pushed toward the conveyor, etc.

The spring mechanism 60 does not deform in an elastic neutral state without external force by appropriately setting the spring constant, and only deforms elastically when an excessive external force is applied.

The ball joint 61 is arranged at the center of the robot mounting portion 46. Therefore, the elastic force of the spring mechanism 60 acts only in the direction of rotation with the ball joint 61 as the center.

Therefore, even if a torque with a larger translation force is applied to the wrist due to the presence of the adsorption parts 41A and 41B at a distance from the wrist of the manipulator 40C (manipulator mounting part 46), the manipulator 40C is prevented from being bent unnecessarily in the translation direction, and the spring mechanism 60 is elastically deformed along the torque received from the adsorption parts 41A and 41B.

By using the robot 40C, even if the tableware 11 cannot be picked up and a part of the robot 40C such as the suction part 41 is pushed toward the conveyor, the reaction force (external force) from the conveyor and other objects of pushing and the pressure from the conveyor and the tableware 11 to the robot 40C can be reduced, so the conveyor, the tableware 11 and the robot 40C can be prevented from being damaged.

In addition, in this embodiment, the elastic force acts only in the direction of rotation around the ball joint 61, but it can also be configured so that the elastic force acts in the axial direction relative to the ball joint 61.

FIG. 9 is an explanatory diagram of the tableware loading section 13 of the present embodiment, FIG. 9A is an explanatory diagram of the stacking type tableware loading section, and FIG. 9B is an explanatory diagram of the bottom stacking type tableware loading section. In the case of the stacking type tableware loading section 13 in FIG. 9A , the upper surface of the tableware loading section 13 is set to be a plane. The multi-joint robot 12 stacks the tableware at a specific location of the tableware loading section 13 according to the type of tableware based on the tableware type information obtained by the tableware information acquisition section 14. In the tableware removal calculation section 30, since the number of tableware loaded in the tableware loading section 13 is calculated according to the type of tableware, the height of the tableware stack corresponding to the type and quantity of the tableware can be predicted. In addition, a mechanism for detecting the presence of tableware according to the position and a mechanism for detecting the height of the tableware stacking can be set in advance in the tableware placing portion 13. For example, by analyzing and photographing the tableware placing portion 13, a tableware placing portion image acquisition portion (not shown) can detect the position, type and number of tableware placed in the tableware placing portion 13.

In the case of the stacking tableware loading section 13 of FIG. 9B , by predetermining the position of the stacked tableware according to the type, a mechanism that rises and falls according to the weight of the tableware is set at each stacking position, so that the next stacked tableware is exactly located at the position on the upper surface of the tableware loading section 13. In this way, since the multi-joint robot 12 knows the location and height of the stacked tableware in advance, each tableware can be stacked correctly. In addition, if a mechanism that rises and falls according to the weight of the tableware is not set at each stacking position, the tableware taking-out calculation section 30 calculates how many tableware are loaded on the tableware loading section 13 according to the type of tableware in advance, and a mechanism that detects the height of the stacked tableware is set in advance, so that each tableware can be loaded until a specific stacking height. Furthermore, if an appropriate mechanism is provided to detect the height of each tableware, the stacking position of each tableware can also be indicated to the multi-joint robot 12.

[About tableware information acquisition department]

In order to stack the tableware in the tableware loading section 13 according to the type of tableware, AR markers are set on the back (bottom) of the tableware according to the type. By setting AR markers on the tableware, the multi-joint robot 12 can pick up the tableware even if the position and posture of the tableware are not fixed. Using the input of multiple tableware detection cameras of the tableware information acquisition section 14, the Kalman filter is used to track each AR marker, thereby improving the accuracy of identifying the position and posture of the tableware. In addition, even when the tableware conveying section 10 is in motion, the multi-joint robot 12 can pick up the moving tableware.

[About the tableware handling department]

Although there are countless robot postures used by the multi-joint robot 12 to pick up tableware, in reality, the robot has limited postures that can be taken or postures that are easy to take. If the robot posture is not properly selected, the robot will stop because it cannot move or is too close to the movable limit. For example, if the robot posture takes a singular point, the robot may stop. Therefore, the multi-joint robot 12 movement ease index, i.e., operability, and proximity to the movable limit, is used to select a posture that is easy for the robot to move from a number of candidate postures to pick up tableware.

Usually, the movement of the multi-joint robot 12 is expected to move the shortest distance from the starting point to the end point, but when picking up tableware, if the shortest distance is moved, the joint may exceed the movable range. Therefore, the movement of the multi-joint robot 12 from the starting point to the end point is simulated in advance, so that the angle of a joint that may exceed the movable range is matched with the angle of the shortest distance to the end point before the movement starts. In this way, it is possible to prevent a specific joint from exceeding the movable range before reaching the end point.

When stacking tableware on the tableware loading part 13, if the height is not appropriate, there is a risk that the tableware will fall and make a loud noise, or the tableware may be damaged by pushing the tableware. Therefore, a force sensor is set between the robot arm of the multi-joint robot 12 and the robot hand 40 to detect the force of the tableware being pushed and prevent the tableware from falling. Compliance control is used to make the front end of the robot arm of the multi-joint robot 12 or the robot hand 40 have a virtual spring property, so that the tableware will not be damaged even if it is pushed. The force sensor detects the direction of stacking deviation and follows that direction.

In the case where the tableware is incorrectly identified and cannot be adsorbed, the tableware or the tableware conveying unit 10 will touch and push at a place other than the adsorption part 41, which may cause the multi-joint robot 12 or the tableware conveying unit 10 to malfunction, or may damage the tableware. Use compliance control, and use a force sensor to detect the direction of stack deviation, and when the applied force exceeds the set threshold of the hand front force, it is considered as a pick-up failure and abnormal processing is performed (skipping the tableware, or measuring the correct position and posture with the sensor again, etc.), and then the processing can continue.

When stacking tableware on the tableware placement part 13, if there is a position deviation on the side of the tableware placement part 13, or the position and posture of the previously placed tableware are deviated, it is possible that a force that pushes the tableware may cause the tableware to be damaged, or the tableware may bounce and fail to be placed. Therefore, by using compliant control, such failure rates can be reduced. In addition, because the direction of the deviation of the front end of the hand is known under compliant control, the deviation of the position and posture itself can be estimated and eliminated by moving the target value of the front end of the hand in that direction, that is, following the trend.

[About prediction]

When detecting AR markers, as a general characteristic of cameras, the accuracy of the camera's optical axis direction (depth direction) is relatively poor. If the position accuracy of the AR marker is relatively poor, the position where the robot picks up the tableware may deviate and there may be a situation where the pick-up fails. In addition, if the direction of the AR marker cannot be clearly distinguished from the camera or it is blocked, the tableware may not be detected. Therefore, multiple cameras are used to track the AR marker. The AR marker is tracked using a Kalman filter. By using a Kalman filter to mix and estimate the position and posture of the marker measured by multiple cameras, even if there is a moment when the AR marker cannot be measured by the camera, its position and posture can be tracked without missing the AR marker.

Using the Kalman filter, the following two improvements were made to achieve more accurate estimation:

(a) In the Kalman filter, the error size (error covariance matrix) can be set as a parameter. Since the error in the camera optical axis direction is known to be larger in principle, the error value in the camera optical axis direction is set larger to improve the estimation accuracy of position and posture.

(b) In the Kalman filter, other parameters can be estimated as state quantities, such as: (1) the size of the marker is different from the assumption; (2) the camera posture is slightly different from the assumption; (3) the direction of the conveyor is slightly different, etc. By using these state quantities, the position and posture of the marker can be estimated with better accuracy.

Regarding the camera configuration, it is known that the optical axis direction of the camera has poor measurement accuracy, so by setting the optical axis direction of the camera to as different directions as possible, the deviation that deteriorates the accuracy can be prevented. Therefore, the two cameras as the image acquisition unit are configured to sandwich the conveyor and the optical axis is directed from both sides toward the center of the conveyor.

When placing tableware into the tableware washing section, there are certain rules, so the position of the tableware when taking it out can also be predicted based on the rules. In addition, when reading the marking information when placing tableware into the dishwasher, the position of picking up the tableware can be predicted with high accuracy because the tableware arrangement on the conveyor is already known in advance.

In addition, the order of picking up tableware can be exemplified as follows: 1) Use multiple cameras to measure the mark and use a Kalman filter to estimate its position; 2) If the mark deviates from the camera's field of view, the position of the mark after the movement is further estimated based on the movement of the conveyor (the movement is directly measured by the encoder installed on the conveyor); 3) Based on the estimated position, the robot picks up the tableware that should be there. On the conveyor, if the tableware collides with other objects or the tableware is unstable on the conveyor, the tableware may still move during the period when the camera does not observe the mark. Therefore, the most accurate approach is to use the camera to continuously observe the mark until the robot picks up the tableware. As long as the tableware can always fall within the camera's field of view, the accuracy of measuring the position of the tableware is further improved. However, even if the dishes always fall within the field of view, the Kalman filter itself still needs to use multiple cameras to improve the observation accuracy. In addition, if the camera is brought close to the robot in order to make the field of view of the camera close to the robot's movement range, the robot may collide with the camera, so it is necessary to work on the overall structure or camera configuration.

[About learning]

When a robot is used to drop tableware, it can accept cases of failed pick-ups and modify the rules for dropping tableware at any time. To prevent the robot from making unreasonable postures, the rules for dropping tableware can be optimized through learning. For example, the method of dropping tableware suitable for picking, the method of dropping tableware with correct rules, or the method of dropping tableware in an orientation that is easy to pick up can be learned through a mechanism such as reinforcement learning to optimize the rules for dropping tableware.

Reinforcement learning is a method of setting rewards for the results of the robot's actions and making it obtain actions that maximize the rewards. As an example of rewards, a vacuum pressure sensor is used to determine the success or failure of tableware picking, and rewards are given when the tableware is picked up successfully. An effective approach is to use the observed tableware images as input, and obtain the trajectory correction amount (determination method) when the robot picks up in a way that maximizes the rewards, so as to perform reinforcement learning.

In addition, if the automatic tableware washing system of this embodiment is to be actually operated while reinforcement learning is being performed, a specific number of trials must be carried out. Therefore, during the period when reinforcement learning is in progress and the picking accuracy needs to be improved, in order to prevent damage or malfunction caused by collision between tableware and the robot arm, it is more effective to adopt the above-mentioned compliance control.

Due to the limitation of the robot's posture, it is sometimes impossible to move the front end of the robot arm in the shortest distance. In order to smoothly pick up the next tableware after picking up the current tableware, the following learning can also be performed, that is, the correction amount of the track of the front end of the robot arm of the multi-joint robot 12 is optimized, or which tableware is selected as the picking object is optimized. For example, if AI (Artificial Intelligence) that has completed learning is used, the robot can quickly pick up tableware with smooth movements like an experienced person from the initial introduction of the dishwasher. An effective approach is, for example, to add a correction value to the trajectory of the robot when picking, and optimize the correction value by strengthening learning in the direction of picking proficiency, or by supervised learning based on DNN (Deep Neural Networks), which also helps to increase the speed of the robot.

[Implementation form 2]

Referring to Figures 10 and 11, the automatic tableware washing system of the second embodiment of the present invention is described. In this embodiment, multiple variations of the markings used in the first embodiment are described.

[Variation 1]

In variation 1, a method for attaching a sticker printed with an AR mark is described. A sheet member is attached to the back (bottom) of the tableware by means of a bonding agent or an adhesive, and the sheet member is, for example but not specifically limited to any of the following: (a) a sheet member printed with an AR mark in black ink on a white-backed sheet member, (b) a sheet member printed with an AR mark in black ink on a transparent sheet member, (c) a sheet member printed with an AR mark in white ink on a black sheet member, or (d) a sheet member printed with an AR mark in white ink on a transparent sheet member.

Here, the sheet member and ink color are not limited to transparent, white or black, as long as the color combination of the AR mark can be read when the printed sheet member is attached to the back of the tableware. By adjusting the image recognition software, various color combinations can be coped with, but because the color of general tableware is mostly white or black, a combination of white background black ink or black text white ink is often used. In this case, the pattern of the mark used for white background tableware can be reversed from black to black and used for black background tableware. In addition, the sheet member (b) above is used for light-colored tableware such as white that can recognize black ink, and the sheet member (d) above is used for dark-colored tableware such as black that can recognize white ink.

As a waterproof sheet, it is desirable to use a sheet member with a waterproof film or waterproof coating on its surface and with high water resistance and heat resistance. In order to be able to wash and use repeatedly, it is also desirable to use an adhesive or adhesive for attaching the sheet member to the tableware with high water resistance and heat resistance. In addition, as long as the sheet member can be attached, the shape or material of the tableware can be arbitrary. In addition, the AR mark must be set on a flat part such as the bottom of the back of the tableware. Since the sheet member can be attached to the existing tableware later, there is no need to prepare special tableware.

It has been explained that the waterproof sheet or coating should be one with high water resistance and heat resistance, but this embodiment is not limited to this. Since the tableware is repeatedly washed and used with food, it is expected that the waterproof sheet or coating is also chemically and physically durable. That is, it is expected that the waterproof sheet or coating is resistant to various foods and detergents, such as alkali resistance and acid resistance, and is difficult to peel off even after repeated washing, and is also durable to friction during use or washing.

[Variation 2]

Using FIG. 10, variation 2 is described. Variation 2 is a method of engraving an AR mark 42 on the back of tableware by laser processing. Tableware used in variation 2 can be made of any material that can be grooved by laser processing, such as melamine. The shape of the tableware can be any shape as long as it has a flat portion only for setting the AR mark, such as tableware with a flat bottom on the back.

For the portion to be filled with ink corresponding to the AR mark on the back of the tableware, for example, a plurality of approximately parallel linear grooves with a shadow line spacing of about 0.3mm are processed by laser. Ink of a color different from the color of the tableware is filled in the groove. The interval of the shadow line is not particularly limited, and can be set to a narrower interval than, for example, about 0.3mm, or a wider interval. As the ink, it is desirable to use water-resistant and heat-resistant ones. Since the ink filled in the groove is retained in the groove and has durability, there is no need to set a coating, etc. However, in order to further improve durability, a waterproof film or protective coating with water resistance and heat resistance can also be set on the AR mark. Since the existing tableware can be laser processed later, there is no need to prepare special tableware.

FIG. 10A is an example of filling black ink into white-bottomed tableware. The size of the mark is not particularly limited, and can be set to about 20 mm square. In order to cope with smaller tableware, it can be about 15 mm square or smaller, and can also be about 30 mm square or larger. FIG. 10B is an example of filling white ink into black tableware. The color of the tableware and the color of the ink are not limited to black or white, as long as they are a combination of colors that can read the AR mark 42. By adjusting the image recognition software, it is possible to cope with various color combinations and various sizes of marks. In addition, in order to simplify the algorithm of image recognition and improve the recognition accuracy of position or posture, it is expected that the size of the AR mark is unified in advance.

[Variation 3]

Variant 3 is described using FIG. 11. Variant 3 is a method of directly printing an AR mark 42 on the back of tableware and then coating it. First, a method of setting an AR mark in the manufacturing step of melamine tableware is described. In the manufacturing step of melamine tableware, the AR mark is printed by a suitable ink, and a coating such as melamine coating is applied on its surface. The material of the coating is not limited to melamine, but in the case where the tableware material is made of melamine, for example, the coating on the surface of the AR mark is also set to a melamine coating. The ink is not particularly limited, and for example, ultraviolet color-developing ink, ultraviolet color-changing ink, ultraviolet curing ink, etc. can be used.

Although melamine tableware has been used as an example for explanation, the AR mark printing method of variation 3 is not limited to melamine tableware, but can also be applied to various tableware such as pottery, glass, and resin, that is, it is provided that appropriate coating corresponding to the material is implemented, such as resin coating. In addition, the AR mark can be printed on the existing tableware later. In this case, since the existing tableware can be laser processed later, it is not necessary to prepare special tableware. In addition, since the AR mark is coated, it has higher durability.

FIG11A is an example of printing black ink on tableware with a white background. Although the color of the black ink may be thinner due to the material of the tableware, the specification or state of the ink, or the specification or state of the coating, the AR mark can be recognized by adjusting the image recognition software even when the color of the black ink is thinner. FIG11B is an example of printing white ink on tableware with a black background. Although the color of the white ink may be slightly brown due to the material of the tableware, the specification or state of the ink, or the specification or state of the coating, the AR mark can still be recognized by adjusting the image recognition software. In addition, the combination of the tableware color and the ink color in Variation 3 is not limited to black or white. As long as the AR marker can be identified, it can be a combination of other colors, and by adjusting the image recognition software, AR markers of various color combinations can be identified.

[Variation 4]

In variation 4, a method of attaching a label with an AR mark to tableware is described. For example, a label with an AR mark printed on it in advance in a coin-like shape (thin round shape) is attached to the back of the tableware, such as the bottom, with an adhesive or adhesive, with the printed surface of the label as the front. On the printed surface of the label, it is desirable to provide a waterproof film or protective coating with water resistance and heat resistance. In addition, for the adhesive or adhesive used to attach the label to the tableware, it is also desirable to use an adhesive with water resistance and heat resistance. The specifications of the shape or material of the label in variation 4 are not limited to the coin shape, and can be any specifications as long as it has an AR mark. Since the label with the AR mark is attached to the existing tableware later, there is no need to prepare special tableware.

[Variation 5]

In variation 5, the types of markers and the positions of markers are described. In variations 1 to 4, the examples of using AR markers as markers are described, but in this embodiment, the markers are not limited to AR markers, and any markers can be used as long as they can identify at least one of the tableware type, tableware position, or tableware posture.

In addition, the marker can be placed on the back of the tableware, for example, at the center of the bottom, but the present embodiment is not limited thereto. For example, as long as the marker is located at a specific position according to the type of tableware, the position of picking up the tableware can be grasped by pre-registering the position in the image recognition software.

[Implementation form 3]

Referring to FIG. 12, the automatic tableware washing system of the third embodiment of the present invention is described. In FIG. 12, the same symbols as those described in FIGS. 1 to 11 are used, and their description is omitted. In embodiments 1 and 2, an example of the tableware information acquisition unit 14 using AR tags to identify tableware has been described, and in embodiment 2, an example of identifying tableware without AR tags is described.

Tableware is identified based on images from multiple tableware detection cameras serving as the tableware information acquisition unit 14, and a 3D point group (e.g., obtained from multiple camera images). Even without AR markers, the position and posture of the tableware can be identified. Although not particularly limited, a depth camera with stereoscopic vision capable of depth measurement, such as the Intel RealSense Depth Camera D400 series, can be used. It is equipped with an image sensor set (2 depth sensors, maximum resolution 1280×720) that can detect the difference between two images and perform real-time depth image processing, and can scale color data, and is equipped with an IR (InfraRed: infrared) projector, and operates with USB (Universal Serial Bus) power supply. For example, when observing a pure white wall, in order to obtain more accurate depth data, the active IR projector that illuminates the object projects a random dot matrix pattern of infrared rays, and the pattern is added to the apparent object, making it easier to obtain the correspondence between the two camera images.

As a method of obtaining a 3D point group, in addition to the above-mentioned stereo camera method, the following methods can also be used: 1) irradiating a pattern of laser light or LED (Light Emitting Diode) and photographing the image of the pattern to obtain a 3D point group (Structured Light); 2) irradiating a laser light or LED and measuring the time it takes to be reflected back (Time of Flight); and 3) exploring the focus position of the image to obtain depth information (focusing method), etc.

Furthermore, as shown in FIG12 , for example, if DNN is used in the image recognition of tableware, the region containing only the image or point group of the tableware can be cropped, thereby also improving the detection accuracy of the position and posture of the tableware. In FIG12 , the recognized tableware is displayed by the detection frame 50 , and its category is further displayed as the category display 51 .

In the embodiment 3, a method of identifying tableware using images from multiple tableware detection cameras and a 3D point group is illustrated, but the embodiment is not limited thereto, and may also be applied to methods such as detecting tableware using 2D images, controlling tableware identification by learning, detecting the bottom of tableware from images, and combining these methods. For example, the position and posture of tableware can be detected even based on 2D images, so even without AR markers, image recognition of tableware can be performed, and based on the image recognition, the robot hand 40 having the adsorption portion 41 of the multi-joint robot 12 adsorbs the bottom of the tableware and picks it up.

[Implementation form 4]

Referring to FIG. 13, the automatic tableware washing system of the embodiment 4 of the present invention is described. FIG. 13 is a top view of the tableware removal part of the embodiment 4. In FIG. 13, the same symbols are used for the same structures as those in FIGS. 1 to 12, and their descriptions are omitted. In the embodiment 4, an example of using the container storage robot 56 together with the multi-joint robot 12 is described.

In the case where a multi-joint robot 12, such as a 6-axis multi-joint robot, is used to take the tableware 11 from the tableware conveying unit 10 and store it in the container 57, it is difficult to stack the tableware horizontally and store it in the container 57 because the movable range of the multi-joint robot 12 is limited. Therefore, if one multi-joint robot 12 is used to realize the entire action of taking the tableware 11 from the tableware conveying unit 10 and storing it in the container 57, the device will be enlarged.

In FIG. 13 , the container storage robot 56 is used together with the multi-joint robot 12 to take the tableware 11 from the tableware conveying section 10 and store it in the container 57. The multi-joint robot 12 takes the tableware 11 from the tableware conveying section 10 and places the tableware 11 on the transfer table 55 in an upside-down state, that is, with the bottom facing upward. The tableware 11 placed on the transfer table 55 is stored in the container 57 by the container storage robot 56. The container storage robot 56 is not particularly limited, and a horizontal multi-joint robot, for example, can be used.

The multi-joint robot 12 and the container storage robot 56 have a mechanism for picking up the tableware 11. For example, the multi-joint robot 12 has a robot arm 40 with a suction portion 41, which can suck the bottom of the tableware 11 by the suction portion 41 to pick up the tableware 11. For example, the container storage robot 56 also has a suction portion 41 on the front end side of the robot arm, which can suck the bottom of the tableware 11 by the suction portion 41 to pick up the tableware 11.

The multi-joint robot 12 takes the tableware 11 conveyed by the tableware conveying unit 10 and places the tableware 11 on the transfer table 55. The container storage robot 56 takes the tableware 11 placed on the transfer table 55 and stores it in the container 57. In FIG13 , four containers are arranged within the range where the container storage robot 56 can store tableware, but the number of containers is not limited thereto, and one or more containers are arranged within the range where the container storage robot 56 can store tableware. The container storage robot 56 can store the tableware 11 in a specific container according to the type of tableware 11. For example, one type of tableware 11 can be stored in one container.

The multi-joint robot 12 and the container storage robot 56 are located at the periphery of the transfer platform 55, where their movable areas overlap. Therefore, the multi-joint robot 12 and the container storage robot 56 are coordinated and controlled in a manner that they do not interfere with each other. In this embodiment, the functions of the actions of the multi-joint robot 12 and the container storage robot 56 are clearly distinguished, and the circumstances that interfere with their actions are limited. Therefore, for the coordinated control of this embodiment, simple interlocking control, etc., which does not necessarily require complex control, can also be applied, and various control methods can be applied as the coordinated control.

When the tableware 11 is stored in the container 57, if the shape of the tableware, that is, the width when it is stacked flat (for example, the radius for round tableware, the length of one side (especially the long side) for square tableware, etc.) is not accurately obtained in advance, there is a risk that the tableware 11 will touch various parts of the automatic tableware washing system and be damaged. If the shape of the tableware 11 cannot be accurately obtained, for example, when the size of the tableware is estimated by image recognition, an additional image acquisition unit or a computer for image processing must be required, which increases the cost.

Therefore, a partition member of a specific shape is set at a specific position of the transfer table 55. When the multi-joint robot 12 unloads the tableware 11 on the transfer table 55 in a horizontal position, the side of the tableware is moved horizontally until it is pressed against the partition member, and the position of the partition member contacted is detected. Various sensors can be used for the detection, such as a force sensor, a vibration sensor, and an acceleration sensor set at the front end of the robot arm of the multi-joint robot 12 or the robot hand 40, or a load current sensor of the driving motor of the multi-joint robot 12, an image acquisition unit, etc. The position of the partitioning member is known in advance, and since the trajectory of the multi-joint robot 12 can also be calculated, the shape of the tableware 11 picked up by the multi-joint robot 12 can be calculated based on the position where it touches the partitioning member. In addition, if the shape of the tableware 11 can be calculated in this way, even if the size of the tableware 11 is not registered, the tableware 11 can be automatically stored in the container 57.

[Implementation form 5]

Referring to FIG. 14, the automatic tableware washing system of the embodiment 5 of the present invention is described. FIG. 14A is a top view of the automatic tableware washing system, and FIG. 14B is a three-dimensional view of the automatic tableware washing system. In FIG. 14, the same symbols are used for the same structures as in FIGS. 1 to 13, and their descriptions are omitted. In embodiment 5, as in embodiment 4, an example of using a container storage robot 56 together with a multi-joint robot 12 is described.

If the multi-joint robot 12 is arranged beside the tableware conveying part 10 (refer to Figure 13), the manipulator 40 of the multi-joint robot 12 will have difficulty reaching the end peripheral area of the tableware conveying part 10 on the opposite side of the multi-joint robot 12, resulting in a narrow range of postures that can be taken. As a result, if the multi-joint robot 12 is arranged beside the tableware conveying part 10, it may be difficult to pick up the tableware 11 due to the conveying area of the tableware conveying part 10.

Furthermore, if the multi-joint robot 12 is placed beside the tableware conveying unit 10, the overall system will become larger and the space efficiency will be reduced.

The automatic tableware washing system of this embodiment is equipped with: a cross-conveying unit stand 70, which is used to place a multi-joint robot 12 at a position above the tableware conveying unit 10 along the width direction of the tableware conveying unit 10.

The cross-transportation part stand 70 is a door-shaped stand, and the multi-joint robot 12 is fixed to the wall surface on the upper part. In this embodiment, the multi-joint robot 12 is fixed to the cross-transportation part stand 70 in such a way that the movable range of the multi-joint robot 12 is a range symmetrical to the center of the width direction of the tableware transport part 10.

In the automatic tableware washing system of the present embodiment 5, the multi-joint robot 12 is fixed to the cross-conveying part frame 70 in such a way that the movable range of the multi-joint robot 12 is a range symmetrical to the center in the width direction of the tableware conveying part 10, so that the multi-joint robot 12 can pick up tableware 11 in a wider range in the width direction of the tableware conveying part 10 and can easily adopt various postures.

In addition, in the automatic tableware washing system of the present embodiment 5, since the multi-joint robot 12 is arranged near the top of the exit of the tableware conveying part 10 which has not been effectively utilized in the past, the overall system can be miniaturized and space-saving such as not occupying the passage can be achieved.

[Implementation form 6]

Referring to FIG. 15 , the automatic tableware washing system of the embodiment 6 of the present invention is described. FIG. 15 is a diagram showing the periphery of the transfer station 55 of the automatic tableware washing system. In FIG. 15 , the same symbols are used for the same structures as in FIGS. 1 to 14 and their descriptions are omitted.

If the multi-joint robot 12 identifies the type of tableware 11 and the height of the tableware 11 when it takes the tableware 11 and places it upside down on the delivery table 55, the tableware 11 can be placed in accordance with the height of the delivery table 55. However, if the type of tableware 11 is incorrectly identified, the tableware 11 may be pushed against the delivery table 55 or fall from the top of the delivery table 55.

In this case, when the type of tableware 11 is incorrectly identified, the container storage robot 56 may be unable to operate due to damage to the tableware 11 or excessive position deviation of the tableware 11 on the transfer table 55.

This problem can be solved by reducing the speed of the multi-joint robot 12 and improving the accuracy of the recognition of the tableware 11, but the processing performance of the tableware 11 cannot be fully utilized.

The automatic tableware washing system of this embodiment is equipped with: a detector 80, which detects whether the tableware 11 passes through a horizontal plane separated from a specific interval above the transfer table 55.

The detector 80 is implemented by a transmission type sensor, for example. When a transmission type sensor is used as the detector 80, the detector 80 projects a straight line laser onto a horizontal plane several centimeters above the transfer table 55 to detect the laser passing through the tableware 11.

When the multi-joint robot 12 takes the tableware 11 and places it upside down on the delivery table 55, the automatic tableware washing system can detect the timing of the tableware 11 passing through the horizontal plane separated from the top of the delivery table 55 by the detector 80 even if the type of the tableware 11 or the height of the tableware 11 is not recognized, so as to decelerate the multi-joint robot 12 and estimate the height of the tableware 11, thereby placing the tableware 11 in accordance with the height of the delivery table 55.

In the automatic tableware washing system of this embodiment 6, since the detector 80 detects the situation that the tableware 11 passes through the horizontal plane separated from the top of the transfer table 55 by a specific interval, it is not necessary to rely on the image recognition of the tableware 11, but to place the tableware 11 in accordance with the height of the transfer table 55, so that the tableware 11 can be processed quickly and accurately.

Several embodiments of the present invention have been described above, but these embodiments are examples of automatic tableware washing systems that embody the technical ideas of the present invention. The present invention is not limited to these embodiments, but can be equally applied to other embodiments, and can also omit, add, change a part of these embodiments, or combine the aspects of each embodiment.

S1~S5: Steps

Claims (18)

A tableware automatic washing system comprises: a tableware washing section, which washes tableware put in from a tableware feeding section for putting in tableware; a tableware conveying section, which conveys tableware washed in the tableware washing section; a tableware information acquisition section, which comprises a plurality of image acquisition sections, which recognizes tableware conveyed by the tableware conveying section and acquires information of the tableware; a tableware loading section, which loads the tableware conveyed by the tableware conveying section; and a tableware taking section, which takes tableware from the tableware conveying section and places the tableware on the tableware loading section; and the system is characterized in that: the information acquired by the tableware information acquisition section includes , including at least one of the tableware type, tableware size, tableware shape, tableware color, tableware position, tableware posture, tableware orientation, a mark attached to the tableware, or a code attached to the tableware; once the tableware information acquisition unit identifies the tableware being transported by the tableware transport unit, the tableware picking unit mixes the information of the identified tableware measured by the plurality of image acquisition units and continuously tracks the position and posture of the identified tableware, and picks up the tableware using the information including the position and posture of the tracked tableware while the tableware is being transported by the tableware transport unit. A tableware automatic washing system comprises: a tableware washing section, which washes tableware put in from a tableware feeding section; a tableware conveying section, which conveys tableware washed in the tableware washing section; a tableware information acquisition section, which identifies tableware conveyed by the tableware conveying section and acquires information of the tableware; a tableware loading section, which loads the tableware conveyed by the tableware conveying section; and a tableware picking section, which uses a multi-joint robot to pick up tableware from the tableware conveying section and load the tableware into the tableware loading section; and the system is characterized in that: the information acquired by the tableware information acquisition section includes tableware type At least one of the following: type, size, shape, color, position, posture, orientation, mark or code attached to the tableware; once the tableware information acquisition unit identifies the tableware being transported by the tableware transport unit, the tableware picking unit simulates the movable range of the multi-joint robot based on the information including the position and posture of the tableware acquired by the tableware information acquisition unit. If a joint has the possibility of exceeding the movable range, the joint is first adjusted to an angle close to the target value, and then the multi-joint robot is driven and controlled to the tableware picking position. A tableware automatic washing system comprises: a tableware washing section, which washes tableware put in from a tableware feeding section for putting in tableware; a tableware conveying section, which conveys tableware washed in the tableware washing section; a tableware information acquisition section, which identifies tableware conveyed by the tableware conveying section and acquires information of the tableware; a tableware loading section, which loads the tableware conveyed by the tableware conveying section; and a tableware picking section, which uses a multi-joint robot including a force sensor to pick up tableware from the tableware conveying section and load the tableware on the tableware loading section; and the tableware information acquisition section is characterized in that the information acquired by the tableware information acquisition section includes tableware information. At least one of the type, size, shape, color, position, posture, orientation, or code of the tableware; once the tableware information acquisition unit identifies the tableware being transported by the tableware transport unit, the tableware picking unit picks up the tableware in accordance with the state of the tableware based on the information acquired by the tableware information acquisition unit and the information of the pushed force detected by the force sensor, and controls the tableware in accordance with the state of the tableware placement unit in a manner that imitates the position and posture of the picked tableware, thereby placing the picked tableware on the tableware placement unit. For example, in the automatic tableware washing system of claim 2 or 3, the tableware information acquisition unit includes one or more image acquisition units. An automatic tableware washing system as claimed in any one of claims 1 to 3, wherein the tableware information acquisition unit includes a camera unit, and the tableware information can be acquired from the image acquired by the camera unit. An automatic tableware washing system as claimed in any one of claims 1 to 3, wherein the tableware information acquisition unit picks up the tableware while the tableware is being transported by the tableware transport unit. In the automatic tableware washing system of any one of claims 1 to 3, if the information acquired by the tableware information acquisition unit is a mark, the mark is at least one of the following: (1) affixed with a sticker; (2) formed by laser processing; (3) given by printing; (4) coated; or (5) affixed to another component on which the mark is printed. An automatic tableware washing system as claimed in any one of claims 1 to 3, wherein tableware is identified without being marked. For example, in the automatic tableware washing system of any one of claim items 1 to 3, the tableware information acquisition unit or the tableware picking unit predicts the location of the tableware. An automatic tableware washing system as claimed in any one of claims 1 to 3, wherein the tableware picking portion has an elastic component. An automatic tableware washing system as claimed in any one of claims 1 to 3, wherein at least one of the tableware input unit, the tableware information acquisition unit, or the tableware retrieval unit is controlled by machine learning. An automatic tableware washing system as claimed in any one of claims 1 to 3, wherein the tableware picking-up unit includes a multi-joint robot that picks up tableware by means of a suction unit; and the suction unit includes a plurality of suction units with different suction directions. An automatic tableware washing system as claimed in any one of claims 1 to 3, wherein a container storage robot is further used to store the tableware in a container. A tableware automatic washing method includes: a tableware washing step, which washes the tableware put in; a tableware conveying step, which conveys the tableware washed in the tableware washing step; a tableware information acquisition step, which identifies the tableware conveyed in the tableware conveying step by using information from a plurality of image acquisition units and obtains information about the tableware; and a tableware picking step, which picks up the tableware conveyed in the tableware conveying step and places the picked up tableware; and the characteristic is that the information obtained in the tableware information acquisition step includes the tableware type, tableware size, tableware shape, tableware color, tableware position, tableware posture, tableware orientation, a mark attached to the tableware, or a code attached to the tableware; in the tableware information acquisition step, if the tableware being transported in the tableware transport step is identified, in the tableware picking step, the information of the identified tableware measured by the plurality of image acquisition units is mixed and the position and posture of the identified tableware are continuously tracked, and the tableware is picked up using the information including the position and posture of the tracked tableware while the tableware is being transported by the tableware transport unit. A tableware automatic washing method comprises: a tableware washing step, which washes the tableware put in; a tableware conveying step, which conveys the tableware washed in the tableware washing step; a tableware information obtaining step, which identifies the tableware conveyed in the tableware conveying step and obtains the information of the tableware; and a tableware picking step, which uses a multi-joint robot to pick up the tableware conveyed in the tableware conveying step and place the picked tableware; and the method is characterized in that: the information obtained in the tableware information obtaining step includes the tableware type, tableware size, tableware shape, tableware color, tableware At least one of the position, tableware posture, tableware orientation, mark attached to the tableware, or code attached to the tableware; in the tableware information acquisition step, if the tableware being transported in the tableware transport step is identified, then in the tableware picking step, based on the information including the position and posture of the tableware acquired by the tableware information acquisition unit, the movable range of the multi-joint robot is simulated, and in the case of a joint that may exceed the movable range, the joint is first made to meet the angle close to the target value, and then the multi-joint robot is driven and controlled to the tableware picking position. A tableware automatic washing method comprises: a tableware washing step, in which the tableware put into the tableware is washed; a tableware conveying step, in which the tableware washed in the tableware washing step is conveyed; a tableware information obtaining step, in which the tableware conveyed in the tableware conveying step is identified and the information of the tableware is obtained; and a tableware picking step, in which a multi-joint robot including a force sensor is used to pick up the tableware conveyed in the tableware conveying step and place the picked up tableware on a tableware placing part; and the tableware automatic washing method comprises: the information obtained in the tableware information obtaining step includes the tableware type, tableware size, tableware shape, tableware At least one of the color, tableware position, tableware posture, tableware orientation, a mark attached to the tableware, or a code attached to the tableware; in the tableware information acquisition step, if the tableware being transported in the tableware transport step is identified, then in the tableware picking step, based on the information acquired by the tableware information acquisition unit and the information of the pushed force detected by the force sensor, the tableware is picked up in accordance with the state of the tableware, and the tableware placement unit is controlled in accordance with the state of the tableware placement unit to imitate the position and posture of the picked tableware, thereby placing the picked tableware on the tableware placement unit. An automatic tableware washing program is characterized in that each step of the automatic tableware washing method as in any one of claims 14 to 16 can be executed by a computer unit. A storage medium that stores the automatic tableware washing program as described in claim 17.