JP6836606B2 - Robot control and system - Google Patents

Description

The present invention relates to a robot control device and a system.

Conventionally, there is known a robot that takes out an article such as a bolt having a metallic luster by a robot hand provided at a tip portion (see, for example, Patent Document 1). As described above, conventionally, a robot hand has been used for taking out an article whose shape does not easily change, thereby improving work efficiency.

Japanese Unexamined Patent Publication No. 2015-182184

In recent years, the realization of the work of taking out articles whose shape easily changes, such as food, by a robot hand has been studied. For example, attempts are being made to take out foods such as cream puffs, breads, and cakes by robot hands. Silicone or rubber robot fingers may be used to grip such articles, but such fingers are often considerably harder than cream puffs, pancakes, and the like. Therefore, if the gripping interval or the gripping force, which is the spacing between the robot fingers in the gripping state, is not appropriate, the article is easily deformed or damaged.

As one of the solutions, it is conceivable to set the gripping interval appropriately. However, articles such as cream puffs, breads, and cakes often have irregular shapes. Therefore, even if the gripping interval is set appropriately when starting the manufacturing on that day, the state of the article such as hardness and shape changes slightly during the manufacturing, and the gripping interval changes due to the change. Often not appropriate.

Further, as described above, it is necessary to grip the article as gently as possible with the robot finger. However, such articles are often heavier than their hardness. Therefore, the range of gripping intervals or gripping forces that can appropriately grip the article is narrow. In order to solve such a problem, it is conceivable to provide a tactile sensor at the tip of each robot finger. However, the structure and control of the robot hand become complicated, which leads to an increase in cost. Furthermore, there is a concern that the sensor may fail due to the components contained in the article to be handled. Further, if the work efficiency is lowered due to the complicated control, the merit of using the robot hand is reduced or eliminated.

In view of the above circumstances, in order to appropriately grip a soft article with a robot hand, a robot control device capable of appropriately changing the gripping interval or gripping force of a plurality of gripping portions such as robot fingers according to the state of the article. And the provision of a system is desired.

The robot control device of the first aspect of the present disclosure is a robot control device that controls a robot hand that grips an article by a plurality of gripping portions, and is based on an image obtained by a visual sensor for detecting the article. A gripping interval that is an interval between the plurality of gripping portions in the gripping state or a gripping force that changes the gripping force of the plurality of gripping portions in the gripping state according to the size information acquiring means for obtaining the size information of the article and the size information. It is provided with an adjusting means.
In this way, since the gripping interval or the gripping force changes according to the size information, a soft article having a non-constant shape can be appropriately gripped by the plurality of gripping portions.

In the above aspect, preferably, the size information is the area value of the region detected as the article in the image.
Here, when the article is a food such as cream puff, bread, cake, etc., the shape, hardness, etc. of the article are not constant. In addition, an article in which a highly viscous liquid such as cream is enclosed often tends to swell the other part of the outer peripheral surface when a part of the outer peripheral surface is dented. In this case, it can be said that the change in the area of the article in the plan view when a part of the outer peripheral surface is recessed is relatively small.
In this aspect, the gripping interval or gripping force changes according to the area value of the region detected as an article in the image, which is advantageous for appropriately gripping the article whose shape is not constant.

In the above aspect, preferably, the size information is dimensional data of a region detected as the article in the image.
The dimensional data is the maximum diameter, average diameter, etc. of the article in the image. Even in this case, it is possible to appropriately grip an article whose shape is not constant.

In the above aspect, it is preferable that the dimensional data is the perimeter data of the region detected as the article in the image.
If the dimensional data is the maximum diameter, average diameter, etc. of the article in the image, there may be variations depending on the direction in which the dimensions are measured, but if the circumference data is used as the dimensional data, the above variations may occur. Can be reduced.

In the above aspect, preferably, the robot hand has a hand control device that changes the gripping interval or the gripping force to a plurality of levels, and a table or a table that associates the value of the size information with the plurality of levels. The hand control device includes a storage unit in which an expression is stored, and the hand control device sets a control signal for the gripping adjustment means to set a level corresponding to the size information among the plurality of levels by using the table or the expression. Send to.

A robot hand that changes the gripping interval or gripping force of a plurality of gripping portions to a plurality of levels is easy to control and simplifies the structure. Therefore, such a robot hand is often used by a manufacturing company of an article to be taken out.
In the above aspect, a control signal for setting the level according to the size information is transmitted to the hand control device using a table or formula that associates the gripping interval or the gripping force with the value of the size information. Therefore, the control on the robot control device side can be simplified, and the time and effort for adjusting the gripping interval or the gripping force can be reduced.

In the above aspect, preferably, the gripping adjusting means sets the value of the size information corresponding to each of the plurality of levels based on the distribution of the size information of each of the plurality of articles in the single image or the plurality of images. Change.
In this way, the gripping adjusting means changes the value of the size information corresponding to each level of the gripping interval or the gripping force based on the distribution of the size information of the article in the image, thereby reducing the error in taking out the article and reducing the article. It is advantageous for reducing damage, deformation, and contact marks that occur in.

A second aspect of the present disclosure is a robot control device that controls a robot hand that grips an article by a plurality of gripping portions, a size information acquisition means for obtaining size information of the article, and gripping according to the size information. When the article is gripped by the gripping interval, which is the interval between the plurality of gripping portions in the state, or the gripping adjusting means for changing the gripping force by the plurality of gripping portions in the gripping state, the size information, and the robot hand. A learning unit that performs learning for improving the gripping state based on the information on the gripping state is provided.

A third aspect of the present disclosure is a management system, comprising a robot control device that controls a robot hand that grips an article, and a higher-level control system that can communicate with the robot control device. A learning unit that performs learning to improve the gripping state based on the size information of the article and information on the gripping state when the article is gripped by the robot hand, and the upper control system that obtains the result of the learning. The higher-level control system has an output unit that outputs to the robot control device, and stores the learning result received from the robot control device.

In the above aspect, in order to appropriately grip a soft article with a robot hand, the gripping interval or gripping force of a plurality of gripping portions such as robot fingers can be appropriately changed according to the state of the article.

It is an overall block diagram of the robot system of one Embodiment. It is a front view of the robot hand used for the robot system of this embodiment. It is operation explanatory drawing of the robot hand of this embodiment. It is a block diagram of the robot control device of this embodiment. It is a flowchart which shows the operation of the control part of the robot control device of this embodiment. It is a flowchart which shows the process of the control part of the robot control device of this embodiment. This is an example of a table stored in the storage unit of the robot control device of the present embodiment. This is an example of a table stored in the storage unit of the robot control device of the present embodiment. This is an example of a table stored in the storage unit of the robot control device of the present embodiment. It is a block diagram of the management system which has the robot control device of this embodiment. It is a block diagram of the management system which has the robot control device of this embodiment. It is a block diagram of the system which has the robot control device of this embodiment. It is a block diagram of the system which has the robot control device of this embodiment. It is a block diagram of the system which has the robot control device of this embodiment.

The robot control device 20 and the robot system according to the embodiment will be described below with reference to the drawings.
In the robot system of the present embodiment, as shown in FIG. 1, an article A conveyed by a transfer device 2 such as a conveyor is taken out from the transfer device 2 by a robot hand 30 attached to the robot 10. In the present embodiment, the article A is a soft product such as cream puff, pancake, or cake whose shape is not constant.

In the present embodiment, the robot 10 is a parallel link robot, and the robot 10 has a main body portion 11, a movable plate 12 arranged below the main body portion 11, and three links connecting the main body portion 11 and the movable plate 12. It has a portion 13 and a wrist portion 14 attached to the movable plate 12. The robot 10 may be a vertical articulated robot, a horizontal articulated robot (SCARA robot), or the like.

As shown in FIG. 1, the main body portion 11 has a base member 11a extending in a substantially horizontal direction and three drive portions 11b fixed to the lower surface of the base member 11a. A servomotor 15 (FIG. 4) with a speed reducer is housed in each drive unit 11b, and one end of the drive link 13a of the link unit 13 is fixed to the output shaft of each drive unit 11b. Further, a servomotor 16 (FIGS. 1 and 4) for rotating the wrist flange 14a of the wrist portion 14 around the vertical axis is attached to one of the plurality of link portions 13.

As shown in FIG. 1, each link portion 13 has a drive link 13a and a driven link 13b whose one end is connected to the other end of the drive link 13a. The other end of the drive link 13a and one end of the driven link 13b are connected by using a spherical bearing. Further, the other end of each driven link 13b is connected to the movable plate 12 by using a spherical bearing.

The base member 11a of the main body 11 is supported by, for example, a frame (not shown), whereby the robot 10 is arranged above the transfer device 2.
The robot hand 30 is supported by the wrist flange 14a. That is, the robot hand 30 rotates around the vertical axis together with the wrist flange 14a. The wrist portion 14 may be a part of the robot hand 30.

As shown in FIGS. 1, 2, and the like, the robot hand 30 has a plurality of grip portions 31, and each grip portion 31 is formed of a material having rubber-like elasticity. As a material having rubber-like elasticity, rubber, silicon, or the like can be used. In the present embodiment, each grip portion 31 is a robot finger having an internal space 31a, and each grip portion 31 bends according to the pressure of air supplied to each internal space 31a (FIG. 3). As the robot hand 30, for example, M4FC, M5FC, etc. manufactured by Soft Robotics can be used.

As shown in FIG. 1, each grip portion 31 of the robot hand 30 is connected to the hand control device 40 via the air supply pipe 32, and the hand control device 40 is connected to an air supply source (not shown). .. The hand control device 40 has a pressure adjusting valve, and the air pressure supplied to each grip portion 31 of the robot hand 30 is adjusted to a plurality of levels by the pressure adjusting valve. In the present embodiment, as an example, the hand control device 40 can adjust the air pressure supplied to each grip portion 31 in five stages from level 1 to level 5. The number of air pressure levels is not limited to five.

When air is supplied to the plurality of grips 31 by the hand control device 40, as shown in FIG. 3, the tips of the plurality of grips 31 approach each other, and the plurality of grips 31 grip the article A. It becomes the gripping state of. When the internal space 31a of the plurality of gripping portions 31 becomes atmospheric pressure or negative pressure by the hand control device 40, the plurality of gripping portions 31 are in a non-grasping state for releasing the article A. In FIG. 3, the article A is deformed undesirably by applying an air pressure higher than an appropriate pressure to each grip portion 31.

When the air pressure supplied from the hand control device 40 to the plurality of grips 31 is set to level 1, there are more than one when the air pressure supplied from the hand control device 40 to the plurality of grips 31 is set to level 5. The distance between the tip portions of the grip portions 31 of the above is reduced, and the gripping force of the article A by the plurality of grip portions 31 is increased. The air pressures at levels 2 to 4 stepwise change the distance between the tips of the plurality of grips 31 between levels 1 and 5. That is, the air pressures of levels 1 to 5 change the gripping interval, which is the distance between the tips of the plurality of gripping portions 31, to a plurality of levels, and the gripping force of the plurality of gripping portions 31 also changes to a plurality of levels.

A visual sensor 50 is provided above the transport device 2. The visual sensor 50 may be any as long as it can obtain an image of the article A on the transport device 2, and a two-dimensional camera, a three-dimensional camera, a three-dimensional sensor or the like is used as the visual sensor 50. The images captured by the visual sensor 50 are sequentially transmitted to the robot control device 20.

As shown in FIG. 4, the robot control device 20 includes a control unit 21 having a processor and the like, a display device 22, a storage unit 23 having a non-volatile storage, a ROM, a RAM, and the like, a keyboard, a touch panel, an operation panel, and the like. It is provided with an input device 24, a transmission / reception unit 25 for transmitting / receiving signals, a servo controller 26 connected to each servomotor 15, and a servo controller 27 connected to the servomotor 16. .. The input device 24 and the transmission / reception unit 25 function as input units.

The system program 23a is stored in the storage unit 23, and the system program 23a has a basic function of the robot control device 20. Further, the storage unit 23 stores an retrieval program 23b, a size information acquisition program (size information acquisition means) 23c, and a grip adjustment program (grip adjustment means) 23d. The control unit 21 controls the servomotors 15 and 16 based on the extraction program 23b, the size information acquisition program 23c, and the grip adjustment program 23d, and sends a control signal for adjusting the grip by the robot hand 30 to the hand control device 40. Send to.

Hereinafter, the processing of the control unit 21 when the robot 10 takes out the article A on the transport device 2 by the robot hand 30 will be described with reference to the flowcharts of FIGS. 5 and 6. The following control is performed in a state where the article A is being transported by the transport device 2.

When the control unit 21 receives the start signal from the input device 24 or the transmission / reception unit 25 (step S1-1), the control unit 21 takes out the article A in the image while performing well-known image processing on the image transmitted from the visual sensor 50. (Step S1-2).

An example of the extraction process in step S1-2 is illustrated in FIG. In the example of FIG. 6, first, the control unit 21 starts the position detection of the article A in the image subjected to the well-known image processing based on the extraction program 23b (step S2-1). Further, the control unit 21 calculates the moving speed of the transport device 2 based on the position of the same article A in a plurality of images based on the extraction program 23b, and position tracking (tracking) of the article A based on the calculated moving speed. Is started (step S2-2).

Then, the control unit 21 obtains the area information (size information) of the article A whose position is being tracked based on the size information acquisition program 23c by using the image processed image (step S2-3). .. When there are a plurality of articles A being tracked, the control unit 21 obtains area information of each of the plurality of articles A. In one example, the image processing is a binarization process, and the area information of the article A is the area value (size information) of the area extracted by the binarization process.

Subsequently, the control unit 21 determines whether the air pressure supplied to the plurality of grip units 31 should be one of level 1 to level 5 based on the grip adjustment program 23d (step S2-4).

In step S2-4, for example, the control unit 21 determines the level of air pressure corresponding to the area value of the article A by using the table 23e as shown in FIG. 7 stored in the storage unit 23. In the table 23e, a plurality of area ranges are set, and the air pressures of level 1 to level 5 correspond to the plurality of area ranges, respectively. The control unit 21 uses the table 23e stored in the storage unit 23 to determine the level of air pressure corresponding to the area value of each article A being tracked. In the table 23e, the meanings of 25 to 27 are 25 or more and less than 27.

Here, the article A such as cream puff is softer than the grip portion 31 of the robot hand 30. Although the grip portion 31 is made of a flexible material, it is considerably harder than the article A. Therefore, it is desirable that the air pressure determined in step S2-4 is an air pressure that does not cause damage, deformation, contact marks, or the like to the article A by the plurality of gripping portions 31. That is, the air pressure that greatly deforms the article A as shown in FIG. 3 is not appropriate. On the other hand, the article A such as cream puff is heavier than its hardness. Therefore, it is preferable that each area range of the table 23e is narrow in order to lift the article A without breaking it by the plurality of grips 31.

Subsequently, the control unit 21 transmits a control signal for taking out the tracked article A to the servo controllers 26 and 27 based on the take-out program 23b (step S2-5). Further, the control unit 21 transmits a control signal for supplying the air pressure of the level determined for the article A to each grip unit 31 based on the grip adjustment program 23d to the hand control device 40 (step S2-6). ). As a result, the article A is taken out from the transport device 2.

Based on the take-out program 23b, the control unit 21 transmits a control signal for placing the gripped article A on the conveyor 3 which is the transfer destination to the servo controllers 26 and 27 (step S2-7). .. Then, the control unit 21 repeats steps S2-3 to S2-7 until the end signal is received from the input device 24 or the transmission / reception unit 25 (step S2-8).

Instead of the table 23e, an expression for determining the air pressure level from the area value of the article A may be stored in the storage unit 23. In this case, in step S2-4, the control unit 21 uses the equation stored in the storage unit 23 to correspond to the area value of each article A being tracked in the air pressure levels 1 to 5. Determine the level of pressure.
It is also possible to determine the air pressure level corresponding to the area value of each article A being tracked among the air pressure levels 1 to 5 by using other well-known means, methods and the like.

As described above, in the present embodiment, the area value of the article A is obtained as the size information of the article A based on the image obtained by the visual sensor 50 for detecting the article A. Then, the control unit 21 changes the gripping interval, which is the interval between the plurality of gripping portions 31 in the gripping state, or the gripping force by the plurality of gripping portions 31 in the gripping state, depending on the area value.
In this way, since the control unit 21 changes the gripping interval or the gripping force according to the size information, the soft article A whose shape is not constant can be appropriately gripped by the plurality of gripping parts 31.

Here, the shape, hardness, and the like of the article A, which is a food product such as cream puff, bread, and cake, are not constant. Further, the article A in which a highly viscous liquid such as cream is sealed has a tendency to swell the other part of the outer peripheral surface when a part of the outer peripheral surface is dented. In this case, it can be said that the change in the area of the article A in the plan view when a part of the outer peripheral surface is recessed is relatively small.
In the present embodiment, the control unit 21 changes the gripping interval or the gripping force according to the area value of the region detected as the article A in the image, which appropriately grips the article A whose shape is not constant. It is advantageous to do.

The control unit 21 can also change the gripping interval or the gripping force according to the dimensional data of the region detected as the article A in the image. The dimensional data is the maximum diameter, average diameter, etc. of the article A in the image. Even in this case, it is possible to appropriately grip the article A whose shape is not constant, but there is a possibility that variations may occur depending on the direction in which the dimensions are measured. On the other hand, when the area value is used, the possibility of the above variation can be reduced. If the circumference data is used as the dimensional data, the possibility of the above variation can be reduced.

In the present embodiment, the robot hand 30 has a hand control device 40 that changes the gripping interval or gripping force to a plurality of levels 1 to 5, and includes an area value that is size information and the plurality of levels 1 to 5. The table 23e or the storage unit 23 in which the formula is stored is provided, and the control unit 21 uses the table 23e or the formula according to the area value of the article A being tracked among the plurality of levels 1 to 5. A control signal for setting the level is transmitted to the hand control device 40.

The robot hand 30 that changes the gripping interval or gripping force of the plurality of gripping portions 31 to a plurality of levels is easy to control, and the structure is also easy to simplify. Therefore, such a robot hand 30 is often used by a manufacturing company of the article A to be taken out.
In the present embodiment, a control signal for setting the level according to the area value is transmitted to the hand control device 40 by using the table 23e or the formula for associating the gripping interval or the gripping force with the area value. Therefore, the control on the robot control device 20 side can be simplified, and the time and effort for adjusting the gripping interval or the gripping force can be reduced.

As described above, the shape, hardness, etc. of the article A, for example, cream puff, bread, cake, etc., are not constant, and the shape, hardness, etc. change daily. Therefore, it is preferable that the gripping interval or the gripping force is adjusted daily.
For this purpose, a plurality of types of tables may be stored in the storage unit 23. For example, as shown in FIG. 8, a table 23f in which the range of the area values corresponding to the levels 1 to 5 is different from that of the table 23e in FIG. 7 may be further stored. The table 23f of FIG. 8 is used, for example, when the article A is harder than usual.

On days when the variation in the shape of the article A is small, the holding pressures of a plurality of levels of the pressure regulating valve of the hand control device 40 may be brought close to each other, and the table 23g of FIG. 9 stored in the storage unit 23 may be used. The adjustment is performed when the variation width of the article A is small.
Instead of the plurality of tables 23e, 23f, 23g, a plurality of expressions having different area values corresponding to the levels 1 to 5 may be stored in the storage unit 23.

Further, the control unit 21 may select and use one of the tables 23e and 23f based on the distribution of the area values of the plurality of articles A obtained based on the single or a plurality of images, and similarly, the storage One of the plurality of expressions stored in the unit 23 may be selected and used.
In this way, when the control unit 21 changes the area value corresponding to each level 1 to 5 based on the distribution of the area value of the article A in the image, the error in taking out the article A is reduced and the damage caused to the article A is caused. It is advantageous for reducing deformation, contact marks, and the like.

It is also possible to further provide an observation imaging device that images the work area of the robot 10 in the transport device 2 or the downstream side thereof. In this case, the article A that could not be taken out by the robot 10 is detected by the imaging device. When the number of such articles A is greater than the threshold value, the control unit 21 changes the range of the area value corresponding to the gripping interval or gripping force of each level 1 to 5. For example, the reference used in step S2-4 is changed from table 23f to table 23e. When this configuration is used, the labor of the adjustment work is reduced and the production efficiency of the article A is improved.

Further, even when the article A has a longitudinal axis such as an eclair, the gripping interval or the gripping force can be adjusted according to the area value as described above, and the same effect as described above can be obtained. In the case of the article A having a longitudinal axis, the control unit 21 detects the position and posture of the article A (direction in which the longitudinal axis is facing) in step S2-1. The control unit 21 may further detect the position of the center of gravity of the article A in step S2-1.

The grip portion 31 of the robot hand 30 may be a well-known robot finger made of metal or hard plastic. In this case, the control unit 21 changes the gripping interval, which is the interval between the tips of the plurality of robot fingers, or the gripping force by the plurality of robot fingers, according to the area value of the article A as the size information of the article A. Even in this case, the same effects as described above can be achieved.

Further, the robot hand 30 may be spherical. In this case, the robot hand 30 has a plurality of spheres approaching each other to grip the article A according to the air pressure enclosed in the robot hand 30, and each of the plurality of spheres functions as a gripping portion.

In the above embodiment, the robot control device 20 may have a learning function. For example, the learning program (learning unit) 23h is stored in the storage unit 23, and the control unit 21 may perform learning based on the learning program 23h.

For example, the learning is performed when the article A is taken out as in steps S2-1 to S2-7 described above. The control unit 21 creates a table or formula corresponding to the area value and the air pressure level using the learning information, and stores the created table or formula in the storage unit 23. The creation of such a table or formula is the result of learning to improve the gripping state.

An example of learning information is a table or formula in which the size information used in the extraction process and the air pressure level correspond to each other. Other examples of learning information are size information and air pressure information. The size information includes information such as the area value, the dimensional data, the peripheral length data, and the position of the center of gravity detected for the article A as described above. In addition, the size information includes information such as the area, size, and weight of each article A obtained by the inspection device in another process, for example, the inspection process. The size information also includes information that the operator inputs to the input device 24 based on the area, dimensions, weight, and the like of each article A obtained by the measurement. Another example of learning information is the specifications of the robot hand 30 and the grip portion 31 used in the retrieval process. Another example of learning information is the specifications of the robot 10 and the transfer device 2 used in the retrieval process. Another example of learning information is the transport speed of the transport device 2 during the retrieval process. Another example of learning information is quality information of goods such as hardness and average weight of each goods A. Another example of learning information is information on the gripping state obtained based on the observation imaging device. The gripping state information includes information on the success or failure of gripping, information on the degree of deformation of each article A, and the like. The created table or formula is stored in the storage unit 23 together with the learning information.

The control unit 21 compares, for example, the relationship between the size information and the air pressure when the take-out process is performed, and the gripping state. The control unit 21 may compare the relationship between the size information and the air pressure when the take-out process is performed, the type of the article A or the quality of the article A, and the gripped state.

The operator may input the specifications of the robot hand 30 and the grip portion 31, the transport speed of the transport device 2, the quality information of the article, and the like into the robot control device 20. Further, even if the operator observes the success or failure of gripping, the degree of deformation of each article A, etc. during the extraction process, the observation result is input to the robot control device 20, and the input information is used for the above learning. Good. Further, the result obtained in the quality inspection step may be input to the robot control device 20, and the input information may be used for the above learning.

When learning is performed in the robot control device 20 in this way, gripping the article A by the robot hand 30 becomes more appropriate.

As shown in FIG. 11, a plurality of robot control devices 20 may be connected to the host control system 100. The host control system 100 is, for example, a computer connected to a plurality of robot control devices 20 by wire, a computer arranged in the same site as the plurality of robot control devices 20, and the like. The host control system 100 is sometimes referred to as a fog computer. The upper control system 100 may be a production management system, a shipping management system, a robot management system, a department management system, or the like.

A plurality of higher control systems 100 may be connected to another higher control system 200 or the like. The host control system 200 is, for example, a cloud server connected to a plurality of host control systems 100 by wire or wirelessly. For example, a management system is formed by the plurality of robot control devices 20 and the host control systems 100 and 200. The upper control system 100 and the upper control system 200 are a control unit having a processor and the like, a display device, a storage unit having a non-volatile storage, a ROM, a RAM, and the like, and an input device such as a keyboard, a touch panel, and an operation panel, respectively. Etc. are provided.

Such a system may include a plurality of edge computers 8, a plurality of superior control systems 100, and a single or plurality of superior control systems 200, as shown in FIG. 12, for example. In the system of FIG. 12, the control device, the robot control device, and the robot can be the edge computer 8. The control device, the robot control device, and a part of the robot may be the host control system 100. The system of FIG. 12 comprises a wired or wireless network.

The plurality of edge computers 8 may transmit the learning information to another edge computer 8 having the learning function, the upper control system 100, or the upper control system 200. For example, the transmission is performed with the configuration shown in FIG. In this case, another edge computer 8 having a learning function, the upper control system 100, or the upper control system 200 performs learning using the received learning information, and an operation parameter for controlling the robot by each edge computer 8. (Learning result), operation program (learning result), etc. can be obtained automatically and accurately.

Further, another edge computer 8 having a learning function, the upper control system 100, or the upper control system 200 uses the received learning information to perform common operation parameters of a plurality of robots controlled by the plurality of edge computers 8. (Learning result), motion program (learning result), etc. can be obtained. That is, the plurality of robots have a common operation parameter or operation program. According to the system, it is possible to improve the learning speed, reliability, etc. by using various data sets.

Further, a plurality of the edge computer 8 having the learning function and the upper control system 100 are created in the other edge computer 8 having the learning function, the upper control system 100, or the upper control system 200 in the learning information and learning. At least one of the training model and the training result may be transmitted. For example, the transmission is performed with the configuration shown in FIG. The other edge computer 8 having the learning function, the upper control system 100, or the upper control system 200 has been newly optimized or made more efficient by performing the process of optimizing the knowledge and improving the efficiency based on the received information. Generate a learning model or learning result. The generated learning model or learning result is distributed to the edge computer 8 that controls the robot. When the learning result is distributed, the distribution destination edge computer 8 does not have to have the learning function.

It is possible to share the learning information, the learning model, and the learning result between the edge computers 8. For example, the sharing is performed with the configuration shown in FIG. This leads to an improvement in the efficiency of machine learning. Further, a learning function is implemented in some edge computers 8 among a plurality of edge computers 8 connected to each other by a network, and the learning results of some edge computers are used for controlling a robot by another edge computer 8. Is also possible. This leads to a reduction in the cost required for machine learning.

An example according to the above embodiment will be described below.
For example, as shown in FIG. 10, the robot control device 20 may be connected to another robot control device 20. In this case, the table (learning result) or formula (learning result) obtained by learning as described above is transferred to another robot control device 20 by the wireless or wired transmission / reception unit (output unit) 25 of the robot control device 20. It may be sent with the required information. The necessary information includes specifications of the robot 10 and the transfer device 2 related to the creation of the table or formula, specifications of the robot hand 30 and the grip portion 31, quality information of the article, and the like. The other robot control device 20 also transmits the learning result to the robot control device 20 in the same manner.
The robot control device 20 and another robot control device 20 can use the received table or formula for the work of taking out the article A. Further, the robot control device 20 and another robot control device 20 can use the received table or formula as learning information.

Further, the table or formula obtained by learning may be sent from the robot control device 20 to the higher control system 100 or the higher control system 200 together with necessary information (FIG. 11). The necessary information includes specifications of the robot 10 and the transfer device 2 related to the creation of the table or formula, specifications of the robot hand 30 and the grip portion 31, quality information of the article A, and the like.

The upper control system 100 or the upper control system 200 can learn by using a plurality of tables or formulas relating to the specifications of the same or the same type of robot hand 30 or grip portion 31, and can create a new table or formula.
The upper control system 100 or the upper control system 200 transmits the created table or formula to the robot control device 20.
The robot control device 20 can use the received table or formula for the work of taking out the article A. Further, the other robot control device 20 can also use the received table or formula as learning information when the robot control device 20 performs learning.

The robot control device 20, the upper control system 100, or the upper control system 200 can also transmit the created table or formula to the robot control device having no learning function. As a result, the gripping of the article by the robot control device having no learning function and the robot hand is improved.

2 Transport device 10 Robot 11 Main body 14 Wrist 14a Wrist flange 15, 16 Servo motor 20 Robot control device 21 Control unit 22 Display device 23 Storage unit 23h Learning program (learning unit)
24 Input device 25 Transmission / reception unit 26, 27 Servo controller 30 Robot hand 31 Grip unit 31a Internal space 32 Air supply pipe 40 Hand control device 50 Visual sensor 100 Upper control system 200 Upper control system A Article

Claims (12)

A robot control device that controls a robot hand that grips an article by a plurality of grips.
A size information acquisition means for obtaining size information of the article based on an image obtained by a visual sensor for detecting the article, and
A gripping adjustment means for changing the gripping interval, which is the interval between the plurality of gripping portions in the gripping state, or the gripping force of the plurality of gripping portions in the gripping state, is provided according to the size information.
The robot hand has a hand control device that changes the gripping interval or the gripping force to a plurality of levels.
A storage unit containing a table or expression for associating the value of the size information with the plurality of levels is provided.
It said gripping adjusting means, the table or using the equation, robot controller that sends a control signal for setting the level according to the size information of the plurality of levels to the hand control device. The robot control device according to claim 1, wherein the size information is an area value of a region detected as the article in the image. The robot control device according to claim 1, wherein the size information is dimensional data of a region detected as the article in the image. The robot control device according to claim 3, wherein the dimensional data is peripheral length data of a region detected as the article in the image. Claims 1 to claim that the gripping adjusting means changes the value of the size information corresponding to each of the plurality of levels based on the distribution of the size information of each of the plurality of articles in the single or plurality of images . 4. The robot control device according to any one of 4. A robot control device that controls a robot hand that grips an article by a plurality of grips.
A size information acquisition means for obtaining size information of the goods, and
A gripping adjusting means for changing the gripping interval, which is the interval between the plurality of gripping portions in the gripping state, or the gripping force of the plurality of gripping portions in the gripping state, according to the size information.
A robot control device including a learning unit that performs learning for improving the gripping state based on the size information and information on the gripping state when the article is gripped by the robot hand. The robot control device according to claim 6 , further comprising an output unit that outputs the learning result. The robot control device according to claim 6 or 7 , which is configured to output the learning result to another robot control device and is configured to receive the learning result from the other robot control device. The robot control device according to claim 6 or 7 , which is configured to output the learning result to a higher-level control system and is configured to receive the learning result from the higher-level control system. It ’s a system,
A robot control device that controls a robot hand that grips an article,
A host control system capable of communicating with the robot control device is provided.
The robot control device is
A learning unit that learns to improve the gripping state based on the size information of the article and the information on the gripping state when the article is gripped by the robot hand.
It has an output unit that outputs the learning result to the upper control system.
The host control system is a system that stores the learning results received from the robot control device. The system according to claim 10 , wherein the upper control system is configured to transmit the result of learning performed in the upper control system to the robot control device using the result of the learning. The upper control system is configured to transmit the learning result or the learning result performed in the upper control system using the learning result to a robot control device having no learning function. , The system according to claim 10 or 11.

Images