JP2020113262A - Learned-model generation device, robot control device, and program - Google Patents

Abstract

To realize proper operations of a robot in various operation environments without defining operations corresponding to all the operation environments.SOLUTION: A learned model generation device to generate a learned model regarding operations of a robot that performs a task consisting of multiple operations comprises: means of acquiring sensor data regarding each of multiple operation environments; means of extracting from the sensor data a feature amount representing an operation environment; means of storing a learning data set in which the extracted feature amount and operations are in association; and means of referring to the learning data set to generate a learned model in which a relation between the operation environment and operations is defined.SELECTED DRAWING: Figure 6

Description

The present invention relates to a learned model generation device, a robot control device, and a program.

In recent years, attention has been focused on the introduction of robots in various fields. For example, introduction of a robot for improving manufacturing is expected to reduce manufacturing cost and improve yield.

In order to introduce a robot, it is important to give an appropriate control command to the robot.
As a technique for giving a control command to a robot, for example, Patent Document 1 discloses a user interface for giving a control command to the robot.

JP, 2016-112651, A

In order to operate a robot, it is necessary to define an operation for causing the robot to execute a task. Generally, the motion of a robot is defined by a combination of a motion environment and a motion to be executed in the motion environment. In order to properly operate the robot, it is necessary to define the motions corresponding to all assumed motion environments.
If there is an operating environment in which the operation is not specified, the operation of the robot will malfunction.

However, in Patent Document 1, displaying the GUI on the display is equivalent to predefining the operating environment. That is, the GUI cannot be displayed on the display for an operating environment that is not specified in advance. Therefore, the robot cannot properly operate in an operation environment in which the operation is not defined.

As described above, conventionally, the robot cannot properly operate in an operation environment in which the operation is not regulated.

An object of the present invention is to realize an appropriate motion of a robot in various motion environments without defining motions corresponding to all motion environments.

One aspect of the present invention is
A trained model generation device for generating a trained model relating to a motion of a robot that executes a task composed of a plurality of motions,
A means for acquiring sensor data for each of a plurality of operating environments of the robot,
A means for extracting a feature amount representing the operating environment from the sensor data,
A means for storing a learning data set in which the extracted feature quantity and the motion are associated with each other,
A means for generating a learned model in which the relationship between the operating environment and the operation is defined with reference to the learning data set,
It is a learned model generation device.

According to the present invention, it is possible to realize an appropriate motion of a robot in various operating environments without defining motions corresponding to all operating environments.

It is a block diagram which shows the structure of the information processing system of this embodiment. It is a functional block diagram of the learned model generation apparatus of FIG. It is a functional block diagram of the robot of FIG. It is a functional block diagram of the learned model generation apparatus of FIG. It is a functional block diagram of the control object robot of FIG. It is explanatory drawing of the outline of this embodiment. It is a figure which shows the data structure of the task database of this embodiment. It is a figure which shows the data structure of the data set for learning of this embodiment. It is a flowchart of the learned model generation process of this embodiment. It is a figure which shows the example of a screen displayed in the process of FIG. FIG. 10 is a network diagram of a learned model generated in the processing of FIG. 9. It is a flow chart of robot control processing of this embodiment. It is a figure which shows the example of a screen displayed in the process of FIG. FIG. 11 is a diagram showing an example of a screen displayed in the information processing of the first modification. It is a figure which shows the data structure of the child task database of the modification 3. 11 is a flowchart of a learned model generation process of modification 3; It is a network diagram of the learned model of the modification 3.

An embodiment of the present invention will be described in detail below with reference to the drawings. In addition, in the drawings for describing the embodiments, the same components are denoted by the same reference symbols in principle, and repeated description thereof will be omitted.

(1) Configuration of Information Processing System The configuration of the information processing system will be described. FIG. 1 is a block diagram showing the configuration of the information processing system of this embodiment.

As shown in FIG. 1, the information processing system 1 includes a learned model generation device 10, a sensor unit 20, a robot 30, a robot control device 50, and a controlled robot 70.

The learned model generation device 10 is connected to the sensor unit 20, the robot 30, and the robot control device 50.

The sensor unit 20 is connected to the learned model generation device 10 and the robot control device 50.

The robot 30 is connected to the learned model generation device 10.
The controlled robot 70 is connected to the robot controller 50.
The robot 30 and the controlled robot 70 are an example of a self-sustained motion device configured to autonomously operate. The robot 30 and the controlled robot 70 include, for example, the following.
・Robot arm ・Machine tool ・Robot cleaner ・Drone ・Self-supporting medical equipment (for example, endoscope)

The robot control device 50 is connected to the learned model generation device 10, the sensor unit 20, and the control target robot 70.

The learned model generation device 10 is configured to generate a learned model for controlling the robot 30. The learned model generation device 10 is, for example, a personal computer or a server computer.

The sensor unit 20 is configured to acquire sensor data regarding operating environments of the robot 30 and the controlled robot 70. The sensor data includes at least one of the following, for example.
-Robot 30 and still images around robot 30, and still images around controlled robot 70 and controlled robot 70-Robot 30 and moving images around robot 30, controlled robot 70 and controlled robot 70 Video around the robot 30. Voice around the robot 30 and the robot 30, and voice around the control target robot 70 and the control target robot 70.

The robot 30 is configured to operate according to a user instruction.

The robot controller 50 is configured to control the controlled robot 70. The robot controller 50 is, for example, a personal computer or a server computer.

The controlled robot 70 is configured to operate under the control of the robot controller 50.

(1-1) Configuration of Learned Model Generating Device The configuration of the learned model generating device 10 will be described. FIG. 2 is a functional block diagram of the learned model generation device of FIG.

As shown in FIG. 2, the learned model generation device 10 includes a storage device 11, a processor 12, an input/output interface 13, and a communication interface 14.

The storage device 11 is configured to store programs and data. The storage device 11 is, for example, a combination of a ROM (Read Only Memory), a RAM (Random Access Memory), and a storage (for example, a flash memory or a hard disk).

The programs include, for example, the following programs.
-OS (Operating System) program-Application (for example, learned model generation application) program that executes information processing

The data includes, for example, the following data.
-Database referred to in information processing-Data obtained by executing information processing (that is, execution result of information processing)

The processor 12 is configured to realize the function of the learned model generation device 10 by activating a program stored in the storage device 11. The processor 12 is an example of a computer.

The input/output interface 13 is configured to acquire a user's instruction from an input device connected to the trained model generation apparatus 10 and output information to an output device connected to the trained model generation apparatus 10. ..
The input device is, for example, a keyboard, a pointing device, a touch panel, or a combination thereof. The input device also includes the sensor unit 20.
The output device is, for example, a display.

The communication interface 14 is configured to control communication between the learned model generation device 10, the robot 30, and the robot control device 50.

(1-2) Configuration of Robot The configuration of the robot 30 of this embodiment will be described. FIG. 3 is a functional block diagram of the robot of FIG.

As shown in FIG. 3, the robot 30 includes a storage device 31, a processor 32, a communication interface 34, and a drive unit 35.

The storage device 31 is configured to store programs and data. The storage device 31 is, for example, a combination of a ROM, a RAM, and a storage (for example, a flash memory or a hard disk).

The processor 32 is configured to realize the function of the robot 30 by activating a program stored in the storage device 31. The processor 32 is an example of a computer.

The communication interface 34 is configured to control communication between the robot 30 and the trained model generation device 10.

The drive unit 35 is, for example, a robot arm having joints. The drive unit 35 is configured to drive under the control of the processor 32.

(1-3) Configuration of Robot Control Device The configuration of the robot control device 50 will be described. FIG. 4 is a functional block diagram of the learned model generation device of FIG.

As shown in FIG. 4, the robot controller 50 includes a storage device 51, a processor 52, an input/output interface 53, and a communication interface 54.

The storage device 51 is configured to store programs and data. The storage device 51 is, for example, a combination of a ROM, a RAM, and a storage (for example, a flash memory or a hard disk).

The programs include, for example, the following programs.
-OS (Operating System) program-Application (for example, robot control application) program that executes information processing

The data includes, for example, the following data.
-Database referred to in information processing-Data obtained by executing information processing (that is, execution result of information processing)

The processor 52 is configured to realize the function of the robot controller 50 by activating a program stored in the storage device 51. The processor 52 is an example of a computer.

The input/output interface 53 is configured to obtain a user instruction from an input device connected to the robot controller 50 and output information to an output device connected to the robot controller 50.
The input device is, for example, a keyboard, a pointing device, a touch panel, or a combination thereof. The input device also includes the sensor unit 20.
The output device is, for example, a display.

The communication interface 54 is configured to control communication between the robot control device 50 and the learned model generation device 10 and the controlled robot 70.

(1-4) Configuration of Control Target Robot The configuration of the control target robot 70 of the present embodiment will be described. FIG. 5 is a functional block diagram of the controlled robot shown in FIG.

As shown in FIG. 5, the controlled robot 70 includes a storage device 71, a processor 72, a communication interface 74, and a drive unit 75.

The storage device 71 is configured to store programs and data. The storage device 71 is, for example, a combination of a ROM, a RAM, and a storage (for example, a flash memory or a hard disk).

The processor 72 is configured to realize the function of the controlled robot 70 by activating a program stored in the storage device 71. The processor 72 is an example of a computer.

The communication interface 74 is configured to control communication between the controlled robot 70 and the robot control device 50.

The drive unit 75 is, for example, a robot arm having joints. The drive unit 75 is configured to drive under the control of the processor 72.

(2) Outline of Embodiment An outline of this embodiment will be described. FIG. 6 is an explanatory diagram of the outline of the present embodiment.

In the present embodiment, the “task” is a work to be completed by the robot 30 and the controlled robot 70.
An "action" is an element required to complete a task.
The “operating environment” is a combination of a situation when performing an action and a place where the action is performed.
That is, the task is completed as a result of the robot 30 and the controlled robot 70 operating in each of a plurality of operating environments.

As shown in FIG. 6, the learned model generating apparatus 10 receives a designation of a motion of the robot 30 according to a motion environment from a user.
The robot 30 operates according to a user instruction.
The sensor unit 20 generates sensor data regarding each of a plurality of operating environments of the robot 30.
The learned model generation device 10 acquires sensor data from the sensor unit 20.
The learned model generation device 10 extracts the feature amount of each operating environment from the sensor data.
The learned model generation device 10 generates a learning data set in which a feature amount (that is, an operating environment) and a motion are associated with each other.
The learned model generation device 10 refers to the learning data set and generates a learned model in which the relationship between the operating environment and the operation is defined.

In the present embodiment, the robot control device 50 that controls the controlled robot 70 transmits a command to the controlled robot 70 by referring to the learned model generated by the learned model generation device 10. The controlled robot 70 executes an appropriate operation according to the operating environment according to the command transmitted from the robot control device 50. As a result, the controlled robot 70 can be caused to execute a task without prescribing operations corresponding to all operating environments.

(3) Data Table The data table of this embodiment will be described.

(3-1) Task Database The task database of this embodiment will be described. FIG. 7 is a diagram showing the data structure of the task database of this embodiment.

The task database of FIG. 7 stores task information regarding tasks.
The task database includes a "task ID" field, a "task name" field, and a plurality of "operating environment" fields ("operating environment A" field, "operating environment B" field...).
Each field is associated with each other.

Task identification information for identifying a task is stored in the "task ID" field.

Information (for example, text) related to the task name is stored in the “task name” field.

The plurality of “operating environment” fields correspond to a plurality of operating environments assumed in the task (for example, operating environment A, operating environment B... ). Each "operating environment" field includes an "image" field and a "command" field.

An image corresponding to each operating environment is stored in the "image" field.

The “command” field stores a plurality of commands that can be assigned in each operating environment. The command is at least one of the following, for example.
-Abstract command that represents a motion (as an example, a command to "grab an object marked with "1" among objects stored in a pallet")
-Drive parameters indicating motion (for example, the value of the joint angle of the joint portion included in the robot 30)

(3-2) Learning Data Set The learning data set of this embodiment will be described. FIG. 8 is a diagram showing the data structure of the learning data set of this embodiment.

Learning data is stored in the learning data set of FIG. The learning data set is associated with the task identification information.
The learning data set includes a "data ID" field, a "time" field, a "sensor data" field, a "feature amount" field, and a "command" field.
Each field is associated with each other.

Learning data identification information for identifying learning data is stored in the “data ID” field.

The time when the operation is detected by the sensor unit 20 is stored in the “time” field.

The sensor data acquired by the sensor unit 20 is stored in the “sensor data” field. The sensor data is, for example, at least one of the following.
・Still image data ・Video data ・Voice data

The “feature amount” field stores a feature amount corresponding to the operating environment of the robot 30.

In the “command” field, a command that is an operation command for the robot 30 is stored.

(4) Information processing The information processing of this embodiment will be described.

(4-1) Learned Model Generation Process The learned model generation process of this embodiment will be described. FIG. 9 is a flowchart of the learned model generation process of this embodiment. FIG. 10 is a diagram showing an example of a screen displayed in the process of FIG. FIG. 11 is a network diagram of the learned model generated in the processing of FIG.

As shown in FIG. 9, the learned model generation apparatus 10 receives a task designation (S110).
Specifically, the processor 12 displays the screen P10 (FIG. 10) on the display.

The screen P10 includes an operation object B10 and a field object F10.
The field object F10 is an object that receives a user input of task identification information.
The operation object B10 is an object for confirming a user input to the field object F10.

When the user inputs arbitrary task identification information to the field object F10 and operates the operation object B10, the processor 12 uses the task identification information input to the field object F10 as a target for generation of a learned model. Task identification information.

After step S110, acceptance of an operation command (S111) is executed.
Specifically, the task database (FIG. 7) is referenced to identify the record associated with the task identification information identified in step S110.
The processor 12 displays the screen P11 based on the combination of the “image” field and the “command” field of the “action A” field of the specified record on the display.

The screen P11 includes operation objects B11a to B11c and an image object IMG11.
The image object IMG11 is an image in the “image” field of the “action A” field (that is, an image corresponding to the action environment A). The operating environment A is an environment in which the objects marked with “1” to “3” are stored in the pallet.
The operation objects B11a to B11c are each assigned a value in the “command” field of the “motion A” field (that is, a command that can be given to the robot 30 in the motion environment A). For example, each of the operation objects B11a to B11c is assigned a command for executing an operation of grasping an object marked with “1” to “3” in the image object IMG11. When the user operates any of the operation objects B11a to B11c, the command assigned to the object operated by the user is specified.

The screen P12 includes operation objects B12a and B12b and an image object IMG12.
The image object IMG12 is an image in the “image” field of the “action B” field (that is, an image corresponding to the operation environment B). The operating environment B is an environment in which the objects to which “1” and “2” are attached are stored in the pallet. In operating environment A, the object marked with “3” is stored in the pallet, whereas in operating environment B, the object marked with “3” does not exist in the pallet. That is, the arrangement of the objects on the palette of the operating environment B is different from that of the operating environment A.
The operation objects B12a and B12b are each assigned a value in the "command" field of the "motion B" field (that is, a command that can be given to the robot 30 in the motion environment B). For example, each of the operation objects B12a and B12b is assigned with a command for executing an operation of grabbing an object marked with "1" or "2" in the image object IMG12. When the user operates any of the operation objects B12a and B12b, the command assigned to the object operated by the user is specified.

The transitions of the screens P11 to P12 are in random order.

After step S111, the learned model generation device 10 determines a command (S112).
Specifically, when the user operates the operation object B11a, the processor 12 identifies the command assigned to the operation object B11a.
The processor 12 sends the specified command to the robot 30.

The processor 32 of the robot 30 generates a control signal corresponding to the command transmitted from the processor 12.
The drive unit 35 drives according to the control signal generated by the processor 32. As a result, the robot 30 operates in the operating environment A according to the control command from the user.

After step S112, the learned model generation device 10 executes acquisition of sensor data (S113).
Specifically, the sensor unit 20 generates sensor data regarding the operating environment of the robot 30 that has operated in step S112.
The processor 12 acquires the sensor data generated by the sensor unit 20.

After step S113, the learned model generation device 10 executes extraction of the characteristic amount (S114).
Specifically, the processor 12 extracts the characteristic amount of the sensor data acquired in step S113.
For example, when the sensor data is a still image or a moving image, the processor 12 extracts the image feature amount corresponding to the operating environment by applying the image analysis algorithm to the sensor data.
For example, when the sensor data is voice, the processor 12 extracts the voice feature amount corresponding to the operating environment by applying the voice analysis algorithm to the sensor data.

After step S114, the learned model generation device 10 generates a learning data set (S115).
Specifically, the task identification information identified in step S110 and the new learning data set (FIG. 8) are stored in the storage device 11 in association with each other.
The processor 12 stores the feature amount extracted in step S114, the time when step S114 was executed, and the command specified in step S112 in a new record of the learning data set in association with each other.

Steps S111 to S115 are repeatedly executed until step S115 is completed for all the predetermined operating environments (S116).
When step S115 has not been completed for all of the predetermined operating environments (S116-NO), step S111 is executed.
When step S115 has been completed for all of the predetermined operating environments (S116-YES), step S117 is executed.

When step S115 has been completed for all of the predetermined operating environments (S116-YES), the learned model generation device 10 generates a learned model (S117).
Specifically, the processor 12 generates a learned model by applying a predetermined learning algorithm to the learning data set (FIG. 8) generated in step S115.
The learning algorithm is, for example, one of the following.
・RNN (Recurrent Neural Network)
・LSTM (Long Short-Term Memory)
・CNN (Convolution Neural Network)
・SVM (Support Vector Machine)

FIG. 11 shows an RNN network which is an example of a learned model.

The RNN network includes an input X, an output Y, and a hidden element S.

For example, the input Xt1 in step t1 is the plurality of feature quantities Xt11 to Xt13 extracted in step S114.
The hidden element St1 in step t1 is a function of the operating environment information (that is, the feature amount) Xt11 to Xt13 in step t1.
The output Yt1 in step t1 is calculated based on the feature quantities Xt11 to Xt13. The output Yt1 is a predicted probability of a motion in the motion environment determined by the feature quantities Xt11 to Xt13. An operation in which the output Yt1 is higher than or higher than a predetermined value means an operation to be executed in the operating environment.

The input Xt2 in step t2 is the plurality of feature quantities Xt21 to Xt23 extracted in step S114.
The hidden element St2 at step t2 is a function of the operating environment information (that is, the feature amount) Xt21 to Xt23 at step t2.
The output Yt2 at step t2 is calculated based on the combination of the feature quantities Xt21 to Xt23 and the hidden element St1. The output Yt2 is a predicted probability of motion in the motion environment determined by the feature quantities Xt21 to Xt23. An operation in which the output Yt2 is higher than or higher than a predetermined value means an operation to be executed in the operating environment.

The input Xt3 in step t3 is the plurality of feature quantities Xt31 to Xt33 extracted in step S114.
The hidden element St3 in step t3 is a function of the operating environment information (that is, the feature amount) Xt31 to Xt33 in step t3.
The output Yt3 in step t3 is calculated based on the combination of the feature quantities Xt31 to Xt33 and the hidden element St2. The output Yt3 is a predicted probability of motion in the motion environment determined by the feature quantities Xt31 to Xt33. The operation in which the output Yt3 is higher than or higher than the predetermined value means the operation to be executed in the operating environment.

The processor 32 stores the task identification information identified in step S110 and the learned model (FIG. 11) in the storage device 11 in association with each other.

(4-2) Robot control processing The robot control processing of this embodiment will be described. FIG. 12 is a flowchart of the robot control process of this embodiment. FIG. 13 is a diagram showing an example of a screen displayed in the process of FIG.

As shown in FIG. 12, the robot control device 50 executes acceptance of designation of a task (S150).
Specifically, the processor 12 displays the screen P20 (FIG. 13) on the display.

The screen P20 includes an operation object B20 and a field object F20.
The field object F210 is an object that receives a user input of task identification information.
The operation object B20 is an object for confirming a user input to the field object F20.

When the user inputs arbitrary task identification information into the field object F20 and operates the operation object B20, the processor 12 uses the task identification information input into the field object F20 as the task identification information of the task to be executed. Identify.

After step S150, the robot control device 50 executes acquisition of sensor data (S151).
Specifically, the sensor unit 20 generates sensor data regarding the operating environment of the controlled robot 70.
The processor 12 acquires the sensor data generated by the sensor unit 20.

After step S151, the robot controller 50 executes the extraction of the characteristic amount (S152).
Specifically, the processor 12 extracts the feature amount of the sensor data acquired in step S151, as in step S114 (FIG. 9).

After step S152, the robot controller 50 executes command generation (S153).
Specifically, the processor 52 accesses the storage device 11 of the learned model generation device 10 and reads the learned model (FIG. 11) associated with the task identification information identified in step S150.
The processor 52 generates a command corresponding to the operating environment of the controlled robot 70 by inputting the feature amount extracted in step S152 to the read learned model.
The processor 52 transmits the generated command to the controlled robot 70.

The processor 72 of the controlled robot 70 generates a control signal corresponding to the command transmitted from the processor 52.
The drive unit 75 drives according to the control signal generated by the processor 72.

Steps S151 to S153 are repeatedly executed until step S153 ends for all of the predetermined operating environments (S154).
When step S153 has not been completed for all of the predetermined operating environments (S154-NO), step S151 is executed.
When step S153 has been completed for all of the predetermined operating environments (S154-YES), the robot control process ends.

According to the present embodiment, the robot control device 50 controls the controlled robot 70 by referring to the learned model generated by the learned model generation device 10. As a result, the controlled robot 70 can be caused to execute a task without prescribing operations corresponding to all operating environments.

(5) Modified Example A modified example of the present embodiment will be described.

(5-1) Modification 1
Modification 1 will be described. Modification 1 is an alternative example of the operating environment. FIG. 14 is a diagram showing a screen example displayed in the information processing of the first modification.

In step S111 (FIG. 9), the processor 12 displays the screen P20 (FIG. 14) based on the combination of the “image” field and the “command” field of the “motion C” field of the specified record on the display.

The screen P20 includes operation objects B20a to B20c and an image object IMG20.
The image object IMG20 is an image in the “image” field of the “motion C” field (that is, an image corresponding to the motion environment C). The operating environment C is an environment in which the objects marked with “4” to “6” are stored in the pallet. In operating environment A, round objects are contained in a pallet having 3 slots, whereas in operating environment B rectangular objects are contained in a pallet having 6 slots. That is, the pallet and the object of the operating environment B are different from those of the operating environment A.
The values of the “command” field of the “motion C” field (that is, the commands that can be given to the robot 30 in the motion environment C) are assigned to the operation objects B20a to B20c, respectively. For example, each of the operation objects B20a to B20c is assigned with a command for executing an operation of grasping an object marked with “4” to “6” in the image object IMG20. When the user operates any of the operation objects B20a to B20c, the command assigned to the object operated by the user is specified.

The transitions of the screens P11 to P20 are in random order.

(5-2) Modification 2
Modification 2 will be described. Modification 2 is an example in which the sensor data is data relating to the physical quantity of the robot 30.

The sensor unit 20 of the second modification acquires sensor data regarding the physical quantity of the robot 30.
The physical quantity of the robot 30 includes at least one of the following, for example.
-The force applied to the force sensor arranged in the robot 30-The torque applied to each axis obtained by the torque sensor arranged in the robot 30-The pressure applied to the contact surface of the pressure sensor arranged in the robot 30-The robot 30 Voltage acquired by the voltage sensor arranged (specifically, voltage generated when moving each axis of the robot 30)
-Temperature acquired by the robot 30 or a temperature sensor arranged around the robot 30

The sensor data of Modification 2 can be replaced or combined with the sensor data of the present embodiment (at least one of a still image, a moving image, and a sound).

According to the second modification, even when the physical quantity of the robot 30 is used, the robot 30 can be caused to execute a task without prescribing operations corresponding to all operating environments.

(5-3) Modification 3
Modification 3 will be described. Modification 3 is an example in which a child task associated with the task of the present embodiment (hereinafter referred to as “parent task”) exists.

(5-3-1) Child Task Database The child task database of Modification 3 will be described. FIG. 15 is a diagram showing a data structure of the child task database according to the modified example 3.

The child task database of FIG. 15 stores child task information regarding child tasks. The child task database is associated with the task identification information. The child task database is an example of a learning data set.
The child task database includes a "child task ID" field, a "time" field, a "command" field, and a "sensor data" field.
Each field is associated with each other.

The "child task ID" field stores child task identification information for identifying the child task.

The time when the operation is detected by the sensor unit 20 is stored in the “time” field.

In the “command” field, a command that is an operation command for the robot 30 (for example, a value of a joint angle of n (n=1 or more) axis 1 to axis n arranged on the robot 30) is stored. ..

The “feature amount” field stores a feature amount corresponding to the operating environment of the robot 30.

(5-3-3) Information Processing The information processing of Modification 3 will be described. FIG. 16 is a flowchart of the learned model generation process of the modified example 3. FIG. 17 is a network diagram of the learned model of Modification 3.

As illustrated in FIG. 16, the learned model generation device 10 executes a motion input (S210).
Specifically, when the user specifies the task identification information and operates the robot 30a, the robot 30a executes the operation at the joint angle according to the user's operation.
The processor 12 acquires the control parameter (for example, the value of the joint angle) of the executed operation from the robot 30a.

After step S210, the learned model generation device 10 executes the operation output (S211).
Specifically, the processor 12 outputs the control parameter acquired in step S210 to the robot 30b.

After step S211, the learned model generation device 10 executes acquisition of sensor data (S212).
Specifically, the robot 30b operates according to the control parameter output in step S211.
The sensor unit 20 produces|generates the sensor data regarding the operating environment of the robot 30 which operated in step S112.
The processor 12 acquires the sensor data generated by the sensor unit 20.

After step S212, the learned model generation device 10 executes extraction of a feature amount (S213).
Specifically, the processor 12 extracts the characteristic amount of the sensor data acquired in step S113.

After step S213, the learned model generation device 10 generates a learning data set (S214).
Specifically, the processor 12 adds a new record to the child task database (FIG. 15) associated with the task identification information designated by the user in step S210. The following information is stored in each field of the new record.
New child task identification information is stored in the "child task ID" field.
The value of the time when the sensor data was acquired in step S212 is stored in the "time" field.
The control parameter acquired in step S210 is stored in the "command" field.
The sensor data acquired in step S212 is stored in the "sensor data" field.

Steps S210 to S214 are repeatedly executed until step S214 is completed for all the predetermined operating environments (S215).
When step S214 has not been completed for all of the predetermined operating environments (S215-NO), step S210 is executed.
When step S214 has been completed for all of the predetermined operating environments (S215-YES), step S216 is executed.

The learned model generation device 10 executes generation of a learned model (S216).
Specifically, the processor 12 generates a learned model by applying a predetermined learning algorithm (eg, RNN or LSTM) to the learning data set (FIG. 15) generated in step S215.

FIG. 17 shows a network of RNNs, which is an example of a trained model of Modification 3. This learned model includes an upper layer network (FIG. 17A) and a lower layer network (FIG. 17B).

The upper layer network (FIG. 17A) is the same as the network (FIG. 11) of this embodiment.

A DCAE (Deep Convolutional. Autoencoder) algorithm is used for the network of the lower layer (FIG. 17B). The lower layer network includes multiple stages of auto encoders. The sensor data generated by the sensor unit 20 is input to the highest-order auto encoder. The auto encoder at each stage extracts the feature amount by compressing the dimension of the sensor data. The extracted feature amount is associated with y (joint angle).

The processor 32 of Modification 3 stores the task identification information designated by the user in step S210 and the learned model (FIG. 17) in the storage device 11 in association with each other.

According to the modified example 3, the learned model is generated from the learning data set prepared in the unit of the detailed small task that constitutes the parent task. As a result, learning can be realized in units of small tasks. In this case, the user only has to give the operation command in units of small tasks, so that it is possible to reduce the difficulty level of giving the user's operation command.

(6) Summary of the present embodiment The present embodiment will be summarized.

The first aspect of the present embodiment is
A trained model generation device 10 for generating a trained model relating to a motion of a robot 30 that executes a task including a plurality of motions,
The robot 30 is provided with a unit (for example, the processor 12 that executes the process of step S113) that acquires sensor data regarding each of a plurality of operating environments,
A unit (for example, the processor 12 that executes the process of step S114) for extracting a feature amount representing an operating environment from the sensor data is provided,
A means (for example, the processor 12 that executes the process of step S115) that stores the learning data set in which the extracted feature amount and the action (for example, command) are associated with each other,
The learning data set is referred to, and means for generating a learned model in which the relationship between the operating environment and the operation is defined (for example, the processor 12 that executes the process of step S117) is provided.
This is the learned model generation device 10.

The second aspect of the present embodiment is
The operating environment is at least one of a moving image, a still image, and a sound,
This is the learned model generation device 10.

The third aspect of the present embodiment is
The means for acquiring the sensor data acquires the sensor data regarding the physical quantity of the robot 30 from the sensor,
This is the learned model generation device 10.

The fourth aspect of the present embodiment is
The sensor data includes at least one of a force applied to the sensor unit, a torque applied to each axis of the robot 30, a pressure applied to a contact surface of the sensor, a temperature, and a voltage generated when moving each axis of the robot 30,
This is the learned model generation device 10.

The fifth aspect of the present embodiment is
A means (for example, the learning data set in FIG. 8) that stores the task identification information for identifying the task and the learning data set in association with each other,
This is the learned model generation device 10.

The sixth aspect of the present embodiment is
A means for storing the task identification information for identifying the task and the learned model in association with each other (for example, the processor 12 that executes the process of step S117),
This is the learned model generation device 10.

A seventh aspect of the present embodiment is
In the learning data set, the feature amount and the action are associated with each other for each of a plurality of child tasks that make up the task,
A means for generating a learned model generates a learned model by inputting a combination of a feature amount and a motion to a higher-order network corresponding to a task and a lower-order network corresponding to a child task,
This is the learned model generation device 10.

The eighth aspect of the present embodiment is
The means for generating is to apply RNN (Recurrent Neural Network), LSTM (Long Short-Term Memory), CNN (Convolution Neural Network), or SVM (Support Vector Machine) to the learning data set, Generate a trained model,
This is the learned model generation device 10.

The ninth aspect of the present embodiment is
A robot control device 50 capable of accessing the learned model generated by the learned model generation device 10 as described above,
A means for acquiring sensor data related to the operating environment of the robot 30 to be controlled,
Equipped with means for extracting the feature amount of sensor data,
By inputting the extracted feature amount to the learned model, a means for generating a command corresponding to the operating environment is provided,
A means for operating the controlled robot 70 by transmitting a command to the controlled robot 70,
The robot controller 50.

A tenth aspect of the present embodiment is a program for causing a computer (for example, the processor 12 or 52) to function as each unit described in any of the above.

(7) Other Modifications Other modifications will be described.

The storage device 11 may be connected to the trained model generation device 10 via a network.
The storage device 51 may be connected to the robot control device 50 via a network.

The learned model generation device 10 and the robot control device 50 may be the same device (that is, may be integrally configured).

The robot 30 used in the learned model generation process (FIG. 9) and the control target robot 70 used in the robot control process (FIG. 12) may be the same robot or different robots. good.

The method of receiving a user's operation command for the robot 30 is not limited to the example of FIG. For example, a motion command may be given to the robot 30 through a user's operation on a haptics device connected to the robot 30.

In the example of FIG. 1, the sensor unit 20 is connected to the learned model generation device 10, but the present invention is not limited to this. The sensor unit 20 may be connected to the learned model generation device 10 via the robot 30. In this case, the learned model generation device 10 acquires sensor data via the robot 30.
The sensor unit 20 may be arranged in the robot 30.

In the example of FIG. 1, the sensor unit 20 is connected to the robot controller 50, but the sensor unit 20 is not limited to this. The sensor unit 20 may be connected to the robot control device 50 via the controlled robot 70. In this case, the robot control device 50 acquires sensor data via the controlled robot 70.
The sensor unit 20 may be arranged in the control target robot 70.

In the present embodiment, the example in which the robot controller 50 executes the extraction of the characteristic amount (S152) and the generation of the command (S153) has been described, but the execution subject of steps S152 to S153 is not limited to this. The controlled robot 70 may execute steps S152 to S153. In this case, the processor 72 of the controlled robot 70 extracts the characteristic amount of the sensor data acquired in step S151, as in step S114 (FIG. 9). The processor 72 inputs the feature amount into the learned model stored in the storage device 11 to generate a control signal corresponding to the operating environment of the controlled robot 70.

Although the embodiments of the present invention have been described above in detail, the scope of the present invention is not limited to the above embodiments. In addition, the above-described embodiment can be variously modified and changed without departing from the gist of the present invention. Further, the above-described embodiments and modified examples can be combined.

1: Information processing system 10: Learned model generation device 11: Storage device 12: Processor 13: Input/output interface 14: Communication interface 20: Sensor unit 20: Sensor 30: Robot 31: Storage device 32: Processor 34: Communication interface 35 : Drive unit 50: Robot controller 51: Storage device 52: Processor 53: Input/output interface 54: Communication interface 70: Controlled robot 71: Storage device 72: Processor 74: Communication interface 75: Drive unit

Claims (10)

A trained model generation device for generating a trained model relating to a motion of a robot that executes a task composed of a plurality of motions,
A means for acquiring sensor data for each of a plurality of operating environments of the robot,
From the sensor data, by means of an auto encoder, a means for extracting a feature amount representing the operating environment,
A means for generating a learning data set in which the extracted feature quantity and the operation are associated with each other,
A means for generating a learned model in which the relationship between the operating environment and the operation is defined with reference to the learning data set,
Trained model generator. The operating environment is at least one of a moving image, a still image, and a sound,
The learned model generation device according to claim 1. The means for acquiring the sensor data acquires sensor data relating to the physical quantity of the robot from the sensor,
The trained model generation device according to claim 1 or 2. The sensor data is at least one of a force applied to the sensor unit, a torque applied to each axis of the robot, a pressure applied to a contact surface of the sensor, a temperature, and a voltage generated when moving each axis of the robot. including,
The trained model generation device according to claim 3. A task identification information for identifying the task, and a means for storing the learning data set in association with each other,
The learned model generation device according to any one of claims 1 to 4. And a means for storing task identification information for identifying a task and the learned model in association with each other,
The learned model generation device according to any one of claims 1 to 5. In the learning data set, the feature amount and the operation are associated with each other for each of a plurality of child tasks constituting the task,
The means for generating the learned model generates the learned model by inputting the combination of the feature quantity and the operation to the upper network corresponding to the task and the lower network corresponding to the child task. To do
The learned model generation device according to claim 1. The means for generating may apply RNN (Recurrent Neural Network), LSTM (Long Short-Term Memory), CNN (Convolution Neural Network), or SVM (Support Vector Machine) to the learning data set. To generate the trained model,
The learned model generation device according to any one of claims 1 to 7. A robot controller capable of accessing a learned model generated by the learned model generating device according to claim 1.
A means for acquiring sensor data regarding the operating environment of the controlled robot to be controlled is provided,
A means for extracting the characteristic amount of the sensor data,
A means for generating a command corresponding to the operating environment by inputting the extracted feature quantity into the learned model,
A means for operating the controlled robot by transmitting the command to the controlled robot,
Robot controller. A program for causing a computer to function as each unit according to claim 1.