JP2006198709A - Creation device and method for teaching motion data to articulated robot - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technique for creating a motion data by effectively utilizing a motion capture device. <P>SOLUTION: The creation device creates the motion data which describes the attitude of an articulated robot with time. The device is equipped with a storing means for storing a first motion data which describes the attitude in executing a scheduled motion with time, the motion capture device for generating a second motion data which measures the attitude of a demonstrator with time and describes the attitude in executing the demonstration motion with time, a motion data integration means which integrates the first motion data and the second motion data, and creates a third motion data which describes the attitude in executing the motion superimposing the demonstration motion on the scheduled motion with time, and a correction means which corrects the third motion data integrated by the motion data integration means to an executable motion data by the articulated robot. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a technique for creating motion data taught to an articulated robot. In particular, the present invention relates to a technique for simplifying creation of motion data using a motion capture device.

The multi-joint robot includes a plurality of links and a plurality of joints. An articulated robot changes its posture over time by adjusting the joint angle of each joint (the rotation angle of the other link with respect to one link), and moves, walks, holds objects, and manipulates objects Execute.
In order for the articulated robot to execute a predetermined operation, operation data describing the posture of the articulated robot over time is required. If the posture is specified with time, the articulated robot changes its posture with time, and consequently operates. The motion data describes the posture with time, but it also describes the change with time of the posture and also describes the motion. The motion data needs to describe the joint angle of each joint of the multi-joint robot, the position of each link, or the direction of each link over time, and creating the motion data requires very troublesome work It is said.

A technique using a motion capture device has been developed when creating motion data to be taught to a humanoid robot or the like. The motion capture device measures the posture of the person performing the performance over time and describes the posture over time while performing the demonstration operation (the data describing the change in posture over time, It is also data that describes the action. Humans can demonstrate the motions they want to perform, and the motion capture device can directly create motion data that describes the motions that the robot wants to perform, by using the motion capture device to convert the changes in the posture of the performer over time. It becomes possible.
Patent Document 1 discloses a device that creates motion data using a motion capture device. In Patent Literature 1, there is a technique for storing in advance an action (change in posture over time) that can be executed by a humanoid robot and correcting the action obtained by the motion capture device to an action that can be executed by a humanoid robot. It is disclosed.
JP 2002-301694 A

Using a motion capture device has the advantage that motion data can be obtained relatively easily, but the performer must faithfully demonstrate the motion that the robot wants to perform. For example, when trying to make a life-size humanoid robot walk a distance of 10 m, the performer must also walk a distance of 10 m. When trying to move the robot over a wide range, the performer must also move over a wide range, and the motion capture device becomes very large. Depending on the operation that the robot wants to perform, the motion capture device becomes very large. For this reason, in the conventional technology, there are limited opportunities to create operation data relatively easily using a motion capture device.
In view of this, there is a need for a technique that allows the user to enjoy the advantage that motion data can be created relatively easily by using a motion capture device.
The present invention solves the above problems. In the present invention, when creating motion data, the motion capture device is used for operations that are best suited for the motion capture device, and the motion data is used for operations that are not suitable for the motion capture device without using the motion capture device. Provide technology to create.

The motion data creation device embodied by the present invention creates motion data that describes the posture of the articulated robot over time. The data that describes the posture over time is data that describes a change in posture over time and data that describes the motion.
The motion data creating apparatus of the present invention includes a storage means storing first motion data that describes a posture at the time of executing a scheduled motion, and a demonstration motion by measuring the posture of the performer over time. A motion capture device that generates second motion data that describes the posture when executing the motion over time, and a motion in which the first motion data and the second motion data are integrated and the demonstration motion is superimposed on the scheduled motion Motion data integration means for creating third motion data describing the posture of the robot over time, and correction means for correcting the third motion data integrated by the motion data integration means into motion data that can be executed by the articulated robot. Yes.
The performer here is not necessarily limited to a person, and may be, for example, an animal or a robot. An example of the scheduled action is an action of walking by moving the limbs normally. An example of the demonstration operation is an operation of reaching out and shaking. In this case, an operation of extending and shaking hands while walking is obtained by superimposing a demonstration operation of extending and shaking hands on a scheduled operation of walking with limbs moving normally. When the demonstration motion is a posture change of a specific part of the body, the scheduled motion data may describe the posture of the same part over time. In this case, the operation in which the demonstration operation is superimposed on the scheduled operation refers to an operation in which the scheduled operation for the same part is replaced with a demonstration operation for the same part. Alternatively, when the demonstration operation is a posture change of a specific part of the body, the scheduled action data may not include data describing the posture of the same part over time. In this case, the operation in which the demonstration operation is superimposed on the scheduled operation is obtained by adding the demonstration operation related to the part not scheduled to the scheduled operation to the scheduled operation. In the case of the former, the demonstration operation is superimposed on the scheduled motion by replacing the motion data related to the part to which the demonstration motion has priority among the first motion data describing the scheduled motion with the second motion data describing the demonstration motion. The third motion data describing the performed motion is obtained. In the latter case, an operation in which the demonstration operation is superimposed on the scheduled motion by adding the second motion data describing the demonstration motion related to the part not included in the scheduled motion to the first motion data describing the scheduled motion. Third motion data to be described is obtained. When integrating the first motion data and the second motion data, it means both replacing the first motion data partially with the second motion data, or adding the second motion data to the first motion data. To do.

According to the motion data creation apparatus of the present invention, motion data for executing a motion in which a demonstration motion is superimposed on a scheduled motion can be created. For example, an operation that moves over a wide range that is difficult to measure with a motion capture device is a scheduled operation, and by preparing operation data that realizes the operation, it is possible to eliminate the need to prepare a large-scale motion capture device. On the other hand, it can be easily demonstrated if it is demonstrated, but on the other hand, it is relatively easy to obtain the operation data for the detailed operation that is very troublesome when preparing the operation data.
According to the motion data creation device of the present invention, since motion data can be obtained by integrating both motion data, motion data can be obtained relatively easily by using a motion capture device on more occasions. You can enjoy the advantage of being able to create.

Preferably, the motion data integration unit selects either the first motion data or the second motion data for each part of the multi-joint robot, and integrates the motion data selected for each part.
For example, the first action data is selected for a part other than the hand, the second action data is selected for the hand, and the first action data for the part other than the hand and the second action data for the hand are integrated. In this case, motion data for executing a scheduled motion for a part other than the hand and performing a demonstration motion for the hand is obtained.
When the priority relationship between the scheduled action and the demonstration action is selected by paying attention to the part, the action in which the demonstration action is superimposed on the scheduled action can be easily specified.

The operation data creating apparatus preferably includes a display unit that reproduces and displays the operation described by the operation data corrected by the correcting unit.
Thus, the performer can adjust the operation to perform while confirming the operation based on the generated operation data, and can more easily generate operation data that realizes the intended operation.

The motion data creating apparatus preferably includes teaching means for teaching the articulated robot the motion data corrected by the correcting means.
As a result, it is possible to operate the articulated robot almost in real time while creating operation data.

The operation data creating apparatus preferably includes a storage unit that stores the operation data corrected by the correction unit.
Thereby, the created operation data can be saved, and new operation data can be created using the created operation data.

The technique of the present invention is also embodied in a method for creating motion data that describes the posture of an articulated robot over time. In this method, the first motion data describing the posture at the time of executing the scheduled motion is stored, and the demonstration posture is executed by measuring the posture of the performer with the motion capture device over time. A step of collecting second motion data describing the posture of the time over time, and the posture of the first motion data and the second motion data integrated to perform the motion in which the demonstration motion is superimposed on the scheduled motion over time And a step of correcting the third motion data into motion data executable by the articulated robot.
According to this method, it is possible to create motion data without using the motion capture device for operations that are not advisable by the motion capture device, using motion capture devices for operations that are best suited for the motion capture device. Become.

  According to the present invention, it is possible to easily create operation data, and it is possible to cause the robot to execute various operations.

First, main forms of the embodiments described below are listed.
(Mode 1) The robot is a humanoid robot.
(Mode 2) The motion data creation device includes a storage unit that stores a plurality of types of first motion data.
(Mode 3) The operation data creation apparatus includes a network connection unit that inputs first operation data and the like via a network.

Example 1
Embodiments of the present invention will be described with reference to the drawings. FIG. 1 schematically shows the configuration of a robot system described in this embodiment. The robot system includes an operation data creation device 10 and a robot 102.
The motion data creation device 10 is a device that creates motion data taught to the robot 102. The motion data is data that describes the posture of the robot 102 over time during execution of a predetermined motion. The motion data describes, for example, the posture of the robot 102 at each time, describes changes in the posture of the robot 102 over time, and describes the motion of the robot 102. Note that data describing the posture of the robot 102 is referred to as posture data. The motion data can be composed of posture data arranged in time series. The motion data can also be composed of posture data at the start of the motion and data describing the posture change over time.
The motion data creation device 10 and the robot 102 are configured to be capable of wireless communication. The robot 102 inputs the motion data from the motion data creation device 10 and sequentially adjusts the posture described by the motion data to execute the motion described by the motion data.

  As shown in FIG. 1, the robot 102 is a humanoid robot made to imitate a person. The robot 102 includes a plurality of parts, and mainly includes a head 104, a torso 106, a right arm 108, a left arm 110, a left leg 112, a right leg 114, and the like. The head 104, the right arm portion 108, the left arm portion 110, the left leg portion 112, and the right leg portion 114 are swingably connected to the trunk portion 106. The right arm unit 108, the left arm unit 110, the left leg unit 112, and the right leg unit 114 are configured by connecting a plurality of links via a plurality of joints. For example, in the right leg 114, the thigh link 114b is connected to the torso 106 via the hip joint 114a, the crus link 114d is connected to the thigh link 114b via the knee joint 114c, and the ankle joint is connected to the crus link 114d. A foot link 114f is connected via 114e. The robot 102 is an articulated robot in which a plurality of links are connected via a plurality of joints. The robot 102 includes an actuator that adjusts the joint angle of each joint, and various operations can be executed by adjusting the joint angle of each joint. For example, the robot 102 can walk biped by swinging the left and right legs 112 and 114. Therefore, it can be said that the robot 102 is a legged mobile robot.

As shown in FIG. 1, the motion data creation device 10 includes a motion capture device 20, a computer 30, and a display device 40.
The motion capture device 20 includes a plurality of sensors 22 and a processing device 24 that processes output signals of the sensors 22 group. As shown in FIG. 1, a group of sensors 22 (indicated by black circles in the figure) are sensors for detecting the posture of the performer H, and can be attached to each part of the performer H. The processing device 24 creates posture data describing the posture of the performer H based on the output signal of the sensor 22 group. The motion capture device 20 detects the posture of the performer H over time, and describes the detected posture (corresponding to the motion data of the robot 102 and corresponds to the second motion data in this specification). Create over time. As the motion capture device 20, a general-purpose device can be used. For example, an optical device, a magnetic device, a device that measures acceleration, a device that measures joint angles, or the like can be used.

  As will be described later in detail, the computer 30 creates motion data that can be executed by the robot 102 using motion data taught from the motion capture device 20. The computer 30 can display the model M of the robot 102 on the display device 40, and can cause the model M of the robot 102 to execute the operation data created. The computer 30 can also teach the created motion data to the robot 102. The computer 30 can also output the created operation data to the outside via, for example, the recording medium 4 or the network 6. The processing by the computer 30 is performed in almost real time, and the operation of the performer H is reflected in the operation of the robot model M and the actual robot 102 in almost real time.

  FIG. 2 is a block diagram functionally showing the configuration of the robot system of the present embodiment. As shown in FIG. 2, the computer 30 functionally includes a scheduled motion data storage unit 52, a robot data storage unit 54, a data selection unit 56, a data integration unit 58, a data correction unit 60, and a network connection. A unit 62, a data storage unit 64, a robot model creation unit 66, a data transmission unit 68, and the like are configured. Each of these components is configured by hardware, software, or the like included in the computer 30.

  The scheduled motion data storage unit 52 stores motion data (corresponding to the first motion data in this specification) that can be executed by the robot 102 and that realizes the motion scheduled for the robot 102. The data storage unit 52 can store operation data using, for example, the recording medium 4 or the like. The scheduled motion data storage unit 52 stores, for example, walking motion data used for the robot 102 to walk. Here, the operation data used by the robot 102 will be described with reference to FIG. FIG. 3A schematically shows a configuration of walking motion data 80 used by the robot 102. As shown in FIG. 3A, the walking motion data 80 includes head motion data 84 describing the posture of the head 104 of the robot 102 over time and the posture of the torso 106 over time. Torso motion data 86, right arm motion data 88 describing the posture of the right arm portion 108 over time, left arm motion data 90 describing the posture of the left arm portion 110 over time, left leg The left leg motion data 92 describing the posture of the unit 112 over time, the right leg motion data 94 describing the posture of the right leg unit 114 over time, and the like. The head movement data 84, the right arm movement data 88, and the left arm movement data 90 describe the joint angle of each joint that determines the posture of each part over time. On the other hand, the torso motion data 86, the left leg motion data 92, and the right leg motion data 94 describe the position and orientation of the representative point of each part over time. When the data corresponding to the time t is read from the operation data 84, 86, 88, 90, 92, 94 of each part as shown in FIG. 3 (a), the robot 102 at the time t as shown in FIG. 3 (b). Can be determined.

The robot data storage unit 52 in FIG. 2 stores robot data describing the specifications of the robot 102 and the like. The robot data describes, for example, the link configuration of the robot 102, the length of each link, the allowable range of joint angles of each joint, the weight of each link and mounted equipment (computer or battery), and the like. In the robot data, information sufficient to calculate a static model or a dynamic model of the robot 102 is described.
The network connection unit 62 is connected to an external computer or the like via the network 6 and can input operation data or the like from the external computer or the like.

The data selection unit 56 inputs operation data from the scheduled operation data storage unit 52 or the motion capture device 20 for each part of the robot 102. Whether the data selection unit 56 inputs the scheduled motion data storage unit 52 or the motion capture device 20 can be set in advance for each part.
The data integration unit 58 integrates the operation data for each part selectively input by the data selection unit 56 to create operation data describing the operation of the entire robot 102.
The data correction unit 60 corrects the integrated operation data created by the integration of the data integration unit 58 into operation data that can be executed by the robot 102. The motion data input from the motion capture device 20 is not necessarily motion data that can be executed by the robot 102. Further, even if each piece of operation data is executable data, the operation data obtained by integrating them is not always executable. The data correction unit 60 refers to the robot data stored in the robot data storage unit 54 and corrects the integrated operation data created by the data integration unit 58. In this correction, for example, the joint angle of each joint is corrected within an allowable range, or the position and posture of the body portion 106 are corrected so that the dynamic balance of the robot 102 is maintained.

The robot model creation unit 66 displays the model M of the robot 102 on the display device 40 by executing a three-dimensional computer graphics process. The robot model creation unit 66 creates the model M of the robot 102 by referring to the robot data stored in the robot data storage unit 54 and inputs the operation data modified by the data modification unit 60 to input the robot 102. Calculate the posture of the model M. As a result, the model M of the robot 102 taking the posture described by the motion data corrected by the data correction unit 60 is displayed on the display device 40. The displayed model M reproduces the motion by changing the posture with time because the posture described by the motion data changes with time.
The data storage unit 64 stores the operation data corrected by the data correction unit 60. The data storage unit 64 stores posture data in time series by sequentially storing posture data together with time (elapsed time from the start of processing) data. That is, the operation data created by the operation data creation device 10 is stored in the data storage unit 64.
The data transmission unit 68 transmits the operation data corrected by the data correction unit 60 to the robot 102. Thereby, the motion data created by the motion data creation device 10 can be taught to the robot 102 in real time.

FIG. 4 is a flowchart showing an operation flow of the operation data creation apparatus 10. The processing flow shown in FIG. 4 is repeatedly executed mainly by the computer 30 every operation cycle Δt. Hereinafter, the operation of the operation data creation device 10 will be described along the processing flow shown in FIG. Here, the case where the robot 102 is walked while swinging the left arm part 110 over the head will be described as an example. FIG. 5 schematically shows changes in posture data processed in the processing flow shown in FIG.
Prior to the processing flow of FIG. 4, the walking motion data 80 is stored in the scheduled motion data storage unit 52, and the performer H is demonstrating the motion of shaking his left hand. The performer H does not have to walk, but can stay on the spot and shake his left hand. Since the performer H remains on the spot, the posture of the performer H can be easily measured by the motion capture device 20.

In step S2, the data selection unit 56 uses the left arm unit 110 in the posture data (A in FIG. 5) corresponding to the time t from the walking motion data 80 stored in the scheduled motion data storage unit 52. Other posture data (C in FIG. 5) is input. That is, since the left arm motion data 90 (see FIG. 3) is not used, it is not always necessary to prepare the left arm motion data 90.
In step S4, the data selection unit 56 receives the posture data (D in FIG. 5) related to the left arm 110 of the robot 102 from the measured posture data (B in FIG. 5) of the performer H from the motion capture device 20. ). That is, since data that is not related to the posture of the left arm 110 is not used, it is not always necessary to measure the posture of the performer H that is not related to the posture of the left arm 110 with the motion capture device 20.
In step S6, the data integration unit 58 receives the partial posture data (C in FIG. 5) input from the walking motion data 80 stored in the scheduled motion data storage unit 52 and the portion input from the motion capture device 20. The integrated posture data (E in FIG. 5) describing the posture of the entire robot 102 is created by integrating the general posture data (D in FIG. 5). The integrated posture data E created by the data integration unit 58 is a simple integration of the partial posture data C and D, and does not require any special calculation. Therefore, it is not guaranteed whether the posture data E created by the data integration unit 58 can be executed by the robot 102.
In step S8, the data correction unit 60 corrects the integrated posture data (E in FIG. 5) created by the data integration unit 58 into posture data (F in FIG. 5) that can be executed by the robot 102. The data correction unit 60 corrects the posture data to be dynamically executable by the robot 102 in consideration of the posture data and the like in the previous operation cycle t−Δt.

In step S <b> 10, the robot model creation unit 66 displays on the display device 40 a robot model M that takes the posture of the posture data corrected by the data correction unit 60. The robot model creation unit 66 inputs the posture data corrected by the data correction unit 60 (F in FIG. 5), and displays the robot model M that takes the posture described by the corrected posture data F.
In step S <b> 12, the data transmission unit 68 transmits the corrected posture data (F in FIG. 5) corrected by the data correction unit 60 to the robot 102 and teaches it. The robot 102 that has received the corrected posture data F takes the posture described by the corrected posture data F. Since the corrected posture data F is corrected to posture data that can be executed by the robot 102, the robot 102 does not cause a processing error or falls. The robot 102 and the model M displayed on the display screen 40 take the same posture.
In step S12, the data storage unit 64 stores the corrected posture data F corrected by the data correction unit 60 in association with the time data.

By repeatedly executing the processing flow of FIG. 4 every operation period Δt, the robot 102 walks while swinging the left arm portion 110 over the head, and the robot model M that performs the same operation as the robot 110 is displayed on the display device 40. The Even when the robot 102 moves out of the field of view of the performer H, the performer H can observe the robot model M of the display device 40 and can adjust his performance.
The data transmission unit 68 teaches the robot 102 the corrected posture data over time for each movement period Δt, thereby teaching the robot 102 data (that is, motion data) describing the change in posture over time. .
The data storage unit 64 stores and saves corrected posture data (that is, motion data) arranged with time.

Specific examples of the present invention have been described in detail above, but these are merely examples and do not limit the scope of the claims. The technology described in the claims includes various modifications and changes of the specific examples illustrated above.
The technical elements described in this specification or the drawings exhibit technical usefulness alone or in various combinations, and are not limited to the combinations described in the claims at the time of filing. In addition, the technology illustrated in the present specification or the drawings achieves a plurality of objects at the same time, and has technical utility by achieving one of the objects.

The schematic diagram which shows the external appearance of a robot system. The block diagram which shows the functional structure of an operation | movement data creation apparatus. The figure which shows the structure of operation | movement data typically. The flowchart which shows the flow of operation | movement of an operation | movement data creation apparatus. The figure which shows the transition of attitude | position data typically.

Explanation of symbols

10. Motion data creation device 20. Motion capture device 30. Computer 52. Scheduled motion data storage unit 54 Robot data storage unit 56 Data selection unit 58 Data integration unit 60 Data correction unit 62..Network connection unit 64..Data storage unit 66..Robot model creation unit 68..Data transmission unit 102..Robot 106..Robo body 110..Robot left arm H..Performer

Claims (6)

An apparatus for creating motion data that describes the posture of an articulated robot over time,
Storage means for storing first motion data that describes the posture when performing the scheduled motion over time;
A motion capture device that measures the posture of the performer over time and generates second motion data that describes the posture when performing the demonstration operation over time;
Action data integration means for integrating the first action data and the second action data, and creating third action data that describes the posture over time when executing the action in which the demonstration action is superimposed on the scheduled action;
Correction means for correcting the third motion data integrated by the motion data integration means into motion data executable by the articulated robot;
An operation data creation device comprising:   The motion data integrating means selects either the first motion data or the second motion data for each part of the articulated robot, and integrates the motion data selected for each part. Data creation device.   3. The operation data creating apparatus according to claim 1, further comprising display means for reproducing and displaying the operation described by the operation data corrected by the correcting means.   4. The motion data creating apparatus according to claim 1, further comprising teaching means for teaching the motion data corrected by the correcting means to the articulated robot.   5. The operation data creating apparatus according to claim 1, further comprising storage means for storing operation data corrected by the correction means. A method of creating motion data describing the posture of an articulated robot over time,
Storing first motion data describing the posture at the time of executing the scheduled motion over time;
Measuring the posture of the performer over time with a motion capture device and collecting second motion data describing the posture over time when performing the demonstration motion;
Integrating the first motion data and the second motion data, and creating third motion data that describes the posture over time when executing the motion in which the demonstration motion is superimposed on the scheduled motion;
Correcting the third motion data into motion data executable by the articulated robot;
An operation data creation method comprising:

Images