JP2008254103A - Presenter action reproducing robot, and controlling method and controlling program of presenter action reproducing robot - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To obtain a presenter action reproducing robot which can make gestures capable of correctly informing an user about an intention of a data explainer, and a controlling method and a controlling program of the presenter action reproducing robot. <P>SOLUTION: An action pattern searching portion 10 is provided, which searches an action pattern corresponding to the action of a presenter analyzed by an action analyzing portion 3 and corresponding to a key word detected by a key word detecting portion, from action patterns of a robot memorized in an action pattern memory portion 9. A body controlling portion 12 controls a body according to the action pattern searched by the action pattern searching portion 10. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to, for example, a presenter motion reproduction robot that reproduces the motion of a presenter that explains a document and the like, a control method for a presenter motion reproduction robot, and a control program.

In Patent Document 1 below, a motion analysis device that detects the motion of the inspection engineer's head is mounted, and the head of the robot is referred to as the inspection engineer's head by referring to the detection result of the motion analysis device. A technique for controlling the head of a robot to perform the same operation is disclosed.
In this robot, the motion analysis device detects the movement of the head of the inspection engineer, but does not analyze the voice of the inspection engineer.
For this reason, the movement of the robot depends only on the movement of the head of the inspection engineer, and there is no correspondence with the voice of the inspection engineer.

JP 2005-118953 A (paragraph numbers [0014] to [0025], FIG. 1)

  Since the conventional robot is configured as described above, the head is controlled so as to perform the same operation as the head of the inspection engineer, but the head is controlled regardless of the voice of the inspection engineer. For this reason, even if the user looks at the operation of the robot, there is a problem that the intention of the operation may not be easily grasped.

  The present invention has been made in order to solve the above-described problems. A presenter motion reproduction robot capable of performing gestures that can accurately convey the intention of a material explainer to a user, and control of the presenter motion reproduction robot The object is to obtain a method and a control program.

  The presenter motion reproduction robot according to the first aspect of the present invention corresponds to the motion analysis means for analyzing the motion of the document explainer, the speech recognition means for recognizing the speech of the document explainer, and the motion and the speech of the document explainer. Corresponding to the motion of the material explainer analyzed by the motion analysis means out of the robot motion storage means storing the robot motion and the robot motion stored in the robot motion storage means, and by the voice recognition means A robot motion search means for searching for a robot motion corresponding to the voice of the recognized document explainer and a body control means for controlling the body according to the robot motion searched by the robot motion search means are provided. .

  According to the first aspect of the invention, there is an effect that it is possible to perform gestures that can accurately convey the intention of the material explainer to the user.

  According to a second aspect of the present invention, there is provided a presenter motion reproduction robot that includes a robot motion storage means for storing a robot motion corresponding to a specific keyword as a robot motion corresponding to a voice of a document explainer, and a voice of the document explainer. A voice recognition means for detecting a specific keyword from the robot, and a specific keyword corresponding to the action of the material explainer and detected by the voice recognition means from the robot actions stored in the robot action storage means And a robot motion search means for searching for a robot motion corresponding to.

  According to the second aspect of the present invention, it becomes possible to associate the motion of the presenter motion reproduction robot with a specific keyword, and to perform gestures that can accurately convey the intention of the material explainer to the user. The effect that can be obtained.

  In the presenter motion reproduction robot according to the third aspect of the present invention, when a specific keyword is not detected by the voice recognition means, the robot motion search means is selected from the robot motions stored in the robot motion storage means by the document explainer. The robot motion corresponding to the motion is searched.

  According to the third aspect of the present invention, even if a specific keyword is not detected, an effect of reproducing the operation of the material explainer can be obtained.

  According to a fourth aspect of the present invention, there is provided a control method for a presenter motion reproduction robot, wherein the motion analysis means analyzes the motion of the document explainer, and the speech recognition means recognizes the speech of the document explainer. The robot motion search means corresponds to the motion of the material explainer analyzed by the motion analysis means from the robot motion storage means in which the robot motion corresponding to the motion and voice of the material explainer is stored, and the voice recognition means A robot motion search step for searching for a robot motion corresponding to the voice of the document explainer recognized by the robot, and a body control step for the body control means to control the body according to the robot motion searched by the robot motion search means. It is a thing.

  According to the fourth aspect of the present invention, there is an effect that it is possible to perform gestures that can accurately convey the intention of the material explainer to the user.

  A control program for a presenter motion reproduction robot according to a fifth aspect of the present invention is a motion analysis processing procedure for analyzing a motion of a material explainer, a speech recognition processing procedure for recognizing a speech of the material explainer, and a motion of the material explainer And from the robot motion storage means storing the robot motion corresponding to the voice, the voice of the material explainer corresponding to the motion of the material explainer analyzed by the motion analysis processing procedure and recognized by the voice recognition processing procedure. Are provided with a robot motion search processing procedure for searching for a robot motion corresponding to the above and a body control processing procedure for controlling the body according to the robot motion searched by the robot motion search processing procedure.

  According to the fifth aspect of the present invention, there is an effect that it is possible to perform gestures that can accurately convey the intention of the material explainer to the user.

  According to the present invention, among the robot motions stored in the robot motion storage means, the material explainer's motion corresponding to the motion of the material explainer analyzed by the motion analysis means and recognized by the voice recognition means. The robot motion search means for searching for the robot motion corresponding to the voice is provided, and the body control means is configured to control the body according to the robot motion searched by the robot motion search means. There is an effect that gestures that can be conveyed to the user can be performed.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing a presenter motion reproduction robot according to Embodiment 1 of the present invention. In FIG. 1, a camera 1 photographs a presenter who is a material explainer, and stores an image of the presenter in a video storage unit 2. .
The video storage unit 2 is a memory that stores the video of the presenter output from the camera 1.
The operation analysis unit 3 is composed of, for example, a semiconductor integrated circuit board on which a CPU or the like is mounted. The operation analysis unit 3 captures a change in the video stored in the video storage unit 2 and performs a process of analyzing the operation of the presenter. .
The camera 1, the video storage unit 2, and the motion analysis unit 3 constitute a motion analysis unit.

The keyword storage unit 4 is a memory that stores a keyword to be detected from the presenter's voice.
The microphone 5 collects the presenter's voice and outputs the voice signal to the voice storage unit 6.
The audio storage unit 6 is a memory that stores an audio signal output from the microphone 5.

The voice recognition processing unit 7 is composed of, for example, a semiconductor integrated circuit board on which a CPU or the like is mounted. The voice recognition processing unit 7 analyzes a voice signal stored in the voice storage unit 6 and performs a process of identifying the presenter's voice. To do.
The keyword detection unit 8 is composed of, for example, a semiconductor integrated circuit board on which a CPU or the like is mounted, and detects a keyword stored in the keyword storage unit 4 from the voice identified by the voice recognition processing unit 7. Perform the process.
The keyword storage unit 4, the microphone 5, the voice storage unit 6, the voice recognition processing unit 7, and the keyword detection unit 8 constitute voice recognition means.

The motion pattern storage unit 9 is a memory that stores the motion of the presenter and the motion pattern (robot motion) corresponding to a specific keyword. The motion pattern storage unit 9 constitutes a robot motion storage means.
The operation pattern search unit 10 is composed of, for example, a semiconductor integrated circuit board on which a CPU or the like is mounted. The presenter analyzed by the operation analysis unit 3 from the operation patterns stored in the operation pattern storage unit 9. And a process of searching for an operation pattern corresponding to the keyword detected by the keyword detection unit 8 is performed. Note that the motion pattern search unit 10 constitutes a robot motion search means.
The operation pattern storage unit 11 is a memory for storing the operation pattern searched by the operation pattern search unit 10.

The body control unit 12 is composed of, for example, a semiconductor integrated circuit board on which a CPU or the like is mounted, and an operation stored in the operation pattern storage unit 11 in synchronization with a timing signal output from the timing generation unit 14. A process of controlling the body according to the pattern is performed. The body control unit 12 constitutes body control means.
The audio reproduction unit 13 is constituted by, for example, an audio device and the like, and is a process for reproducing the presenter's audio in accordance with the audio signal stored in the audio storage unit 6 in synchronization with the timing signal output from the timing generation unit 14. To implement.
When the timing generation unit 14 receives a robot operation start request from the outside, the timing generation unit 14 establishes a synchronization between the body control unit 12 and the sound reproduction unit 13 (for example, a pulse signal of a predetermined frequency, a start trigger signal, etc.). Is output to the body control unit 12 and the sound reproduction unit 13.

FIG. 2 is a block diagram showing the motion analysis unit 3 of the presenter motion reproduction robot according to Embodiment 1 of the present invention. In FIG. 2, the change site detection unit 21 detects changes in the video stored in the video storage unit 2. Capture the presenter's moving parts.
The change site analysis unit 22 performs a process of analyzing the movement (for example, the moving direction) of the site detected by the change site detection unit 21.

FIG. 3 is a block diagram showing the speech recognition processing unit 7 of the presenter motion reproduction robot according to the first embodiment of the present invention. In FIG. 3, the feature extraction unit 31 corresponds to the speech signal stored in the speech storage unit 6. Thus, for example, by performing LPC analysis, 34-dimensional feature parameters including logarithmic power, 16th-order cepstrum coefficient, Δlogarithmic power, and Δ16th-order cepstrum coefficient of the speech signal are extracted.
The HMM memory 32 is composed of a hidden Markov model (a hidden Markov model is composed of a plurality of states and arcs indicating transitions between the states. Each arc has a transition probability between states and an input code (feature parameter). The output probability is stored).
The phoneme matching unit 33 performs phoneme matching processing using a hidden Markov model stored in the HMM memory 32, thereby generating phoneme data from the feature parameters extracted by the feature extraction unit 31.

The language model storage unit 34 is a memory that stores a statistical language model.
The speech recognition unit 35 refers to the statistical language model stored in the language model storage unit 34 and executes, for example, the “One Pass DP algorithm”. That is, the phoneme data generated by the phoneme matching unit 33 is processed from left to right without backtracking, and a speech recognition process is performed to determine a word with a higher probability of occurrence as a speech recognition result (presenter's speech). .

FIG. 4 is a front view showing the presenter motion reproduction robot according to Embodiment 1 of the present invention. FIG. 5 is a side view showing the presenter motion reproduction robot according to Embodiment 1 of the present invention.
4 and 5, the motion of the presenter is reproduced by moving the actuators of the upper limbs (left upper arm portion 41L, upper right arm portion 41R, left lower arm portion 42L, right lower arm portion 42R) and neck (neck joint portion 46N). An example of a presenter motion reproduction robot is shown.
4 and 5, one end of the left upper arm portion 41L of the presenter motion reproduction robot is movably attached to the left shoulder joint portion 43L, and one end of the left lower arm portion 42L is movably attached to the left elbow joint portion 44L. ing.
Under the instruction of the body control unit 12, the left shoulder joint portion 43L is a mechanical element including, for example, an actuator that rotates the left upper arm portion 41L in the direction of arrow A, and an actuator that swings the left upper arm portion 41L in the direction of arrow B.
The left elbow joint portion 44L is a mechanical element composed of, for example, an actuator that rotates the left lower arm portion 42L in the direction of arrow C under the instruction of the body control unit 12.

One end of the upper right arm portion 41R of the presenter motion reproduction robot is movably attached to the right shoulder joint portion 43R, and one end of the right lower arm portion 42R is movably attached to the right elbow joint portion 44R.
The right shoulder joint portion 43R is a mechanical element including, for example, an actuator that rotates the upper right arm portion 41R in the arrow A direction, an actuator that swings the upper right arm portion 41R in the arrow B direction under the instruction of the body control unit 12. .
The right elbow joint portion 44R is a mechanical element composed of, for example, an actuator that rotates the right lower arm portion 42R in the direction of arrow C under the instruction of the body control unit 12.

The moving rollers 45L and 45R are moving mechanisms that move the presenter motion reproduction robot under the instruction of the body control unit 12.
The neck joint portion 46N is a mechanical element composed of an actuator or the like that turns the neck of the presenter motion reproduction robot up and down or rotates left and right under the instruction of the body control unit 12.

In the example of FIG. 1, the camera 1, the motion analysis unit 3, the microphone 5, the speech recognition processing unit 7, the keyword detection unit 8, the motion pattern search unit 10, the body control unit 12, and the voice playback, which are components of the presenter motion reproduction robot. It is assumed that the unit 13 and the timing generation unit 14 are each configured by dedicated hardware. However, when the presenter motion reproduction robot is configured by a computer, the camera 1, the motion analysis unit 3, the microphone 5, A program describing the processing contents of the voice recognition processing unit 7, the keyword detection unit 8, the motion pattern search unit 10, the body control unit 12, the voice reproduction unit 13, and the timing generation unit 14 is stored in a computer memory. The CPU may execute a program stored in the memory.
FIG. 6 is a flowchart showing the processing contents of the presenter motion reproduction robot according to Embodiment 1 of the present invention.

Next, the operation will be described.
The camera 1 photographs a presenter who is a material explainer, for example, every predetermined time (for example, every 5 seconds), and stores the video of the presenter in the video storage unit 2 (step ST1).
Here, if the video shot by the camera 1 is an intermittent video, the latest video at the shooting time T is P _T , the video at the shooting time T-1 previously shot by the camera 1 is P _T-1 , and the camera 1 The video at the shooting time _TN taken N times before is expressed as _PTN .
If the video taken by the camera 1 is a continuous video, the video of the latest frame is expressed as P _T , the video before one frame is expressed as P _T-1 , and the video before N frames is expressed as P _TN .

When the camera 1 stores the video of the presenter in the video storage unit 2, the motion analysis unit 3 stores the video P _T , P _T-1 , P _T-2 ,..., P _TN stored in the video storage unit 2. The movement of the presenter is analyzed by capturing the change (step ST2).
Hereinafter, the operation analysis processing by the operation analysis unit 3 will be described in detail.
However, for simplification of description, an example in which the images P _T , P _T-1 and P _T-2 are compared as shown in FIG. 7 will be described.

The change site detection unit 21 of the motion analysis unit 3 detects changes in the images P _T , P _T-1 , and P _T-2 stored in the video storage unit 2 and detects the site of the moving presenter.
That is, the change portion detection section 21, as shown in FIG. 8, in the video P _T and the image P _T-1 of the differential image S _T (picture P _T, only the part that differs from the image P _T-1 with obtaining the image) indicating the difference image S _T-1 video P _T-1 and the image P _T-2 (in the video P _T-1, the image showing only parts that are different from the image P _T-2) Are extracted, the contours of the difference images S _T and S _T-1 are extracted, and the features of the contours are analyzed.
For example, when the motion detection target parts are the presenter's “neck” and “hand”, the change part detection unit 21 has a contour feature quantity and a preset presenter “neck” feature quantity (or “ If the difference between the two feature values is smaller than a predetermined threshold, it is determined that the moving presenter's part is the “neck” (or “hand”).

Here, although the change part detection part 21 implemented the process which extracts the outline of a difference image, the process which extracts the feature-value, etc., it showed about the part of the presenter who is moving, but it did not restrict to this Instead, for example, by using a known face recognition algorithm or the like, it is determined whether or not the difference video is a face image, thereby determining whether or not the moving presenter's part is the “neck”. It may be.
Known face recognition algorithms are disclosed in, for example, “The Institute of Electronics, Information and Communication Engineers Journal D-II vol. J88-D-II No. 8 pp. 1339-1348 2005”.

When the change part detection unit 22 detects the part of the presenter where the change part detection unit 21 is moving, the change part analysis unit 22 compares the difference images S _T and S _T-1 and analyzes the movement of the part.
That is, the change site analysis unit 22 analyzes the direction in which the site detected by the change site detection unit 21 is moving.
If the moving presenter part is the “neck”, the movement of the upper or lower direction of the neck or the rotation of the right or left direction is detected.
If the moving presenter's part is “hand”, movement of the hand in the right or left direction is detected.

The microphone 5 collects the presenter's voice in parallel with the shooting by the camera 1, and stores the voice signal in the voice storage unit 6 (step ST3).
When the microphone 5 stores the voice signal in the voice storage unit 6, the voice recognition processing unit 7 analyzes the voice signal and identifies the presenter's voice (step ST4).
Hereinafter, the voice identification processing by the voice recognition processing unit 7 will be specifically described.

The feature extraction unit 31 of the speech recognition processing unit 7 performs, for example, LPC analysis on the speech signal stored in the speech storage unit 6, so that the logarithmic power of the speech signal, the 16th-order cepstrum coefficient, Δ 34-dimensional feature parameters including logarithmic power and Δ16th-order cepstrum coefficients are extracted.
When the feature extraction unit 31 extracts a feature parameter, the phoneme matching unit 33 performs a phoneme matching process using a hidden Markov model stored in the HMM memory 32, thereby extracting the feature parameter extracted by the feature extraction unit 31. Phoneme data is generated.

When the phoneme matching unit 33 generates phoneme data, the speech recognition unit 35 refers to the statistical language model stored in the language model storage unit 34 and executes, for example, the “One Pass DP algorithm”.
That is, the speech recognition unit 35 processes the phoneme data from the left to the right without backtracking, and determines a word having a higher occurrence probability (for example, a noun or a verb) as a speech recognition result (presenter's speech). Perform voice recognition processing.

When the speech recognition processing unit 7 identifies the presenter's speech, the keyword detection unit 8 detects the keyword stored in the keyword storage unit 4 from the presenter's speech (step ST5).
For example, when keywords such as “how about”, “new product”, “congratulations” are stored in the keyword storage unit 4, the words (or combinations of words) that make up the presenter's voice, A keyword (or a combination of words) matching the keyword is detected by comparing with a keyword such as “How is it”?

When the motion analysis unit 3 analyzes the presenter's motion and the keyword detection unit 8 detects a keyword, the motion pattern search unit 10 selects a motion analysis unit from the motion patterns stored in the motion pattern storage unit 9. The motion pattern corresponding to the presenter motion analyzed in step 3 and corresponding to the keyword detected by the keyword detection unit 8 is searched (step ST6), and the motion pattern is stored in the motion pattern storage unit 11.
Here, as shown in FIG. 9, the motion pattern storage unit 9 stores motion patterns for each part of the robot, such as the right hand, left hand, and neck of the robot.

For example, when the analysis result of the motion analysis unit 3 indicates that “the right hand of the presenter is moving from right to left” and the keyword detection unit 8 detects the keyword “new product”, the motion pattern search unit 10 From the motion pattern for the right hand (see FIG. 9A), a motion pattern of “moving the robot's right hand from right to left and then turning the right hand toward the product” is searched.
When the analysis result of the motion analysis unit 3 indicates that “the presenter's left hand is moving from left to right” and the keyword detection unit 8 detects the keyword “how?”, The motion pattern search unit 10 The motion pattern for “moving the left hand from the left to the right after turning the robot's left hand toward the customer” is searched from the motion patterns for use (see FIG. 9B).

However, if no keyword is detected by the keyword detection unit 8, an operation pattern corresponding to the presenter operation analyzed by the operation analysis unit 3 is searched from the operation patterns stored in the operation pattern storage unit 9.
For example, if the analysis result of the motion analysis unit 3 indicates that “the presenter's right hand is moving from right to left”, but the keyword detection unit 8 does not detect a keyword, the motion pattern search unit 10 is for the right hand. An operation pattern “move the right hand of the robot from right to left” is searched from the operation patterns (see FIG. 9A).
Further, the analysis result of the motion analysis unit 3 indicates that “the presenter's left hand is moving from left to right”. However, if the keyword detection unit 8 does not detect a keyword, the motion pattern search unit 10 is for the left hand. An operation pattern “move the left hand of the robot from the left to the right” is searched from the operation patterns (see FIG. 9B).

  Here, since the analysis result of the motion analysis unit 3 indicates the motion of “presenter's right hand” or “presenter's left hand”, the motion pattern search unit 10 performs the motion pattern of “presenter's right hand” or “presenter's left hand”. For example, when the analysis result of the motion analysis unit 3 indicates movements of a plurality of parts simultaneously such as “presenter's right hand”, “presenter's left hand”, “presenter's neck”, etc. The operation pattern search unit 10 simultaneously searches for operation patterns such as “presenter's right hand”, “presenter's left hand”, and “presenter's neck”.

  When the motion pattern search unit 10 stores the motion pattern in the motion pattern storage unit 11 and receives a robot motion start request from the outside, the timing generation unit 14 performs the robot control by the body control unit 12 and the voice playback unit 13. In order to synchronize the sound reproduction, a timing signal (for example, a pulse signal of a predetermined frequency, a start trigger signal, etc.) is output to the body control unit 12 and the sound reproduction unit 13.

When body controller 12 receives the timing signal from timing generator 14, body controller 12 controls the body according to the motion pattern stored in motion pattern storage 11 in synchronization with the timing signal.
That is, the body control unit 12 outputs a control signal that causes the robot to operate according to the operation pattern to the actuator of the robot (step ST7).
For example, if the movement pattern is “move the robot's right hand from right to left”, the control signal is output to the actuators related to the right shoulder joint portion 43R and the right elbow joint portion 44R, thereby moving the robot's right hand from the right. Move to the left.

  When receiving the timing signal from the timing generation unit 14, the audio reproduction unit 13 reproduces the presenter's audio according to the audio signal stored in the audio storage unit 6 in synchronization with the timing signal.

As apparent from the above, according to the first embodiment, the motion corresponding to the presenter analyzed by the motion analysis unit 3 among the motion patterns of the robot stored in the motion pattern storage unit 9, and An operation pattern search unit 10 for searching for an operation pattern corresponding to the keyword detected by the keyword detection unit 8 is provided, and the body control unit 12 controls the body according to the operation pattern searched by the operation pattern search unit 10. Since it comprised, there exists an effect which can perform the gesture which can tell a user's intention correctly to a user.
That is, since the motion pattern search unit 10 searches not only the presenter's motion analyzed by the motion analysis unit 3 but also the motion pattern corresponding to the keyword detected by the keyword detection unit 8, the motion of the robot is controlled. As a result, it is possible to associate with a specific keyword, and as a result, it is possible to realize a gesture that can accurately convey the presenter's intention to the user.

  Further, according to the first embodiment, when a specific keyword is not detected by the keyword detection unit 8, an analysis is performed by the motion analysis unit 3 from the robot motion patterns stored in the motion pattern storage unit 9. Since the operation pattern corresponding to the operation of the presenter is searched, there is an effect that the operation of the presenter can be reproduced even when the keyword detection unit 8 does not detect a specific keyword.

In the first embodiment, the motion analysis unit 3 analyzes the operation of the presenter at regular intervals, the speech recognition processing unit 7 recognizes the speech of the presenter at regular intervals, and the keyword detection unit 8 selects keywords. However, the present invention is not limited to this. For example, every time a processing start request signal is received from the outside, the motion analysis unit 3 analyzes the motion of the presenter, and the speech recognition processing unit 7 May be recognized, and the keyword detection unit 8 may detect the keyword.
For example, every time the presenter utters a sentence, the motion analysis unit 3 analyzes the motion of the presenter, the speech recognition processing unit 7 recognizes the speech of the presenter, and the keyword detection unit 8 detects the keyword. You may do it.

Embodiment 2. FIG.
In the first embodiment, when the motion pattern search unit 10 stores the motion pattern in the motion pattern storage unit 11 and the timing generation unit 14 receives a robot motion start request from the outside, the timing signal is transmitted to the body control unit 12 and Although what was output to the audio reproduction unit 13 has been shown (in this case, the robot control is normally performed asynchronously with the operation of the presenter after the presenter finishes the operation), the operation pattern search unit 10 operates the operation pattern. As soon as it is stored in the pattern storage unit 11, the timing generation unit 14 may output a timing signal to the body control unit 12 and the sound reproduction unit 13.
In this case, when the presenter is operating, real-time robot control that follows the operation of the presenter can be performed.

Embodiment 3 FIG.
In the first embodiment, the camera 1 stores the video of the presenter in the video storage unit 2 and the microphone 5 stores the audio signal in the audio storage unit 6. For example, the presenter is photographed by a video camera. The recorded data may be stored in the video storage unit 2 and the audio storage unit 6.
In this case, the motion analysis unit 3 acquires the presenter's video from the recorded data stored in the video storage unit 2, and the voice recognition processing unit 7 uses the presenter's audio from the recorded data stored in the audio storage unit 6. To get.

It is a block diagram which shows the presenter operation | movement reproduction robot by Embodiment 1 of this invention. It is a block diagram which shows the motion analysis part 3 of the presenter motion reproduction robot by Embodiment 1 of this invention. It is a block diagram which shows the speech recognition process part 7 of the presenter operation | movement reproduction robot by Embodiment 1 of this invention. It is a front view which shows the presenter operation | movement reproduction robot by Embodiment 1 of this invention. It is a side view which shows the presenter operation | movement reproduction robot by Embodiment 1 of this invention. It is a flowchart which shows the processing content of the presenter operation | movement reproduction robot by Embodiment 1 of this invention. It is explanatory drawing which shows the image | video image | photographed with the camera. It is explanatory drawing which shows the some difference image image | photographed with the camera. It is explanatory drawing which shows the operation | movement pattern for every part of a robot. It is a chart.

Explanation of symbols

1 Camera (motion analysis means)
2 Video storage (motion analysis means)
3. Motion analysis unit (motion analysis means)
4 Keyword storage (voice recognition means)
5 Microphone (voice recognition means)
6 Voice storage (voice recognition means)
7 Voice recognition processing part (voice recognition means)
8 Keyword detector (voice recognition means)
9 Operation pattern storage (robot operation storage means)
10 Motion pattern search unit (robot motion search means)
11 motion pattern storage unit 12 body control unit (body control means)
DESCRIPTION OF SYMBOLS 13 Voice reproduction part 14 Timing generation part 21 Change part detection part 22 Change part analysis part 31 Feature extraction part 32 HMM memory 33 Phoneme collation part 34 Language model storage part 35 Speech recognition part 41L Left upper arm part 41R Upper right arm part 42L Left lower arm part 42R Right lower arm portion 43L Left shoulder joint portion 43R Right shoulder joint portion 44L Left elbow joint portion 44R Right elbow joint portion 45L, 45R Moving roller 46N Neck joint portion

Claims (5)

A motion analysis means for analyzing the behavior of the document explainer;
A voice recognition means for recognizing the voice of the material explainer;
Robot operation storage means for storing the robot operation corresponding to the operation and voice of the material explainer;
The robot motion stored in the robot motion storage means corresponds to the motion of the material explainer analyzed by the motion analysis means, and corresponds to the voice of the material explainer recognized by the voice recognition means. Robot motion search means for searching for robot motion to perform,
A presenter motion reproduction robot comprising: body control means for controlling the body according to the robot motion searched by the robot motion search means. The robot operation storage means for storing the robot operation corresponding to the specific keyword as the robot operation corresponding to the voice of the document explainer;
The voice recognition means for detecting a specific keyword from the voice of the document explainer;
The robot operation for searching for a robot operation corresponding to the operation of the material explainer and corresponding to the specific keyword detected by the voice recognition unit from among the robot operations stored in the robot operation storage unit The presenter motion reproducing robot according to claim 1, further comprising a search unit.   The robot motion search means is characterized in that, when a specific keyword is not detected by the voice recognition means, the robot motion corresponding to the material explainer's motion is searched from the robot motions stored in the robot motion storage means. The presenter motion reproducing robot according to claim 2. A motion analysis step in which the motion analysis means analyzes the motion of the document explainer;
A voice recognition step in which the voice recognition means recognizes the voice of the material explainer;
The robot motion search means corresponds to the motion of the material explainer analyzed by the motion analysis means from the robot motion storage means in which the robot motion corresponding to the motion and voice of the material explainer is stored, and the voice A robot motion search step for searching for a robot motion corresponding to the voice of the material explainer recognized by the recognition means;
A control method for a presenter motion reproduction robot, comprising: a body control step in which the body control means controls the body according to the robot motion searched by the robot motion search means. Action analysis processing procedure to analyze the document explainer's action,
A voice recognition processing procedure for recognizing the voice of the document explainer,
Corresponding to the motion of the material explainer analyzed by the motion analysis processing procedure from the robot motion storage means storing the motion of the material explainer and the robot motion corresponding to the speech, and by the speech recognition processing procedure A robot motion search processing procedure for searching for a robot motion corresponding to the voice of the recognized document explainer;
A control program for a presenter motion reproduction robot for causing a computer to execute a body control processing procedure for controlling a body according to the robot motion searched by the robot motion search processing procedure.

Images