JP2001188551A - Device and method for information processing and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To properly generate a response sentence as an answer to a user's speech. SOLUTION: When the user speaks a prescribed command, a robot recognizes the spoken command and operates according to the recognition result. For example, the robot speaks the answer sentence as an answer to the speech of the user. In this case, the robot distinguishes proper use of words, such as 'this', 'that', and 'it' corresponding to the positional relation between the robot and user, the robot and an object, or the user and object. When the object is near the robot, the robot represents and pronounces object with a word, such as 'this box' or 'it', showing a nearby body, while when the object is at a distance, the robot represents and pronounces the object with a word, such as 'that box' or 'that ', indicating a distance body.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing apparatus and method, and a recording medium, and more particularly, to an information processing apparatus and method capable of appropriately generating a response sentence as a response to a user's utterance. And a recording medium.

[0002]

2. Description of the Related Art A robot that recognizes a command spoken by a user by using a voice recognition technique and takes various actions based on the recognition result is being put into practical use. For example, the robot utters a response sentence (message) as a response to the utterance of the user.

[0003]

However, the response sentence uttered by the robot is uniquely generated based on the content of the given dialogue, for example, the robot and the user, or the robot and the object indicated in the dialogue. It is not generated on the basis of the positional relationship with. As a result, there has been a problem that the dialog between the robot and the user becomes unnatural.

[0004] For example, when the robot utters a response sentence for confirming the content of the utterance as a response to the utterance of the user, the user is far away from the user. "
When you say, "Take that thing,"
The robot utters "Are you?" Regardless of whether the "thing" is far or near the robot. If the "thing" is far from the robot, this utterance ("that is") is appropriate, but if it is near, it is better to say "this is?" A more natural conversation.

[0005] The present invention has been made in view of such a situation, and aims to generate a response sentence according to a positional relationship.

[0006]

According to a first aspect of the present invention, there is provided an information processing apparatus, comprising: an acquisition unit for acquiring a target indicated in a speech with a robot, a robot and a user, or a positional relationship between a user and a target; If the response sentence includes a word that is properly used in accordance with the positional relationship, a selecting unit that selects a word corresponding to the positional relationship acquired by the acquiring unit, and using the word selected by the selecting unit, Generating means for generating a response sentence.

The positional relationship may be a distance between the robot and the object, a robot and the user, or a distance between the user and the object, or a direction or a direction in which the robot, the user, or the object is located.

If the utterance contains words that have been properly used in accordance with the positional relationship, the utterance further includes estimating means for estimating the positional relationship based on the words, and the acquiring means includes a position estimated by the estimating means. A process of detecting a user or a target from within a range specified by the relationship is executed, and a positional relationship can be acquired based on a detection result of the executed process.

[0009] When the response sentence is output by voice, the generation means can adjust the content of the response sentence based on the positional relationship acquired by the acquisition means.

According to a fifth aspect of the present invention, there is provided an information processing method comprising: obtaining an object designated in a utterance with a robot, a robot and a user, or obtaining a positional relationship between the user and the object; When words that can be properly used according to the positional relationship are included, a response step is performed by using a selecting step of selecting a word corresponding to the positional relationship acquired in the processing of the acquiring step, and a word selected in the processing of the selecting step. And a generating step of generating

According to a sixth aspect of the present invention, there is provided a program for a recording medium, comprising: an acquisition step of acquiring an object indicated in a utterance with a robot, a robot and a user, or a positional relationship between a user and an object; If a word that is properly used in accordance with the positional relationship is included, a response step is performed by using the word selected in the processing of the selecting step and a selecting step of selecting a word corresponding to the positional relationship acquired in the processing of the acquiring step. And a generating step of generating a sentence.

An information processing apparatus according to claim 1,
In the information processing method according to the above, and the program of the recording medium according to the sixth aspect, the target indicated in the utterance with the robot, the robot and the user, or the positional relationship between the user and the user is acquired and generated. When the response sentence includes a word that is properly used in accordance with the positional relationship, a word corresponding to the acquired positional relationship is selected, the selected word is used, and a response sentence is generated.

[0013]

FIG. 1 shows an example of using a robot to which the present invention is applied. Each of the robot, the user, and the object (in this case, a box) is located at an arbitrary position within a predetermined range (for example, in a room). In addition,
The robot and the user can move within the range.

When a user speaks a predetermined command, the robot recognizes the speech by voice and operates based on the recognition result. For example, the robot utters a predetermined response sentence as a response to the utterance of the user. in this case,
The robot appropriately uses words such as "this", "that", and "it" according to the positional relationship between the robot and the user, the robot and the target, or the user and the target. For example, if the object is near the robot, the object will appear in the response text as "this box" or "this"
Etc., is indicated (indicated) by words indicating something near, while, when it is far away, it is indicated (indicated) by words indicating things that are far away, such as "that box" or "that". In the following, "this", "that",
A word that needs to be used properly depending on the positional relationship with a conversation partner or an object during a conversation, such as “it,” is referred to as a distance / direction-dependent word when it is not necessary to distinguish each word.

FIG. 2 shows an example of the external configuration of the robot shown in FIG. FIG. 3 shows an example of the electrical configuration.
Leg units 3A, 3B, 3C, and 3D are connected to the front, rear, left, and right sides of the body unit 2, respectively, and a head unit 4 and a tail unit 5 are provided at the front end and the rear end of the body unit 2, respectively. It is constituted by being connected.

The tail unit 5 is drawn out of the base unit 5B provided on the upper surface of the body unit 2 so as to bend or swing with two degrees of freedom.

The body unit 2 contains a controller 10 for controlling the entire robot, a battery 11 as a power source of the robot, and an internal sensor unit 14 including a battery sensor 12 and a heat sensor 13.

The head unit 4 includes a microphone (microphone) 15 corresponding to “ears” and a CCD corresponding to “eyes”.
(Charge Coupled Device) A camera 16, a touch sensor 17 corresponding to tactile sensation, a speaker 18 corresponding to a "mouth", and the like are arranged at predetermined positions. The head unit 4 also has a GPS (Global Positioning System) receiver 1
9, a gyrocompass 20, a geomagnetic sensor 21, and an ultrasonic sensor 22 are respectively arranged at predetermined positions.

The joint portions of the leg units 3A to 3D, the connecting portions of the leg units 3A to 3D and the body unit 2, the connecting portions of the head unit 4 and the body unit 2, and the tail unit 5 As shown in FIG. 3, actuators 3AA1 to 3AAK, 3BA1 to 3B
AK, 3CA1 to 3CAK, 3DA1 to 3DAK, 4A1
4AL, 5A1 and 5A2 are provided so that each connecting portion can rotate with a predetermined degree of freedom.

The microphone 15 in the head unit 4 collects surrounding sounds (sounds) including utterances from the user, and sends the obtained sound signals to the controller 10. CCD
The camera 16 captures an image of the surroundings, and sends the obtained image signal to the controller 10.

The touch sensor 17 is provided, for example, above the head unit 4 and “strokes” from the user.
It detects the pressure received by a physical action such as tapping or tapping, and sends the detection result to the controller 10 as a pressure detection signal.

The GPS receiver 19 receives a radio wave transmitted from a predetermined artificial satellite and sends it to the controller 10.
The gyro compass 20 and the geomagnetic sensor 21 detect the direction in which the head unit 4 is facing, and send the detection result to the controller 10. The ultrasonic sensor 22 outputs an ultrasonic wave, receives the reflected wave, and sends the reflected wave to the controller 10. In the following, when it is not necessary to distinguish each signal transmitted from the GPS receiving unit 19 to the ultrasonic sensor 22 to the controller 10, these signals are collectively referred to as a position information signal.

The battery sensor 12 in the body unit 2 detects the remaining amount of the battery 11 and sends the detection result to the controller 10 as a battery remaining amount detection signal. The heat sensor 13 detects heat inside the robot,
The detection result is sent to the controller 10 as a heat detection signal.

The controller 10 has a CPU (Central Pro
(Processing Unit) 10A, a memory 10B, and the like. The CPU 10A performs various processes by executing a control program stored in the memory 10B.

That is, the controller 10 controls the microphone 15
, A CCD camera 16, a touch sensor 17, a GPS receiver 19, a gyro compass 20, a geomagnetic sensor 21, an ultrasonic sensor 22, a battery sensor 12, and a sound signal, an image signal, a pressure detection signal, and position information provided from a heat sensor 13. Based on the signal, the remaining battery detection signal, and the heat detection signal, it is determined whether there is a surrounding situation, a command from the user, an action from the user, or the like.

Further, the controller 10 determines a subsequent action based on the determination result and the like, and based on the determined result, the actuators 3AA1 to 3AAK, 3BA1.
To 3BAK, 3CA1 to 3CAK, 3DA1 to 3DA
K, 4A1 to 4AL, 5A1, and 5A2 are driven, whereby the head unit 4 can be swung up and down, left and right, the tail unit 5 can be moved, and the leg units 3A to 3D can be moved. Drive to make the robot perform an action such as walking.

Further, the controller 10 generates a synthesized sound (a synthesized sound corresponding to the response sentence) as necessary, and supplies the synthesized sound to the speaker 18 for output.

As described above, the robot can take an autonomous action based on the surrounding situation and the like.

In the present specification, the present invention will be described by taking a dog-shaped robot as shown in FIGS. 1 and 2 as an example. However, the present invention is not limited to other forms of robot (for example, human shape). Of course can also be applied.

FIG. 4 shows an example of a functional configuration of the controller 10 of FIG. The functional configuration shown in FIG.
The CPU 10A is realized by executing a control program stored in the memory 10B.

The input processing unit 51 includes a microphone 15 and a CCD.
Camera 16, touch sensor 17, GPS receiver 19, gyro compass 20, geomagnetic sensor 21, ultrasonic sensor 2
The audio signal, the image signal, the pressure detection signal, the position information signal, and the like from the second and the like are received and transmitted to the control unit 50.

The speech recognition / analysis unit 52 includes an input processing unit 51
Performs predetermined speech recognition processing and language analysis processing on the audio signal from the microphone 15 input through
The voice signal is converted into a format (text, parse tree, or frame for dialogue) that can be processed by the dialogue manager 56 and sent to the controller 50.

The image recognizing section 53 performs an image recognizing process on an image signal input via the input processing section 51 and taken in by the CCD camera 16. Then, when the image recognition unit 53 detects, for example, “a rectangular parallelepiped object”, “fingertip”, “human face”, or the like as a result of the processing,
"There is a box", "Direction indicated by fingertip", "Face size"
Is notified to the control unit 50.

The position detecting section 54 receives a voice signal, a position information signal, and an image recognizing section 53 inputted through the input processing section 51.
For example, the position of the robot, the user, and the target object, the distance between each of them, and the direction in which the robot and the user are facing are detected using the image recognition result by, and the detection result is sent to the control unit 50. Send out.

The position detecting section 54 also detects, for example, based on the detected position of the robot, the user, or the object, for example,
A positional relationship (hereinafter, referred to as a near-far state) in properly using the distance / direction-dependent words is detected, and the detection result is notified to the control unit 50. For example, of the distance / direction-dependent words, a distance / direction-dependent word indicating a distant word should be used (hereinafter, referred to as a far state), or a distance / direction-dependent word indicating a near word. It is detected whether or not the user is in a near / far state where words are to be used (hereinafter referred to as a near state).

The user identification unit 55 extracts a voiceprint from a voice signal input via the input processing unit 51 or extracts a voice image from a face image recognized by the image recognition unit 53.
The feature amount is extracted, the user is identified based on the extracted voiceprint or feature amount, and the identification result is notified to the control unit 50.

The dialogue management unit 56 determines the action to be taken by the robot based on each signal input through the input processing unit 51, the passage of time, and the like, and uses the determined action as action command information. , To the control unit 50. For example, when the robot takes an action of uttering a response sentence, for example, a response sentence is generated using an appropriate distance / direction-dependent word corresponding to the perspective state detected by the position detection unit 54, and the robot's head The control unit 50 is notified together with information for operating the limbs and the like. For example, the control unit 50 is notified of the action command information that “Speaks“ No ”and shakes the head left and right”.

If the user's utterance subjected to the speech recognition and language analysis by the speech recognition / analysis unit 52 includes a distance / direction dependent word, the dialog management unit 56 Based on this, for example, the distance between the robot and the user, the robot and the object, or the distance between the user and the object is detected (determined).

When the action command information generated by the dialogue management unit 56 includes a command to utter a response sentence, the output generation unit 57 outputs a synthesized sound corresponding to the response sentence (for example, “No”). Is generated and output to the audio output unit 58. The output generation unit 57 also generates an operation pattern (an operation pattern for the robot to swing the head left and right) based on the action command information, and sends it to the operation control unit 59.

The audio output section 58 causes the speaker 18 to output the synthesized sound generated by the output generation section 57. Thereby, the robot utters a response sentence, for example, “No”.

The operation control section 59 determines whether or not the actuator 3 has been operated based on the operation pattern generated by the output generation section 57.
A control signal for driving AA1 to 5A1 and 5A2 is generated, and actuators 3AA1 to 5A1 and 5A
Send to 2. Thus, the actuators 3AA1 to 5A1 and 5A2 are driven according to the control signals. This allows the robot to act autonomously,
For example, shake your head left or right.

The control section 50 controls each section described above.

Next, the processing of the position detecting section 54 for detecting the positions of the robot, the user, and the object, the distance between them, and the directions of the robot and the user will be described.

First, a processing procedure for detecting the position of the robot will be described. In this case, the position detector 54
Is the GPS receiving unit 1 input through the input processing unit 51.
9, based on the radio wave from the predetermined satellite received
The latitude and longitude corresponding to the current position of the robot are calculated, and the absolute position of the robot is detected.

The position detecting section 54 is also provided within a predetermined range (for example, in a room) inputted through the input processing section 51.
The relative position of the robot in the range (in the room) is detected using the reflected wave of the ultrasonic wave transmitted from the ultrasonic sensor 22 and reflected from the object.

A process for detecting the direction in which the robot (for example, the tip of the head unit 4) faces will be described. In this case, the position detection unit 54 detects the direction (direction) in which the robot is facing, based on the detection results of the gyrocompass 20 and the geomagnetic sensor 21 input via the input processing unit 51.

Next, processing for detecting the position of the user (position viewed from the robot) will be described. Position detector 5
4 detects the relative position of the user using the reflected wave of the ultrasonic wave transmitted from the ultrasonic sensor 22 and reflected from the object identified as the user by the user identification unit 55.

The position detecting section 54 also receives the sound from the user (for example, when the user speaks), the microphone 15 input via the input processing section 51,
In this case, the direction in which the user (sound source) is located is detected based on a volume difference, a phase difference, a tone color difference, or the like of the audio signal from the plurality of microphones 15. That is, the position detection unit 54 uses the identification result by the user identification unit 55 to
It can also be determined whether the sound source is a user.

Processing for detecting the direction in which the user (for example, face) is facing (the direction viewed from the robot) will be described. The position detection unit 54 extracts a feature amount from the face image subjected to the image recognition processing by the image recognition unit 53, and detects a direction in which the user is facing based on the extracted feature amount.

The position detecting section 54 detects a change (in which direction and how much the user has moved) of the image of the user subjected to the image recognition processing by the image recognizing section 53 and turns the moving direction. Direction.

If neither the user's face image nor the moving direction can be obtained, the user's direction is obtained from the user's position obtained by the method assuming that the user is facing the robot. be able to.

Next, a process for detecting the distance between the robot and the user will be described. The position detection unit 54 detects the distance between the robot and the user by using the reflected wave of the ultrasonic wave transmitted from the ultrasonic sensor 22 and reflected from the object identified as the user by the user identification unit 55.

The position detecting section 54 is also provided with the CCD camera 16.
(In this case, a plurality of CCD cameras 16), parallax is detected from a plurality of user images captured from different angles, and a distance between the robot and the user is detected based on the parallax.

The position detector 54 further includes a CCD camera 1
6, when the user's face is imaged, the size of the user's face is detected from the image subjected to the image recognition processing by the image recognition unit 53, and the distance between the robot and the user is determined based on the detection result. To detect.

Next, the processing for detecting the position of the object will be described. The position detection unit 54 includes the image recognition unit 5
The relative position of the target object is detected using the reflected wave of the ultrasonic wave transmitted from the ultrasonic sensor 22 and reflected from the object identified as the target object by 3.

When the user's finger is being imaged by the CCD camera 16, the position detection unit 54 indicates the finger from the image subjected to the image recognition processing by the image recognition unit 53. Direction is detected, thereby detecting the direction in which the object is located.

Next, a process for detecting the distance between the robot and the object will be described. The position detection unit 54 uses the reflected wave of the ultrasonic wave transmitted from the ultrasonic sensor 22 reflected from the object recognized as the target object by the image recognition unit 53 to determine the distance between the robot and the target object. To detect.

The position detecting section 54 is also provided with the CCD camera 16.
(In this case, a plurality of CCD cameras 16), parallax is detected from a plurality of images of the object taken from different angles, and a distance to the object is detected based on the parallax.

The processing for detecting the distance between the user and the object will be described. From the position of the user detected by the method described above and the position of the object detected by the method described above, their distances are detected.

Next, based on the positions of the robot, the user, and the object detected by the methods described above and above, the robot and the user, as viewed from the robot,
Alternatively, the process of the position detection unit 54 when detecting (determining) the near / far state between the robot and the target object will be described.

As shown in FIG. 5, the position detector 54 sets the point corresponding to the position of the robot (the center position of the head unit 4 in the example of FIG. 5) as the origin, and the y-axis (FIG. For example, a two-dimensional space including a middle axis and a vertical axis) and a z axis (horizontal axis in the figure) is set in the memory 10B.
A perspective determination pattern A is formed (drawn) on the dimensional space. In this case, the perspective determination pattern A includes, for example, a point on the z-axis (a point on the plus side on the z-axis) corresponding to a position 2 m away from the robot, and a distance of 1 m, for example, in the left-right direction from the robot. A point on the y-axis (a point on the plus side and a minus side on the y-axis) corresponding to a position away from the robot, and a point on the z-axis (z (A minus point on the axis).

The position detection unit 54 converts the detected position of the user or the object into a position (coordinates) in the two-dimensional space in FIG. 5 and determines whether the coordinates are within the range of the perspective judgment pattern A. If it is determined that the robot is present in the perspective determination pattern A, it is determined that the robot and the user, or the robot and the object are in a close state, while it is determined that the robot and the user are not present in the perspective determination pattern A. Is determined to be far away.

For example, as shown in FIG. 6, when the robot and the user are close to each other (when the user
m), the coordinates of the detected position of the user are present in the perspective determination pattern A, so that the robot and the user are determined to be in a close state. on the other hand,
As shown in FIG. 7, when the robot and the user are far apart (when the user is more than 2 m ahead of the robot), the coordinates of the detected position of the user are outside the perspective determination pattern A. , The robot and the user are determined to be far away.

In this judgment, a perceptron (a neural network for judging whether the object is inside or outside the perspective judgment pattern A) is prepared, and the position (coordinates) of the user and the object with respect to the robot is in a state close to the position or coordinates. The correspondence with the distant state may be learned in advance and used.

Further, in the above description, using the perspective judgment pattern A, the robot and the user,
Alternatively, the case where the distance between the robot and the target object is determined has been described. However, the position detection unit 54 may be configured to use the two-dimensional space shown in FIG. As shown in FIG. 8, a circular perspective determination pattern B centered on a point corresponding to the position of the user is further formed on the space, and the perspective state of the user and the target when viewed from the robot ( "Is the object near the user as seen from the robot?" In this case, the perspective determination pattern B indicates a range of a circle centered on the user. This is because the range of a state close or far from the robot is not related to the orientation of the user. It is. That is, the shape does not depend on the direction.

For example, as shown in FIG. 9, a range where the perspective judgment pattern A and the perspective judgment pattern B overlap (in the figure,
When the coordinates of the position of the target object exist in the range 1), it is determined that the robot and the target object and the user and the target object are close to each other when viewed from the robot. On the other hand, as shown in FIG. 10, when there is no area where the perspective determination pattern A and the perspective determination pattern B overlap each other, the robot and the object, and the user and the object are viewed from the robot.
There is no case where it is determined that they are close to each other.

The range of the perspective judgment pattern A (FIG. 10)
The middle and the range 2) and the range of the perspective determination pattern A that do not overlap with the perspective determination pattern B (range 2 in FIG. 9)
When the coordinates of the object exist, the robot and the object are close to each other when viewed from the robot, and the user and the object are
It is determined that the state is far.

The range of the perspective judgment pattern B that does not overlap with the perspective judgment pattern A (range 3 in FIG. 9) and the range of the perspective judgment pattern B (range 3 in FIG. 10)
When the coordinates of the object exist, the robot and the object are far from each other when viewed from the robot, and the user and the object are
It is determined that the state is close. Furthermore, when the coordinates of the object exist in a range (range 4 in FIGS. 9 and 10) that is not included in both the perspective determination pattern A and the perspective determination pattern B, the robot and the object, and The user and the object are each determined to be far away.

The determination method described with reference to FIGS. 5, 8, 9, and 10 is a method for determining the distance from the viewpoint of the robot. As shown in (1), by using the perspective judgment pattern A and the perspective judgment pattern B in reverse, the perspective state of the user and the target object or the robot and the target object when viewed from the user can be determined.

In this case, as shown in FIG. 11, the position detector 54 sets the robot position (coordinates) as the origin and sets the center in the two-dimensional space including the y-axis and the z-axis passing through the origin. A perspective determination pattern B is formed, and a perspective determination pattern A centered on the position (coordinates) of the user is formed as shown in FIG.

For example, as shown in FIG. 13, when the coordinates of the position of the object exist in a range (range 1 in the figure) where the perspective judgment pattern A and the perspective judgment pattern B overlap,
From the user's point of view, it is determined that the user and the object and the robot and the object are close to each other. on the other hand,
As shown in FIG. 14, when there is no range in which the perspective determination pattern A and the perspective determination pattern B overlap, as viewed from the user,
There is no case where it is determined that the user and the target object and the robot and the target object are close to each other. FIG.
3 and 14, range 1 range 4 corresponds to the case in FIGS. 9 and 10, and the determination result when the coordinates of the object are in range 2 to range 4 is the same as that in FIGS. Therefore, description of the determination result is omitted.

In the above description, the method of determining the perspective state using the perspective determination pattern A or the perspective determination pattern B has been described. However, for example, the perspective state is determined simply based on the actual distance. You can also. For example,
When the distance between the robot and the target object or the user is less than 2 m, it can be determined that the robot is close, and when it is longer than that, it can be determined that the robot is far.

Further, when determining the perspective state based on the actual distance as described above, the distance between the robot and the object or the distance between the robot and the user is used as one variable, and the degree is determined for each of the near and far cases. A method is also possible in which the calculated degree is calculated as the distance state.

For example, the function of the solid line in FIG. 15 is prepared as the degree of closeness, and the function of the dotted line in FIG. 15 (degree of distance = 1−degree of closeness) is prepared as the degree of distance. The solid line indicates that the degree is 1 when the distance is 0 m to 2 m, the degree is 0 when the distance is 5 m or more, and that the distance is 2 m to 5 m.
It is a function in which the degree decreases linearly from 1 to 0 as the distance increases.

On the other hand, the dotted line has a distance of 0 m, contrary to the solid line.
The function is a function in which the degree is 0 from 1 m to 2 m, the degree is 1 from 5 m or more, and the degree linearly increases from 0 to 1 as the distance increases from 2 m to 5 m.

When the distance is 3 m, the degree of closeness (solid line) = ２, the degree of distance (dotted line) = 1/3, and the degree of closeness> the degree of distance , "Close state". Also, when the distance is 4 m, the degree of closeness (solid line) = 1/3 (solid line), the degree of distance (dotted line) = 2/3, and the degree of closeness <the degree of distance, It is determined that the state is “distant”.

As described above, when calculating the degree of the near state or the degree of the distant state, in addition to the distance,
For example, the orientation of the robot or the user can also be a variable, whereby the range in which the state is determined to be close is widened in the front and narrowed in the rear as in the perspective determination pattern A (for example, FIG. 5). Can also.

Next, details of the dialogue management section 56 will be described. FIG. 16 shows an example of a configuration of a part for generating a response sentence of the dialog management unit 56 and determining a perspective state.

The dialogue processing unit 111 includes, in addition to the user's utterance (for example, text) that has been subjected to the voice recognition processing and the language analysis processing by the voice recognition / analysis unit 52, and the determination result of the perspective state by the position detection unit 54. Is input via the control unit 50. The dialog processing unit 111 generates a response sentence corresponding to the content of the input utterance with reference to the interaction rule table 112 and the position / word correspondence table 113, and the response sentence includes a distance / direction dependent word. If
The dialogue processing unit 111 generates a response sentence using an appropriate distance / direction-dependent word corresponding to the result of the determination of the perspective state by the position detection unit 54.

Each signal input through the input processing unit 51, the speech recognition / analysis unit 52
, The image recognition result by the image recognition unit 53, and the like are arranged and stored as a conversation history.

The distance / word correspondence table 113 includes a distance / calling word correspondence table 121 (FIG. 17) and a distance / sword word correspondence table 122 (FIG. 18). The distance / calling word correspondence table 121 is
The words (hereinafter referred to as “calling words”) used when calling the “Hey” or “Hey” partner in response to the near state and the far state seen from the robot are shown.

For example, as shown in FIG. 6, when the user and the robot are close to each other, when one calls the other, a call word for calling a close partner such as "Hey" is used. On the other hand, as shown in FIG. 7, when the robot and the user are far from each other, when one of them calls the other, a call word for calling a distant partner such as “hey” is used. That is, the distance / calling word correspondence table 121 in FIG. 17 shows the positional relationship (perspective state) of the calling party and the corresponding relation between the calling words appropriate in the perspective state.

Distance / Shoreado Word Correspondence Table 122
Are four column-side items (column item 1 to column item 4) indicating respective combinations of the perspective state of the robot and the object and the perspective state of the user and the robot, and the words used in the utterance of the robot. A line-side item (line item 1) and a line-side item (line item 2) indicating a word used in the user's utterance are arranged. In addition,
However, the word means a word representing a predetermined object, such as "this", "that", or "it".

The data area corresponding to the column item 1 indicating the combination of the robot and the object and the user and the object being in the close state and the line item 1 indicating the words used in the utterance of the robot are stored in the data area. Is a word that indicates something close to it, such as "this", "this guy", "this", "here". The data area corresponding to the column item 1 and the row item 2 indicating the word used in the utterance of the user is also indicated by the same word area as the data area corresponding to the column item 1 and the row item 1. Have been.

As shown in FIG. 19, when viewed from the robot,
When the robot and the object, and the user and the object are close to each other, the robot and the user usually place the object (in this example, a box) in a state such as “this” indicating a nearby object. Express using words. That is, column item 1 in the distance / sword word correspondence table 122
In each data area corresponding to line item 1 and line item 2, the words used in the perspective state shown in FIG. 19 are shown.

Column item 2 of a combination in which the robot and the object are close and the user and the object are far,
In the data area corresponding to the line item 1, words such as “this”, “this”, “this”, and “here” are shown to indicate nearby things. Further, in the data area corresponding to the column item 2 and the row item 2, words that indicate distant items such as "it", "that", "the", "there" are shown.

As shown in FIG. 20, as viewed from the robot,
When the robot and the object are close to each other and the user and the object are far from each other, the robot usually expresses the object using words such as “this” indicating a nearby object. On the other hand, the user expresses the object using words, such as "the" indicating a distant object. That is, the words used in the perspective state shown in FIG. 20 are shown in the data area corresponding to the column item 2 and the row items 1 and 2.

A column item 3 of a combination in which the robot and the object are far away and the user and the object are close,
In the data area corresponding to the row item 1, the words indicating the distant things such as "it", "that", "the", "there" are shown. Item 2
In the data area corresponding to, "this", "this guy", "
Words are shown to indicate something near, such as "this" or "here".

As shown in FIG. 21, as viewed from the robot,
When the robot and the object are in a distant state and the user and the object are in a close state, usually, the robot expresses the object using words such as “the” indicating a distant object. On the other hand, the user expresses the target object using words such as “this” indicating a nearby object.
That is, the words used in the perspective state shown in FIG. 21 are shown in the data areas corresponding to the column item 3 and the row items 1 and 2.

A column item 4 of a combination in which the robot and the object, and the user and the object are respectively far from each other,
In each data area corresponding to the line item 1 and the line item 2, other words indicating a distant thing such as "that", "that guy", "that", "that there" are shown. .

As shown in FIG. 22, as seen from the robot,
When the robot and the object and the user and the object are far from each other, usually, the robot and the user
The object is expressed using words, such as "that", which indicates a distant object. That is, in each data area corresponding to column item 4 and line item 1 and line item 2, FIG.
The words are used when used in the perspective state shown in FIG.

The distance / calling word correspondence table 1
The call words shown in FIG. 21 and the word words shown in the distance / short word correspondence table 122 are words that can be properly used depending on the positional relationship with the conversation partner and the object as described above. That is, it is one of the distance / direction dependent words.

That is, the dialogue processing unit 111 sends an appropriate distance / direction-dependent word to the position / word correspondence table 113 (distance / calling word correspondence table 121, distance / A response sentence is generated by selecting from the word correspondence table 122).

When the input utterance includes a distance / direction dependent word, the dialog processing unit 111 refers to the position / word correspondence table 113 to determine the distance / direction corresponding to the distance / direction dependent word. The state is detected, and for example, the perspective state of the robot and the user, the robot and the object, or the user and the object is determined.

Next, a processing procedure when the robot utters a response sentence will be described with reference to a flowchart of FIG. As shown in FIG. 20, when the robot and the object are in a close state and the user and the object are in a distant state, the user utters “sit in front of the box” and responds thereto. In the following, an example will be described in which the robot speaks "This box" and then sits in front of the box.

In step S1, the position detector 54
Starts the process of detecting the position and orientation of the robot by the method described above and in the above, and based on the detected position and orientation of the robot, the center of the robot as shown in FIG. Is formed.

In step S2, the position detector 54
Waits until an audio signal is input from the microphone 15,
That is, the process waits until the user speaks, and when the user speaks and the voice signal is input, the process proceeds to step S3.

In step S3, the position detector 54
Detects the position of the uttering user by the method described above, detects the direction of the user by the other method, and, based on the detected position and direction of the user, as shown in FIG. Is formed.

At this time, the output generation unit 57
Required actuators 3 to 5 actuators 3AA1 to 5A1 and 5 so that the CCD camera 16 corresponding to “eyes” can capture the face of the uttered user.
An operation pattern for driving A2 is generated, and the operation control unit 5
9 is output. The operation control section 59 controls the necessary actuators 3AA1 to 5A1 and 5A2 based on the operation, and directs the CCD camera 16 to the user. This allows
An image of the uttered user is captured by the CCD camera 16.

In step S4, the user identification section 55
Identifies the uttering user. Specifically, the user identification unit 55 extracts a voiceprint from the audio signal input via the input processing unit 51, and extracts a feature amount from the image of the user that has been subjected to the image recognition processing by the image recognition unit 53. The user is identified based on the extracted voiceprint or feature amount.

Next, in step S5, the voice recognition /
The analysis unit 52 performs a voice recognition process and a language analysis process on the voice signal corresponding to the utterance of the user input in step S2.

In step S6, the dialogue management section 56
Determines in step S5 whether or not the user's utterance speech-recognized by the speech recognition / analysis unit 52 includes a distance / direction-dependent word (calling word or sword word), and includes If it is determined that it has been performed, the process proceeds to step S7. In this example, since the user's utterance includes the distance / direction-dependent word “that”,
Proceed to step S7.

In step S7, the position and orientation of the object indicated by the distance / direction dependent word are detected. Specifically, first, the dialog management unit 56 (the dialog processing unit 111)
Refers to the distance / word correspondence table 113 and detects a perspective state corresponding to the distance / direction-dependent word included in the utterance. Since the distance / direction-dependent word used in the user's utterance in this example is "that", the distance / word correspondence table 113 includes "the" of the distance / sword word correspondence table 121 in the distance / word correspondence table 113. The column item 2 corresponding to the row item 2 is referred to, and the perspective state (close state) between the robot and the object and the perspective state (distant state) between the user and the object indicated in the column item 2 are detected.

Next, the position detecting section 54 centers on the position of the robot formed in step S1 which is specified by the distance between the robot and the object detected by the dialog managing section 56 or between the user and the object. Perspective determination pattern B or perspective determination pattern A centered on the position of the user
For example, a process of detecting the position of the object designated by the distance / direction dependent word from the range described above is performed by the method described above.

In the case of this example (FIG. 20), the robot and the object are in a close state, and the user and the object are in a distant state. Therefore, in the example of FIG. In the part B, and in the example of FIG. 14, the position (coordinates) of the object exists in the perspective determination pattern B. That is, for example, an ultrasonic wave is emitted from the ultrasonic sensor 22 to the range specified by this. Processing for detecting an object is performed. Thereby, the position of the target object can be detected more efficiently by specifying the predetermined range.

When it is determined in step S6 that the distance / direction-dependent word is not included in the speech recognition result (user's utterance), or after the position of the object is detected in step S7, step S8 is performed. The processing for performing an operation as a response to the utterance of the user is executed.

Details of the processing in step S8 are shown in FIG.
4 is shown in the flowchart.

In step S21, the dialog management unit 56
Generates action command information corresponding to the user's utterance input in step S2. If the action command information includes a command to utter a response sentence including a distance / direction-dependent word, this At this point, the portion of the response sentence where the distance / direction-dependent word is used is information indicating that the distance / direction-dependent word is used (eg, an abstracted word).
It has become. In the case of this example, the action command information is generated, saying “Speaks“ distance / direction dependent word ”box and sits in front of the box”.

In step S22, the dialog management unit 56
Indicates whether the action command information generated in step S21 includes a command to utter a response sentence, and whether a distance / direction-dependent word is used in the response sentence (for example, indicates that the distance / direction-dependent word is used). Whether the abstracted words are included in the response sentence) and utter the response sentence,
When it is determined that the response sentence includes the distance / direction dependent word, the process proceeds to step S23. In the case of this example, the process proceeds to step S23.

In step S23, a distance / direction dependent word is selected. Specifically, first, as shown in FIG. 5, the position detection unit 54 forms a perspective determination pattern A centered on the position of the robot detected in step S1, and the position of the object detected in step S7. (Coordinates)
It is determined whether or not it exists in the perspective determination pattern A,
Determine the distance between the robot and the object. In this example, since the object (box) is near the robot, the coordinates of the object are determined to be present in the perspective determination pattern A, and the robot and the object are determined to be close.

Next, the dialog management unit 56 (the dialog processing unit 11)
1) stores the distance / direction-dependent words used in the response sentence generated in step S21 corresponding to the perspective state determined by the position detection unit 54 in the distance / word correspondence table 1.
Select from 13. In the case of this example, the distance / sword word correspondence table 1 in the distance / word correspondence table 113
Reference is made to 22, and “this” included in the data area of the row item 1 corresponding to the column item 1 and the column item 2 indicating that the robot and the object are close to each other is selected.

In step S24, the dialogue manager 56
Incorporates the distance / direction dependent word selected in step S23 into the "portion where distance / direction dependent word is used" in the response sentence of the action command information generated in step S21, and completes the response sentence. In this example, a response sentence indicating “is this box?” Is generated.

In step S22, the action command information includes a command to utter a response sentence, and the response sentence includes a distance /
If it is not determined that the direction-dependent word is used, or after the response sentence is completed in step S24, the process proceeds to step S25, where the action command information generated in step S21 or the response completed in step S24 The robot operates based on action command information including a command to utter a sentence. Specifically, the output generation unit 57 generates an operation pattern based on the action command information and sends the operation pattern to the operation control unit 59. The operation control unit 59 controls the actuators 3AA1 to 5A1 based on the operation pattern. And 5
A control signal for driving A2 is generated and sent to the actuators 3AA1 to 5A1 and 5A2. Thereby, the actuators 3AA1 to 5A1 and 5A2 are
Driven according to the control signal, in this example the robot walks and sits close to the object.

Thereafter, the processing ends, and the flow advances to step S9 in FIG.

In step S9, it is determined whether or not the utterance is continuously input. If it is determined that the utterance is input, the process returns to step S3, and the subsequent processing is executed. On the other hand, if it is determined that no utterance has been input, the process ends.

In the manner described above, appropriate distance / direction-dependent words are used according to the distance between the robot and the user, the robot and the object, or the distance between the user and the object. A natural dialogue takes place.

In the above description, the case where there is one object has been described as an example. However, as shown in FIG. 25 or FIG. 26, the present invention can be applied to the case where there are a plurality of objects. In such a case, for example, when the user points to "that box" and an object, the robot may say "there are multiple equivalents", "is it right or left?" Can also. Also, the robot
As shown in FIG. 26, it is possible to move to one of the objects and speak “Here?”.

In the above description, the words change as the robot moves, but the method of the present invention can correctly handle the words. That is,
At the time of FIG. 25, the user expresses the target object as “that box”, and at the time of FIG. 26, it expresses it as “it” (or “the box” or the like). By using the discrimination pattern (FIGS. 11 and 12) and the distance / sword word table 122 (FIG. 18), "it" at the time of FIG. 26 is the same as "that box" at the time of FIG. Can be recognized.

In step S7 described above, a process of detecting an object is performed within the range specified by the determined perspective state.
If the direction-dependent word is not appropriate, for example, if the object is located near the user but indicates "the box" and the object, the object is not detected. In this case, the robot may say, for example, "No equivalent found."
Can also be spoken, or the user can move closer to an object located outside the range corresponding to the determined perspective state, and say "What is this?" Further, the robot may determine that the user is lying (for example, making fun of the robot) and perform an operation corresponding to the determination. Further, when the case where the object cannot be detected continues, the range of the perspective judgment pattern B shown in FIG. 11 or the perspective judgment pattern A shown in FIG. The distance / direction-dependent words used by the user and the actual perspective can also be adapted.

In the above description, the case where the perspective state is determined by using the perspective determination pattern and an appropriate word is selected from the distance / right word correspondence table 112 based on the determination result is described. As explained,
As described with reference to FIG. 15, the degree of the near state or the degree of the far state can be used to select an appropriate word.

For example, when the distance between the robot and the object is 3 m
In the case where the distance between the user and the object is 4 m, according to FIG. 15, the degree of the distance between the robot and the object (in the figure,
The value of the dotted line corresponding to the distance 3 m) is 1/3, and the degree of the close state (the value of the solid line corresponding to the distance 3 m in the figure) is 2
/ 3. On the other hand, the degree of the distant state between the user and the object (the value of the dotted line corresponding to a distance of 4 m in the figure) is 2/3, and the degree of the close state (the value of the implementation corresponding to the distance of 4 m in the figure) Becomes 1/3. Therefore, each value is multiplied by a combination corresponding to the column item of the distance / sword word correspondence table 122. The result of the multiplication is as follows.

2/3 (degree of close state between robot and target object) × 1/3 (degree of close state between user and target object)
= 2/9 (computation result of combination corresponding to column item 1) 2/3 (degree of close state between robot and target) x 2 /
3 (degree of distant state between user and target) = 4/9 (computation result of combination corresponding to column item 2) 1/3 (degree of distant state between robot and target) × 1 /
3 (degree of close state between user and object) = 1/9 (computation result of combination corresponding to column item 3) 1/3 (degree of far state between robot and object) × 2 /
3 (degree of distant state between user and target object) = 2/9 (computation result of combination corresponding to column item 4)

The largest value among the calculation results is shown in the data area corresponding to the column item (column item 2 in this example) of the distance / sword word correspondence table 122 indicating the combination of the perspective and near states. That is when words are selected.

In the above description, the case where the distance / direction dependent words are appropriately selected based on the perspective state has been described as an example. For example, when the robot and the user are far apart, the robot Phone number, address,
It is possible not to utter personal information such as age. That is, information to be uttered can be appropriately selected based on the perspective state.

In the above description, the distance / direction-dependent words (calling words and sneak words are distance / direction-dependent words) that can be properly used in accordance with the positional relationship between the robot, the user, and the object are appropriately selected. However, for example, words such as “go” or “come” that can be properly used according to the positional relationship between the robot, the destination, and the user can be appropriately selected.

In the above description, the case where the response sentence is output by voice has been described as an example. However, for example, a display unit may be further provided to display the response sentence.

The above series of processing can be realized by hardware, but can also be realized by software. When a series of processing is realized by software, a program constituting the software is installed in a computer, and the program is executed by the computer, whereby the above-described robot is functionally realized.

FIG. 27 is a block diagram showing the configuration of an embodiment of a computer 501 functioning as a robot as described above. An input / output interface 516 is connected to the CPU 511 via a bus 515.
When a user inputs a command from the input unit 518 including a keyboard, a mouse, and the like via the input / output interface 516, the ROM 11 reads, for example, a ROM (Read Only Memory).
y) 512, hard disk 514, or drive 5
Magnetic disk 531 and optical disk 53 mounted on
2. Magneto-optical disk 533 or semiconductor memory 53
4 is stored in a storage medium such as RAM.
(Random Access Memory) 513 is loaded and executed. Thereby, the various processes described above (for example, FIG.
24 or the processing shown by the flowchart in FIG. 24). Further, the CPU 511 outputs the processing result to a display unit 517 such as an LCD (Liquid Crystal Display) via the input / output interface 516 as necessary. The program is stored in the hard disk 514 or the ROM 512 in advance and provided to the user integrally with the computer 501, or provided as package media such as the magnetic disk 531, the optical disk 532, the magneto-optical disk 533, and the semiconductor memory 534. Or can be provided to the hard disk 514 via a communication unit 519 from a satellite, a network, or the like.

In the present specification, the step of describing a program provided by a recording medium may be performed not only in chronological order according to the described order but also in chronological order. This also includes processing executed in parallel or individually.

[0130]

According to the information processing apparatus according to the first aspect, the information processing method according to the fifth aspect, and the program of the recording medium according to the sixth aspect, the robot and the object specified in the utterance In this case, a word corresponding to the positional relationship between the robot and the user or the user and the target is selected according to the positional relationship, so that a more appropriate response sentence can be generated as a response to the utterance of the user. .

[Brief description of the drawings]

FIG. 1 is a diagram showing a usage example of a robot to which the present invention is applied.

FIG. 2 is a diagram illustrating a configuration example of an appearance of the robot in FIG. 1;

FIG. 3 is a block diagram showing an example of the internal configuration of the robot shown in FIG. 1;

FIG. 4 is a block diagram illustrating a functional configuration example of a controller 10 of FIG. 3;

FIG. 5 is a diagram illustrating a perspective determination pattern A.

FIG. 6 is a diagram illustrating a positional relationship between a robot and a user.

FIG. 7 is another diagram showing the positional relationship between the robot and the user.

FIG. 8 is a diagram illustrating a perspective determination pattern B;

FIG. 9 is a perspective judgment pattern A and a perspective judgment pattern B.
It is a figure explaining the example of utilization of.

FIG. 10 is another diagram illustrating a use example of the perspective determination pattern A and the perspective determination pattern B;

FIG. 11 is another diagram illustrating a perspective determination pattern B;

FIG. 12 is another diagram illustrating a perspective determination pattern A;

FIG. 13 is another diagram illustrating an example of use of the perspective determination pattern A and the perspective determination pattern B.

FIG. 14 is another diagram illustrating an example of use of the perspective determination pattern A and the perspective determination pattern B;

FIG. 15 is a diagram showing calculation results of the degree of the near state and the degree of the far state.

FIG. 16 is a diagram illustrating a configuration example of a dialog management unit 56 in FIG. 4;

17 is a diagram showing a distance / calling word correspondence table 121. FIG.

FIG. 18 is a diagram showing a distance / sword word correspondence table 122.

FIG. 19 is a diagram illustrating a positional relationship between a robot, a user, and an object.

FIG. 20 is another diagram showing the positional relationship between the robot, the user, and the target object.

FIG. 21 is another diagram showing a positional relationship between a robot, a user, and an object.

FIG. 22 is another diagram showing a positional relationship between a robot, a user, and a target object.

FIG. 23 is a flowchart showing a processing procedure when the robot operates in response to a user's dialogue.

FIG. 24 is a flowchart illustrating details of a process in step S8 of FIG. 23;

FIG. 25 is another diagram showing a positional relationship between a robot, a user, and a target object.

FIG. 26 is another diagram showing the positional relationship between the robot, the user, and the target object.

FIG. 27 is a block diagram illustrating a configuration example of a computer 501.

[Explanation of symbols]

10 controller, 10A CPU, 10B memory, 15 microphone, 16 CCD camera, 17
Touch sensor, 18 speaker, 19 GPS receiver, 20 gyro compass, 21 geomagnetic sensor, 22 ultrasonic sensor, 51 input processor,
52 voice recognition / analysis unit, 53 image recognition unit, 54
Position detection unit, 55 user identification unit, 56 dialogue management unit, 57 output generation unit, 58 audio output unit, 5
9 Operation control unit

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G10L 15/00 G10L 3/00 545C 15/22 551H 13/04 571T 5/02 J (72) Inventor Kishi Hideki 6-7-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation (72) Inventor Table Masanori 6-7-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation (72) Inventor Kazuhiko Tajima Tokyo 6-7-35 Kita-Shinagawa, Shinagawa-ku, Sony Corporation (72) Inventor Masayoshi Takeda 6-7-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation F-term (reference) 3F059 AA00 BA00 BB06 DA05 DC00 DC08 DD00 DD18 FC00 5B091 AA15 AB15 BA02 BA16 BA19 CA12 CA21 CB01 CB12 CB32 CD15 DA03 5D015 AA03 HH04 KK02 LL00 5D045 AA20 AB11 AB30 9A001 BB04 BB06 DZ15 HH17 HH18 HH19 HHZHHZ 62

Claims (6)

[Claims] 1. An information processing apparatus that generates a response sentence as a response to a user's utterance output by a robot, wherein the robot and a target indicated in the utterance, the robot and the user, or the user Acquiring means for acquiring the positional relationship of the object; if the generated response sentence includes a word which can be properly used in accordance with the positional relationship, the word corresponding to the positional relationship acquired by the acquiring means; An information processing apparatus comprising: a selection unit that selects a response sentence; and a generation unit that generates the response sentence by using the word selected by the selection unit. 2. The positional relationship may be based on a distance between the robot and the target, the robot and the user, or a distance between the user and the target, or a direction or a direction in which the robot, the user, or the target is located. The information processing apparatus according to claim 1, wherein: 3. When the utterance includes the words properly used in accordance with the positional relationship, the utterance further includes an estimating unit that estimates the positional relationship based on the words. Executing a process of detecting the user or the target from within a range specified by the positional relationship estimated by the estimating unit, and acquiring the positional relationship based on a detection result of the executed process. The information processing apparatus according to claim 1. 4. The method according to claim 1, wherein the generating unit adjusts the content of the response sentence based on the positional relationship acquired by the acquiring unit when the response sentence is output as voice. 2. The information processing device according to 1. 5. An information processing method of an information processing apparatus for generating a response sentence as a response to a user's utterance output by a robot, wherein the robot and a target indicated in the utterance, the robot and the user, Or, an acquiring step of acquiring a positional relationship between the user and the target; and, if the generated response sentence includes a word that can be properly used in accordance with the positional relationship, the position acquired in the processing of the acquiring step An information processing method, comprising: a selection step of selecting the word corresponding to a relationship; and a generation step of generating the response sentence using the word selected in the processing of the selection step. 6. A program for information processing for generating a response sentence as a response to a user's utterance, which is output by a robot, wherein the robot and a target indicated in the utterance, the robot and the robot An acquisition step of acquiring a positional relationship between a user or the user and the object; and a case where the generated response sentence includes a word that can be properly used in accordance with the positional relationship. A computer comprising: a computer that performs processing including: a selection step of selecting the word corresponding to the positional relationship; and a generation step of generating the response sentence using the word selected in the processing of the selection step. A recording medium on which a program to be executed is recorded.