JP5982840B2 - Dialogue device, dialogue program, and dialogue method - Google Patents

Description

  The present invention relates to a dialogue apparatus, a dialogue program, and a dialogue method.

  An interactive device has been developed that performs an interactive interaction with a person through conversation or gestures. As an aspect of this interactive apparatus, a therapy robot, an entertainment robot, or the like can be given.

  By the way, the dialogue apparatus may not understand the content of the utterance input by the person. As described above, various techniques have been proposed for continuing the conversation even when the content of the utterance input by the person cannot be understood. For example, there has been proposed a technique for discriminating a topic in a current conversation and outputting a response according to the topic when an utterance input by a person cannot be recognized. In addition, for example, when the user's command or the like cannot be understood, the robot is allowed to perform a predetermined action such as tilting the head or placing a hand on the ear, and the voice recognition reliability is low for the user. Techniques have been proposed to notify recognition failure.

JP 2002-169590 A JP 2002-116792 A JP 2002-189488 A JP 2003260681 A

  However, according to the conventional technology, even if the dialogue with the person is continued, the utterance of the dialogue apparatus may become unnatural and the dialogue with the person may become unnatural.

  It is an object of the disclosed technology to provide an interactive apparatus, an interactive program, and an interactive method capable of suppressing an unnatural conversation with a person.

  In one aspect, the dialog device disclosed in the present application includes a detection unit, a calculation unit, a presentation unit, and a change unit. The detection unit detects at least one of an utterance and an action of the conversation target person. The calculation unit calculates an evaluation related to the dialogue with the dialogue target person based on at least one of the utterance and the action detected by the detection unit. The exposing unit constantly expresses the attitude to the conversation target person by gesture. When the state where the evaluation calculated by the calculation unit is high continues, the change unit changes the attitude expressed by the display unit to a state of high understanding with respect to the dialogue. Moreover, a change part changes the attitude | position expressed by the said display part to the state with a low understanding with respect to the said dialog, when the state with the said low evaluation continues.

  According to one aspect of the dialogue device disclosed in the present application, it is possible to suppress an unnatural dialogue with a person.

FIG. 1 is a perspective view illustrating the configuration of the robot according to the first embodiment. FIG. 2 is a block diagram illustrating an internal configuration of the robot according to the first embodiment. FIG. 3 is a block diagram illustrating a functional configuration of a control unit included in the robot according to the first embodiment. FIG. 4 is a diagram illustrating an example of a configuration of conversation data according to the first embodiment. FIG. 5 is a diagram illustrating an example of a configuration of non-language operation data according to the first embodiment. FIG. 6 is a diagram illustrating an example of a configuration of whispering operation data according to the first embodiment. FIG. 7 is a diagram illustrating an example of a configuration of steady operation frequency data according to the first embodiment. FIG. 8 is a diagram showing an example of the flow of dialogue. FIG. 9 is a flowchart illustrating the procedure of the dialogue processing according to the first embodiment. FIG. 10 is a flowchart illustrating the procedure of the expression attitude determination process according to the first embodiment. FIG. 11 is a flowchart of the response process according to the first embodiment. FIG. 12 is a flowchart illustrating a procedure of steady attitude expression processing according to the first embodiment. FIG. 13 is a diagram for describing an example of a computer that executes a dialogue processing program.

  Embodiments of a dialogue apparatus, a dialogue program, and a dialogue method disclosed in the present application will be described below in detail with reference to the drawings. Note that this embodiment does not limit the disclosed technology. Each embodiment can be appropriately combined within a range in which processing contents are not contradictory. In the following embodiments, a case where the present invention is applied to a robot will be described as an aspect of an interactive device.

[Entire robot configuration]
First, the overall configuration of the robot according to the present embodiment will be described. FIG. 1 is a perspective view illustrating the configuration of the robot according to the first embodiment. As shown in FIG. 1, the robot 1 adopts the appearance of a stuffed doll having a skin shape covered with soft hair and a body shape close to that of an infant. By adopting the appearance of the stuffed toy of such a cub, the robot 1 can promote skinship and produce an intermediate presence between animals and humans (infants). Can have a dialogue.

  The robot 1 has a head 2, a torso 3, a right arm 4R, a left arm 4L, a right leg 5R, and a left leg 5L. When the robot 1 is placed with the head 2 in the vertical direction, the head 2 is rotated in the front-rear direction with the left / right axis (X axis) and the yaw rotated in the left / right direction with the upper / lower axis (Z axis). Operation in the roll direction that rotates in the left-right direction about the direction and the front-rear axis (Y-axis) is possible. That is, the robot 1 can rotate the head 2 in the three axis directions of the pitch direction, the yaw direction, and the roll direction. The robot 1 can perform various operations by rotating the head 2 in three axis directions. For example, the robot 1 performs an operation of turning the head 2 in the pitch direction, an operation of rotating the head 2 in the yaw direction and turning the face sideways, and an operation of tilting the head 2 by rotating the head 2 in the roll direction. It can be carried out.

  Further, the right arm 4R and the left arm 4L are capable of rotating in the front-rear direction around the body 3 and the connection portion when the robot 1 is placed with the head 2 in the upward direction. That is, the robot 1 can rotate the right arm 4R and the left arm 4L in one pitch direction. The robot 1 can move the right arm 4R and the left arm 4L back and forth by moving the right arm 4R and the left arm 4L in the pitch direction.

  Further, the right leg 5R and the left leg 5L can be rotated in the front-rear direction around the body 3 and the connection portion when the robot 1 is disposed with the head 2 in the upward direction. The robot 1 can move the right leg 5R and the left leg 5L back and forth by moving the right leg 5R and the left leg 5L in the pitch direction.

  The head 2 is provided with a right ear 6R, a left ear 6L, two eyes 7, a nose 8, and a mouth 9. The right ear 6R and the left ear 6L are capable of rotating in the front-rear direction around the head 2 and the connection portion. The eye 7 is capable of rotating the eyeball in the left-right direction, and can move the line of sight to the left and right. The eye 7 is provided with an eyelid 7A. The eyelid 7A can move in the vertical direction and can blink. The mouth 9 can be opened and closed. Further, the nose 8 incorporates a camera 41 described later. The camera 41 is arranged with its optical axis substantially coincident with the viewing direction of the robot 1. Furthermore, the head 2 incorporates a microphone 42 and a speaker 43 described later.

  The robot 1 according to the present embodiment performs a dialogue with a person by utterance and gesture. When the robot 1 according to the present embodiment detects at least one of a person's utterance and action, the robot 1 calculates an evaluation related to the conversation with the person to be talked based on at least one of the detected utterance and action. The robot 1 according to the present embodiment performs a response operation by utterance or gesture according to the evaluation regarding the dialogue with the dialogue target person. Further, the robot 1 according to the present embodiment constantly expresses an attitude to the conversation target person by gesture even when there is no utterance from a person. Then, the robot 1 according to the present embodiment changes the attitude to be expressed to a state of high understanding with respect to the conversation when the state of high evaluation continues in the utterance of the conversation target person, and the state of low evaluation is If it continues, the attitude to be expressed is changed to a state of low understanding for dialogue. For example, the robot 1 according to the present embodiment understands the content of the dialogue, and has a sense of listening that indicates that the user is highly interested, an unknown feeling that indicates that the content of the dialogue is not understood, and a content of the dialogue. First, any attitude of boredom indicating boring is expressed by gestures. And the robot 1 which concerns on a present Example will change the attitude to express to a feeling of listening when a state with high evaluation continues, and will change to a feeling of unknown and bored when a state with low evaluation continues. As described above, the robot 1 according to the present embodiment changes the attitude to be expressed in accordance with the understanding state with respect to the utterance of the conversation target person. For this reason, according to the robot 1 according to the present embodiment, it is possible to determine whether the robot 1 is in a state of high understanding or low understanding of the dialogue from the attitude of the robot 1 to express. Thereby, according to the robot 1 according to the present embodiment, even if the robot 1 is in a state of low understanding with respect to the dialogue and the utterance becomes unnatural, the conversation target person has a low understanding of the dialogue of the robot 1 in advance. Since it can be determined that the state is a state, it is possible to suppress an unnatural conversation with a person.

  Next, the internal configuration of the robot 1 according to the present embodiment will be described. FIG. 2 is a block diagram illustrating an internal configuration of the robot according to the first embodiment.

  As shown in FIG. 2, the head 2, the right arm 4R, the left arm 4L, the right leg 5R, and the left leg 5L are connected to the body 3 of the robot 1 through motors 33P, 33R, 33Y, 34R, 34L, 35R, and 35L. Is done. The head 2 of the robot 1 is provided with small motors 36R, 36L, 37, 38, and 39.

  The motors 33P, 33R, and 33Y are provided at the neck joint that connects the head 2 and the body 3 of the robot 1. The motor 33P rotates the head 2 of the robot 1 in the pitch direction, the motor 33R rotates the head 2 of the robot 1 in the roll direction, and the motor 33Y rotates the head 2 of the robot 1 in the yaw direction.

  The motors 34R and 34L are provided at shoulder joints that connect the right arm 4R and the left arm 4L of the robot 1 and the body 3. The motor 34R rotates the right arm 4R of the robot 1 in the front-rear direction of the robot 1. The motor 34L rotates the left arm 4L of the robot 1 in the front-rear direction of the robot 1.

  The motors 35R and 35L are provided at the hip joint that connects the right leg 5R and the left leg 5L of the robot 1 and the body 3. The motor 35R rotates the right leg 5R of the robot 1 in the front-rear direction of the robot 1. The motor 35L rotates the left leg 5L of the robot 1 in the front-rear direction of the robot 1.

  Further, the motors 36R and 36L are provided at the connection points of the right ear 6R and the left ear 6L of the head 2 of the robot 1. The motor 36R rotates the right ear 6R in the front-rear direction. The motor 36L rotates the left ear 6L in the front-rear direction.

  The motor 37 is provided in the portion of the eyes 7 of the head 2 of the robot 1. The motor 37 rotates the eyeballs of the two eyes 7 in the left-right direction using a transmission mechanism (not shown). That is, the motor 37 changes the line-of-sight direction of the robot 1 by rotating the eyeball of the eye 7 in the left-right direction.

  The motor 38 is provided in the eye 7 portion of the head 2 of the robot 1. The motor 38 moves the eyelids 7A of the two eyes 7 in the vertical direction using a transmission mechanism (not shown). That is, the motor 38 blinks by operating the eyelid 7A in the vertical direction.

  The motor 39 is provided in the mouth 9 portion of the head 2 of the robot 1. The motor 39 opens and closes the opening 9 using a transmission mechanism (not shown).

  Here, hereinafter, the motors 33P, 33R, 33Y, 34R, 34L, 35R, 35L, and the motors 36R, 36L, 37, 38, and 39 may be collectively referred to as “motor 30” in some cases. The motor 30 is connected to a motor control unit 40 that controls the motor. The motor control unit 40 performs various overall controls such as power control of the motor 30 that drives the head 2, the right arm 4R, the left arm 4L, the right leg 5R, and the left leg 5L, and monitoring of a current value flowing through the motor 30. In addition, various overall controls such as power control of the motor 30 that drives the right ear 6R, the left ear 6L, the eyes 7, and the mouth 9 and monitoring of a current value flowing through the motor 30 are performed.

  Each motor 30 transmits power according to an instruction from the motor control unit 40, so that the head 2, the right arm 4 </ b> R, the left arm 4 </ b> L, the right leg 5 </ b> R, the left leg 5 </ b> L, the right ear 6 </ b> R, the left ear 6 </ b> L, The eyelid 7A and the mouth 9 are respectively driven.

  The head 2 of the robot 1 is provided with a camera 41, a microphone 42, and a speaker 43. The camera 41 is disposed in the nose 8 portion of the head 2 and photographs the front direction of the head 2 at a predetermined cycle. The microphone 42 can be attached at an arbitrary position, but is preferably installed on the front side of the head 2 in order to collect the utterances of the person performing the dialogue. The speaker 43 can also be attached at an arbitrary position, but is preferably installed on the front side of the head 2 in order to output a sound to a person who performs a dialogue.

  The body 3 of the robot 1 is provided with a control unit 10 that controls the entire robot 1. The motor control unit 40, the camera 41, the microphone 42 and the speaker 43 are connected to the control unit 10. The motor control unit 40 performs control to operate each motor 30 in accordance with an instruction from the control unit 10. The camera 41 outputs the captured image data to the control unit 10. The microphone 42 outputs the collected audio data to the control unit 10. The speaker 43 outputs sound based on control from the control unit 10.

  Next, the functional configuration of the control unit included in the robot according to the present embodiment will be described. FIG. 3 is a block diagram illustrating a functional configuration of a control unit included in the robot according to the first embodiment. As shown in FIG. 3, the control unit 10 is a processing unit that performs overall control of the robot 1. A storage unit 11 is connected to the control unit 10.

  The storage unit 11 stores an OS (Operating System) executed by the control unit 10 and various programs used for controlling the robot 1. Further, the storage unit 11 stores various data necessary for executing the program executed by the control unit 10. As an example of such data, the storage unit 11 stores conversation data 11a, non-language motion data 11b, whispering motion data 11c, and steady motion frequency data 11d.

  The conversation data 11a is information indicating the utterance content when a conversation is performed. As an example, the conversation data 11 a is registered in advance at the creation source of the robot 1. As another example, the conversation data 11a is referred to by a response determination unit 10f described later in order to speak.

  FIG. 4 is a diagram illustrating an example of a configuration of conversation data according to the first embodiment. As shown in FIG. 4, the conversation data 11 a includes items of “topic”, “question”, “attitude”, and “next question”. The topic item is an area for storing information indicating a topic for conversation. The question item is an area for storing information indicating the content of a question to a person. The attitude item is an area for storing information indicating the attitude to be expressed. The next question item is an area for storing information indicating the question content to be questioned next. The conversation data 11a stores the question content for the next question for each topic, question content, and attitude. Also, in the record in which “-” is stored in both the question and attitude items, the content of the question to be asked first when the topic is changed is stored.

  In the example of FIG. 4, it is shown that the question “Question a1” is first performed on the topic “A”. Further, in the example of FIG. 4, when the question “Question a1” of the topic “A” is asked and the answer is obtained, if the attitude to be expressed is “listening”, then the question “Question a2” is asked next. Indicates what to do. In the example of FIG. 4, when the question “Question a1” of the topic “A” is asked and the answer is obtained, when the attitude to be expressed is “Unknown”, the question “Question a3” is then asked. Indicates what to do. Further, in the example of FIG. 4, when the question “Question a1” of the topic “A” is asked and the answer is obtained, when the attitude to be expressed is “feeling bored”, the question of “Topic B” is asked next. Indicates what to do. In the example of FIG. 4, the question “question b1” is first performed on the topic “B”.

  The non-language action data 11b is information indicating action contents for each attitude to be expressed. As an example, the non-language motion data 11b is registered in advance at the creation source of the robot 1. As another example, the non-language motion data 11b is referred to by a non-language response determination unit 10h and a steady non-language operation determination unit 10i described later in order to instruct a non-language expression operation.

  FIG. 5 is a diagram illustrating an example of a configuration of non-language operation data according to the first embodiment. As illustrated in FIG. 5, the non-language motion data 11 b includes items of “setting element”, “listening feeling”, “unknown feeling”, and “boring feeling”. The item of setting element is an area for storing information indicating an element to be set for expressing an attitude. The item of listening feeling is an area for storing information indicating setting contents in each element when expressing listening feeling. The item of unclearness is an area for storing information indicating setting contents in each element when unclearness is expressed. The bored feeling item is an area for storing information indicating the setting contents of each element when expressing bored feeling.

  The setting element item stores an element that shows a difference in operation when a person expresses a sense of listening, an unknown feeling, or a bored feeling. In each of the items of listening feeling, unknown feeling, and feeling bored, information indicating the action contents that become prominent is stored. As an aspect, in this embodiment, settings are made for “volume”, “pitch”, “speech speed”, “line of sight”, “blink”, and “motion, facial expression” as setting elements. The volume setting element indicates the volume setting when outputting sound such as speech from the speaker 43. In the present embodiment, “low” in the volume indicates that the volume is lower than the standard volume of the utterance. “Middle” of the volume indicates that the standard volume of the utterance is set. “Large” of the volume indicates that the volume is higher than the standard volume of the utterance. The standard volume of utterance may be determined, for example, by analyzing the sounds of a large number of people in advance, or a predetermined volume may be determined as the standard volume.

  The pitch setting element indicates the setting of the voice pitch when the utterance is output from the speaker 43. In this embodiment, “low” of the voice pitch indicates that the voice pitch of the utterance is set lower than the standard voice pitch of the utterance. “Middle” of the voice pitch indicates that the voice level of the utterance is set as the standard voice level of the utterance. “High” of the voice pitch indicates that the voice pitch of the utterance is set higher than the standard voice pitch of the utterance. The standard voice level of the utterance may be determined, for example, by analyzing a number of human sounds in advance, or the predetermined voice level may be determined as the standard pitch.

  The utterance speed setting element indicates the setting of the utterance speed when the utterance is output from the speaker 43. In the present embodiment, “slow” in the speaking speed indicates that the speaking speed is made slower than the standard speaking speed. “Normal” of utterance speed indicates that the utterance speed is set to the standard speed of utterance. The standard utterance speed may be determined, for example, by analyzing the sounds of a large number of people in advance, and the predetermined utterance speed may be determined as the standard utterance speed.

  The line-of-sight setting element indicates the setting of the line-of-sight direction of the eye 7. In the present embodiment, “user gaze” of the line of sight indicates that the line of sight remains in the direction of the user. “Slightly divert” the line of sight indicates that the line of sight is directed toward the user, but the line of sight is periodically moved in a direction different from the user. The line of sight “not seeing the user so much” indicates that the line of sight is directed in a direction different from the user and the line of sight is periodically moved in the direction of the user. The period for moving the line of sight may be, for example, a fixed period or a random period within a predetermined range.

  The blink setting element indicates a blink setting for moving the eyelid 7A up and down. In this embodiment, blinking “normal” indicates blinking at a standard frequency. “Frequently” blinking means that blinking is performed at a frequency higher than the standard frequency. The standard blink frequency may be determined, for example, by analyzing the blink frequency of a large number of people in advance, or the predetermined frequency may be determined as the standard blink frequency.

  The action / expression setting elements indicate settings related to the action and expression expressed by the robot 1. In the present embodiment, “smile” in action and expression indicates that each part of the robot 1 is operated to make it feel like a smile. For example, the right ear 6R and the left ear 6L are in the most upright state, the head 2 and eyes 7 are directed toward the user, the eyelid 7A and the mouth 9 are fully opened, and the right arm 4R and the left arm 4L are moved upward. The action and expression “tilt the neck” indicate that each part of the robot 1 is operated to make it feel that the head is tilted. For example, the mouth 9 is closed, the right arm 4R and the left arm 4L are moved downward, and the head 2 is rotated in the yaw direction or the pitch direction to tilt the neck. “Let's sleep” in action and expression indicates that each part of the robot 1 is operated to make it feel sleepy. For example, the eyelid 7A is half open, the right arm 4R and the left arm 4L are moved downward, the head 2 is rotated at a pitch, and the head 2 is directed downward.

  In the example of FIG. 5, when expressing a sense of listening, the volume is set to “high”, the pitch is set to “medium”, the speaking speed is set to “normal”, the line of sight is set to “user gaze”, and the blink is set to “ “Normal” and “Smile” as the action and facial expression. In the example of FIG. 5, when an unclear feeling is expressed, the volume is set to “low”, the pitch is set to “high”, the speed of the utterance is set to “normal”, and the line of sight is changed from time to time. It indicates that blinking is “frequently” and movement and facial expression are “tilt neck”. Further, in the example of FIG. 5, when boring is expressed, the volume is set to “medium”, the pitch is set to “low”, the speaking rate is set to “slow”, and the line of sight is set to “not seeing the user very much”. ”,“ Normal ”for blinking, and“ sleepy ”for action and facial expression.

  The whispering motion data 11c is information indicating settings related to whispering when performing a response. As an example, the whispering motion data 11 c is registered in advance at the creator of the robot 1. As another example, the whispering motion data 11c is referred to by a non-language response determination unit 10h described later in order to instruct a whispering motion.

  FIG. 6 is a diagram illustrating an example of a configuration of whispering operation data according to the first embodiment. As illustrated in FIG. 6, the whispering motion data 11 c includes items of “setting element”, “feeling of listening”, “feeling of unknown”, and “feeling of boredom”. The item of setting element is an area for storing information indicating an element to be set for the whirling operation. The item of listening feeling is an area for storing information indicating setting contents in each element when expressing listening feeling. The item of unclearness is an area for storing information indicating setting contents in each element when unclearness is expressed. The bored feeling item is an area for storing information indicating the setting contents of each element when expressing bored feeling.

  In the setting element field, a setting element related to the whirling operation is stored. In each item of the sense of listening, the feeling of ambiguity, and the feeling of boredom, information indicating the content of the whispering operation is stored. As an aspect, in the present embodiment, settings are made for “blowing depth”, “blowing speed”, and “blowing delay” as setting elements. The setting element for the pitching depth indicates the setting of the magnitude of the amplitude of the up and down motion when rolling. The setting element for the sowing speed indicates the setting of the soaking speed. The whispering delay setting element indicates the setting of a period from when the user's utterance is detected until he / she speaks.

  Here, a person's movement changes according to the state with respect to the dialogue. For example, when a person listens to the other person's story, he or she responds according to the other person's utterance. In addition, when a person cannot fully understand the other person's speech, the person speaks according to the other person's speech, but the amplitude of the whispering action is small, the speed of whispering is increased, and the frequency of whispering is increased. In addition, when a person cannot understand the other person's story and is bored, the person speaks according to the other person's utterance, but the delay is large, and the amplitude of the whispering action becomes smaller. Therefore, in this embodiment, conditions for deriving the depth, speed, and delay from the understanding level and interest level described later are set for each of the sense of listening, unknownness, and boredom.

  In the example of FIG. 6, when the attitude to be expressed is “listening”, the depth of craving is derived from the coefficient g × max (understanding degree, interest degree), and the speed of craving is derived from the coefficient i × interest degree. This indicates that the whirling delay is 0.0. Note that max (degree of understanding, degree of interest) is a function that returns the largest value from the degree of understanding and degree of interest assigned in (). Further, in the example of FIG. 6, when the attitude to be expressed is “Unknown”, the depth to be whispered is derived from the coefficient h × understanding degree, the whirling speed is set to 1.0, and the coefficient j × (1.0 -Demonstrate that the lag is derived from the degree of understanding. In the example of FIG. 6, when the attitude to be expressed is “feeling bored”, the crawl depth is 0.1, the crawl speed is 0.5, and the coefficient k × (1.0−interest degree) It shows that the whirling delay is derived. In this embodiment, the depth, speed, and delay are determined in the range of 0.0 to 1.0, and when smaller than 0.0, 0.0 is greater than 1.0. In this case, 1.0 is set. In addition, the coefficient g is set to a value larger than the coefficient h in order to increase the depth of the listening feeling rather than the unknown feeling. The coefficient k is set to a value larger than the coefficient j in order to increase the delay in boring with a feeling of boredom than with an unclear feeling.

  The steady operation frequency data 11d is information indicating a setting relating to a frequency at which an attitude is constantly expressed in a state where there is no utterance from a person. As an example, the steady operation frequency data 11 d is registered in advance at the creation source of the robot 1. As another example, the steady motion frequency data 11d is referred to by a steady non-linguistic motion determination unit 10i, which will be described later, in order to instruct the motion timing of a motion of a steady non-language expression.

  FIG. 7 is a diagram illustrating an example of a configuration of steady operation frequency data according to the first embodiment. As shown in FIG. 7, the whispering motion data 11c has items of “setting element”, “feeling of listening”, “feeling of unknown”, and “feeling of boredom”. The item of the setting element is an area for storing information indicating an operation for constantly expressing the attitude. The item of listening feeling is an area for storing information indicating setting contents in each element when expressing listening feeling. The item of unclearness is an area for storing information indicating setting contents in each element when unclearness is expressed. The bored feeling item is an area for storing information indicating the setting contents of each element when expressing bored feeling.

  In the setting element item, there is stored an operation in which a difference appears when a person expresses a sense of listening, unknown, or bored. Conditions for deriving the frequency from the degree of understanding and the degree of interest, which will be described later, are set for each item of the sense of listening, the feeling of ambiguity, and the feeling of boredom. As an aspect, in this embodiment, settings are made for “blink frequency” and “gaze frequency” as setting elements. The setting element of the whispering frequency indicates setting of a cycle for whispering. The line-of-sight frequency setting element indicates the setting of a cycle for aligning the line of sight with the user.

  In the example of FIG. 7, when the attitude to be expressed is “listening”, the frequency of whispering is derived from the degree of understanding × the coefficient a, and the frequency of matching the line of sight with the user is derived from the degree of understanding × the coefficient d. Show. In the example of FIG. 7, when the attitude to be expressed is “Unknown”, the frequency of whispering is derived from the degree of understanding × the coefficient b, and the frequency of matching the user's line of sight is derived from the degree of interest × the coefficient e. It shows that. Further, in the example of FIG. 7, when the expressed attitude is “feeling bored”, the frequency of whispering is derived from the degree of interest × coefficient c, and the frequency of matching the line of sight with the user is derived from the degree of interest × coefficient f. It shows that. Here, as will be described later, in this embodiment, the degree of interest increases when the degree of understanding is high, and the degree of interest is calculated from the degree of understanding so that the degree of interest decreases when the degree of understanding decreases. It is determined from the degree of interest whether the state of dialogue is a sense of listening, an unknown feeling, or a bored feeling. When the conversation state is a sense of listening with a high degree of interest, the whispering frequency and the gaze frequency are derived from the degree of understanding. Also, when the conversation state has a high degree of interest and a low level of understanding, the whispering frequency is derived from the understanding level, and the line-of-sight frequency is derived from the interest level. When the conversation state is boring with a low degree of interest, the whispering frequency and the line-of-sight frequency are derived from the degree of interest. Note that the coefficient a, the coefficient b, and the coefficient c are set so that coefficient a> coefficient b> coefficient c in order to increase the whispering frequency with an unclear feeling more than a bored feeling and with a sense of listening rather than an unclear feeling. Set. Further, in order to increase the line-of-sight frequency with an unknown feeling more than bored feeling and with a sense of listening more than an unknown feeling, the coefficient d, the coefficient e, and the coefficient f are values such that the coefficient d> the coefficient e> the coefficient f. Set.

  Returning to FIG. 3, the control unit 10 has an internal memory for storing programs and control data that define various processing procedures, and executes various processes using these. As shown in FIG. 3, the control unit 10 includes an utterance analysis unit 10a, an utterance understanding unit 10b, a non-linguistic expression analysis unit 10c, an evaluation calculation unit 10d, and a state determination unit 10e. Further, the control unit 10 includes a response determination unit 10f, an utterance response determination unit 10g, a non-language response determination unit 10h, a steady non-language operation determination unit 10i, an utterance generation unit 10j, and a non-language operation instruction unit 10k. It has further.

  Note that various types of integrated circuits and electronic circuits can be employed for the control unit 10. For example, an ASIC (Application Specific Integrated Circuit) is an example of the integrated circuit. Examples of the electronic circuit include a central processing unit (CPU) and a micro processing unit (MPU).

  The utterance analysis unit 10 a performs voice recognition on the voice data acquired by the microphone 42. The speech recognition method may be any method such as a statistical method, a dynamic time expansion / contraction method, or a method using a hidden Markov model. The speech analysis unit 10a performs voice recognition on the voice data, and converts the recognized voice into text data.

  The utterance understanding unit 10b calculates the ratio of words that can be understood with respect to the utterance content. As one aspect, the utterance understanding unit 10b calculates the ratio of words that can be recognized by the utterance analysis unit 10a from the voice data. Moreover, as another aspect, for example, for each topic to be interacted with, words included in the conversation when the conversation of each topic is performed are previously stored as topic-specific word data. Then, the utterance understanding unit 10b calculates a ratio in which the text data converted by the utterance analysis unit 10a includes a word related to the current topic stored in the topic-specific word data.

  The non-linguistic expression analysis unit 10c analyzes the interest state for the dialogue from the non-linguistic expression, and generates a measurement value indicating the interest state. For example, when a person is talking happily, the person is highly interested in the topic. Thus, as one aspect, the non-linguistic expression analysis unit 10c performs image processing including face recognition processing on the image data obtained from the camera 41 to detect the mouth corners and the corners of the eyes. Then, the non-linguistic expression analysis unit 10c calculates, as a measurement value, a smile degree that indicates a high possibility that the facial expression of a person is a smile based on the detected change in the position of the mouth corner and the corner of the eye and the degree of mouth spread. In addition, when a person is eagerly speaking, the gesture becomes larger or the gesture becomes faster. Therefore, the non-linguistic expression analysis unit 10c tracks the movement of the person's hand included in the image data, thereby obtaining measurement values indicating the size of the gesture indicated by the movement of the person's hand and the speed of the movement. It may be calculated. When a person is eagerly speaking, the voice is louder, the intonation is greater, and the frequency of utterance increases. Therefore, as another aspect, the non-linguistic expression analysis unit 10c applies, for example, voice analysis processing to voice data obtained by the microphone 42, so that, for example, human voice inflection or voice volume, The frequency of utterances or the like may be generated as a measured value indicating the strength of non-visual non-verbal reaction. Note that the robot 1 may be provided with a contact sensor for detecting contact in addition to the camera 41 and the microphone 42 as a sensor for detecting the non-language expression of the user. And the non-linguistic expression analysis part 10c may produce | generate the measured value which shows an interest state from the contact period and contact frequency of a user detected with a contact sensor. For example, the non-linguistic expression analysis unit 10c calculates the measurement value higher as the contact period of the user is longer and the contact frequency is higher.

  The evaluation calculation unit 10d calculates an evaluation related to the dialogue with the conversation target person based on the ratio of the understood words calculated by the utterance understanding unit 10b and the measurement value obtained by the analysis by the non-linguistic expression analysis unit 10c. . As an aspect, the evaluation calculation unit 10d calculates an understanding level for an utterance as an evaluation related to a dialogue. Then, the evaluation calculation unit 10d calculates an interest level representing the degree of interest in the dialogue from the calculated understanding level. For example, when the percentage of words that can be understood about the speech input at time t by the utterance understanding unit 10b is p, the evaluation calculation unit 10d calculates from the degree of understanding U (t-1) at time t-1. The degree of understanding U (t) at time t is calculated from the following equation (1).

U (t) = U (t−1) × (1−α) + p × α (1)
Here, α is a coefficient for updating.

  For example, when the measured value for the non-linguistic expression i detected at time t is set to qi by the speech understanding unit 10b, the evaluation calculation unit 10d determines from the understanding level U (t-1) at the time t−1. The degree of understanding U (t) at time t is calculated by the following equation (2).

U (t) = U (t−1) × (1−β) + qi × β (2)
Here, β is a coefficient for updating.

  Then, the evaluation calculation unit 10d calculates the degree of interest so that it rises when a state where the calculated degree of understanding continues is high and decreases when a state where the degree of understanding is low continues. For example, when the understanding level U (t) calculated at the time t is used, the evaluation calculation unit 10d calculates the interest level I (t-1) at the time t-1 from the following equation (3) to the time t The degree of interest I (t) is calculated.

I (t) = I (t−1) − (1.0−U (t)) × γ + A (3)
Here, γ is a coefficient for updating, and γ> A. A is a constant (0 <A <1.0).

  In this embodiment, the degree of understanding and the degree of interest are set in the range of 0.0 to 1.0. The evaluation calculation unit 10d sets 0.0 when the calculated U (t) and I (t) are smaller than 0.0, and sets 1.0 when larger than 1.0. In the initial state where the conversation is started, 1.0 is set as the initial value for the degree of understanding and the degree of interest.

  The state determination unit 10e determines the attitude to be expressed based on the evaluation calculated by the evaluation calculation unit 10d. Moreover, when the state where evaluation is high continues, the state determination part 10e changes the attitude to express to a state with high understanding with respect to a dialog. Moreover, when the state where evaluation is low continues, the state determination unit 10e changes the attitude to be expressed to a state of low understanding with respect to the dialogue. As one aspect, the state determination unit 10e determines whether or not the degree of interest is greater than a predetermined threshold T1. Then, when the degree of interest is equal to or less than the threshold value T1, the state determination unit 10e determines that the attitude to be expressed is bored. If the degree of interest is greater than the threshold value T1, the state determination unit 10e determines whether or not the degree of understanding is greater than a predetermined threshold value T2. Then, when the degree of understanding is greater than the threshold value T2, the state determination unit 10e determines the attitude to be expressed as a sense of listening. On the other hand, when the understanding level is equal to or less than the threshold value T2, the state determination unit 10e determines the attitude to be expressed as an unknown feeling. Note that the threshold value T1 and the threshold value T2 are appropriately determined so as to correspond to the calculated degree of understanding and the degree of interest. For example, the threshold value T1 is a value within the range of 0.3 to 0.5. The threshold T2 is a value in the range of 0.3 to 0.7. The threshold values T1 and T2 may be adjustable from the outside. Thereby, the state determination part 10e changes a state to express to a bored feeling, an unknown feeling, and a listening feeling one by one when a state with a high degree of understanding continues and the state of interest changes. In addition, the state determination unit 10e sequentially changes the attitude to be expressed to an audible feeling, an unknown feeling, and a bored feeling when the degree of understanding continues and the degree of interest changes to a low state.

  The response determination unit 10f determines the utterance content when performing a dialogue with a person. As one aspect, the response determination unit 10f selects any topic stored in the conversation data 11a for the first utterance content that has started the conversation, and reads and reads out the question content to be asked at the beginning of the selected topic. The question content is determined as the first utterance content. In addition, the response determination unit 10f asks the user on any topic, and when the utterance understanding unit 10b analyzes the user's utterance, the response determination unit 10f performs the following processing. That is, the response determination unit 10f reads the next question content corresponding to the attitude determined by the state determination unit 10e for the question immediately before the current topic from the conversation data 11a, and the read question content is the next utterance. Decide with content.

  The utterance response determination unit 10g determines the sound quality of the utterance. As one aspect, the utterance response determination unit 10g reads the volume, voice pitch, and utterance speed settings corresponding to the attitude from the non-language action data 11b, and determines the read settings as sound quality settings related to the utterance.

  The non-language response determination unit 10h determines the operation content to be expressed. As one aspect, the non-linguistic response determination unit 10h reads out the conditions for deriving the whirling depth, whirling speed, and whispering delay corresponding to the attitude from the whispering motion data 11c. Then, the non-linguistic response determination unit 10h uses the understanding level and the interest level calculated by the evaluation calculation unit 10d to calculate the whirling depth, the whirling speed, and the whirling delay from the read out derivation conditions. For example, if the determined attitude is a sense of listening, the non-linguistic response determination unit 10h derives the depth obtained from the coefficient g × max (understanding degree, interest degree), and determines the speed obtained from the coefficient i × interest degree. Derived and set the throwing delay to 0.0. In addition, when the determined attitude is unclear, the non-language response determination unit 10h derives the depth to be uttered from the coefficient h × understanding, sets the crawl speed to 1.0, and the coefficient j × (1.0 -Deriving the whispering delay from the degree of understanding. If the determined attitude is boring, the non-verbal response determination unit 10h sets a depth of 0.1 and sets a speed of 0.5 as a lag from a coefficient k × (1.0−degree of interest). Is derived. Note that the non-language response determining unit 10h sets 0.0 when the calculated depth, rate, and delay are smaller than 0.0, and 1.0 when larger than 1.0.

  Further, the non-language response determination unit 10h reads the setting of the line of sight, blink, and action expression corresponding to the determined attitude from the non-language action data 11b, and determines the read setting as the setting related to the action.

  The stationary non-verbal motion determining unit 10i determines an attitude that is constantly expressed by whispering and gestures with respect to a conversation target person even in a state where there is no utterance from a person. As one aspect, the steady non-linguistic motion determination unit 10i reads the squeezing frequency and line-of-sight frequency derivation conditions corresponding to the determined attitude from the steady motion frequency data 11d. Then, the non-linguistic response determination unit 10h calculates the stroking frequency and the line-of-sight frequency from the read out derivation conditions using the degree of understanding and the degree of interest calculated by the evaluation calculation unit 10d. For example, if the determined attitude is a sense of listening, the non-language response determination unit 10h derives the frequency of whispering from the degree of understanding × coefficient a, and derives the frequency of matching the user's line of sight from the degree of understanding × coefficient d. . Further, when the determined attitude is unclear, the non-language response determination unit 10h derives the frequency of whispering from the degree of understanding × the coefficient b, and derives the frequency of matching the line of sight with the user from the degree of interest × the coefficient e. . In addition, when the determined attitude is boring, the non-language response determination unit 10h derives the frequency of whispering from the interest degree × coefficient c, and derives the frequency of matching the user's line of sight from the interest degree × coefficient f. . Then, the steady non-linguistic action determination unit 10i derives the period for performing the whispering as being shorter as the whispering frequency is higher. In addition, the steady non-linguistic action determining unit 10i derives a cycle for aligning the line of sight with the user, assuming that the higher the frequency of aligning the line of sight, the shorter. The period may be derived from the frequency by calculation. For example, the period is derived from the value x / frequency. This value x may be a constant or may be changed depending on the dialog state. Moreover, the periodic information which defined the period for each frequency may be stored in the storage unit 11, and the steady non-linguistic action determining unit 10 i may derive the period according to the frequency from the periodic information.

  Further, the steady non-linguistic motion determining unit 10i reads the squeezing depth, the squeezing speed, and the squeezing delay derivation conditions corresponding to the determined attitude from the stumbling motion data 11c, and reads them using the understanding level and the interest level. Based on the derived conditions, the depth, speed, and delay are calculated. In addition, the steady non-language motion determination unit 10i reads gaze, blink, and motion facial expression settings corresponding to the determined attitude from the non-language motion data 11b, and determines the read settings as settings related to the motion.

  The utterance generation unit 10 j generates an audio signal corresponding to the utterance content and outputs it to the speaker 43. As one aspect, when the utterance content and the volume, voice pitch, and utterance speed related to the utterance are determined, the utterance generation unit 10j generates the voice of the determined utterance content at the determined volume, voice pitch, and utterance speed. An audio signal is generated and output to the speaker 43. Thereby, the speaker 43 outputs a sound corresponding to the sound signal and speaks.

  The non-language operation instruction unit 10k generates an operation instruction for performing a non-language operation such as gesture and outputs the operation instruction to the motor control unit 40. As one aspect, the non-language operation instruction unit 10k determines the volume, voice pitch, and speed of speech related to the utterance content and utterance, and when speaking, An operation instruction to perform a whirling operation at a speed and a whispering delay is output to the motor control unit 40. In addition, the non-language operation instruction unit 10k outputs to the motor control unit 40 an operation instruction for performing an operation with the line of sight, blinking, and operation expression determined by the non-language response determination unit 10h. For example, the non-language operation instruction unit 10k moves the eyes 7 or the head 2 in the direction in which the face is recognized by the non-language expression analysis unit 10c, and controls the operation instruction to operate with the determined line of sight, blink, and operation expression. To the unit 40. The motor control unit 40 controls the motor 30 according to the operation instruction, and causes the robot 1 to perform an operation according to the operation instruction. Thereby, the robot 1 responds to the utterance from the user by a non-language operation.

  In addition, the non-language movement instruction unit 10k performs the following when the period of performing the whisper calculated by the steady non-language movement determination unit 10i and the period of matching the line of sight from the previous whispering movement, the movement of aligning the line of sight with the user, An operation instruction for performing a steady operation is output to the motor control unit 40. For example, the non-language operation instruction unit 10k outputs to the motor control unit 40 an operation instruction for causing the operation to be performed with a stroke depth, a stroke speed, and a roll delay determined by the steady non-language operation determination unit 10i. In addition, the non-language motion instruction unit 10k outputs to the motor control unit 40 an operation instruction for performing an operation with the line of sight, blinking, and motion expression determined by the steady non-language operation determination unit 10i. The motor control unit 40 controls the motor 30 according to the operation instruction, and causes the robot 1 to perform an operation according to the operation instruction. As a result, the robot 1 constantly expresses the current state as an attitude by non-verbal operation to the user.

  Next, the flow of dialogue by the robot 1 will be described. FIG. 8 is a diagram showing an example of the flow of dialogue. As illustrated in FIG. 8, when the utterance C1 is made by the user, the robot 1 performs a response C2 based on the utterance and the non-language operation on the utterance C1. For example, when the determined state is a sense of listening, the robot 1 makes a response C2 while whispering, and expresses the sense of listening by gazing at the user.

  Further, the robot 1 constantly expresses the attitude C3 by non-verbal operation. Further, when the determined state is bored as a result of detecting the user's speech or the user's attitude, the robot 1 sometimes turns the line of sight away from the user and expresses boredom in a non-verbal operation. When the user utters C4 and cannot fully understand the contents of the utterance C4, the robot 1 responds to the utterance C4 by expressing a sense of ambiguity by tilting the head along with the utterance.

  In addition, when the robot 1 detects the user's utterance or the user's attitude and the determined state is boring, the robot 1 expresses an attitude C6 such as a sleepy expression with a steady non-verbal motion and feels bored. Is expressed.

  As described above, the robot 1 according to the present embodiment changes the attitude that is constantly expressed in the non-language operation according to the understanding state with respect to the utterance of the conversation target person. For this reason, according to the robot 1 according to the present embodiment, it is possible to determine whether the robot 1 is in a state of high understanding or low understanding of the dialogue from the attitude of the robot 1 to express.

  Next, the flow of processing for performing a dialogue with the robot 1 according to the present embodiment will be described. FIG. 9 is a flowchart illustrating the procedure of the dialogue processing according to the first embodiment. This dialogue processing is started when a predetermined operation for instructing the robot 1 to start dialogue is performed. For example, the interactive process is started when the power of the robot 1 is turned on and a predetermined initial process is completed.

  As illustrated in FIG. 9, the evaluation calculation unit 10d sets 1.0 as an initial value for the degree of understanding and the degree of interest (step S10). The evaluation calculation unit 10d determines whether or not a dialogue input from the user has been detected (step S11). For example, if the user's utterance is detected as a result of speech recognition by the utterance analysis unit 10a, the evaluation calculation unit 10d determines that a dialogue is input by voice. Further, the evaluation calculation unit 10d determines that the dialogue is input with the non-linguistic expression when the non-linguistic expression analysis unit 10c detects the facial expression or action of the user. When the input of the dialogue from the user is not detected (No at Step S11), the process proceeds to Step S11 and waits for the input of the dialogue from the user. When the input of the dialogue from the user is detected (Yes at Step S11), the evaluation calculation unit 10d analyzes the ratio of words that can be understood included in the speech utterance content of the user that has been voice-recognized and the non-linguistic expression of the user. Based on the obtained measurement value, the degree of understanding is updated (step S12). Then, the evaluation calculation unit 10d updates the degree of interest based on the degree of understanding (step S13). The state determination unit 10e determines whether to continue the dialogue (step S14). For example, when the degree of interest is 0, the state determination unit 10e determines that the conversation cannot be continued. If the dialogue cannot be continued (No at Step S14), the process is terminated. When the dialogue can be continued (Yes at Step S14), the state determination unit 10e performs an expression attitude determination process for determining an attitude to be expressed (Step S15). The response determination unit 10f determines the utterance content to be responded from the conversation data 11a (step S16). And the response process which responds by utterance is performed (step S17), and it transfers to step S11 again.

  Next, the flow of the expression attitude determination process according to the present embodiment will be described. FIG. 10 is a flowchart illustrating the procedure of the expression attitude determination process according to the first embodiment. This expression attitude determination process is invoked and started from step S15 of the dialogue process.

  As illustrated in FIG. 10, the state determination unit 10e determines whether or not the degree of interest is greater than a predetermined threshold T1 (step S30). When the degree of interest is equal to or less than the threshold T1 (No at Step S30), the state determination unit 10e determines that the attitude to be expressed is bored (Step S31). On the other hand, when the degree of interest is greater than the threshold T1 (Yes at Step S30), the state determination unit 10e determines whether the degree of understanding is greater than a predetermined threshold T2 (Step S32). When the understanding level is equal to or lower than the threshold T2 (No at Step S32), the state determination unit 10e determines that the attitude to be expressed is unknown (Step S33). On the other hand, when the comprehension level is greater than the threshold value T2 (Yes at Step S32), the state determination unit 10e determines the attitude to be expressed as a sense of listening (Step S34).

  Next, the flow of response processing according to the present embodiment will be described. FIG. 11 is a flowchart of the response process according to the first embodiment. This response process is invoked and started from step S17 of the interactive process.

  As shown in FIG. 11, the speech response determination unit 10g and the non-language response determination unit 10h determine an operation parameter when performing a response (step S40). For example, the utterance response determination unit 10g determines the volume, voice pitch, and utterance corresponding to the attitude from the non-language action data 11b. Further, the non-language response determining unit 10h determines the whirling depth, the whirling speed, and the whirling delay from the whispering motion data 11c. The utterance generation unit 10j generates an audio signal for generating the audio of the determined utterance content at the determined volume, voice pitch, and utterance speed, and outputs the audio signal to the speaker 43 (step S41). The non-language operation instruction unit 10k outputs to the motor control unit 40 an operation instruction for performing a whirling operation with the determined whispering depth, whispering speed, and whispering delay (step S42).

  Next, a flow of steady attitude expression processing in which the robot 1 according to the present embodiment constantly expresses attitude will be described. FIG. 12 is a flowchart illustrating a procedure of steady attitude expression processing according to the first embodiment. This steady-state attitude expression process is started when a predetermined operation for instructing the robot 1 to start a dialogue is performed. For example, in the steady attitude expression process, the process is started when the power of the robot 1 is turned on and a predetermined initial process is completed.

  As illustrated in FIG. 12, the steady non-language operation determining unit 10 i determines whether or not a predetermined operation for instructing the end of the process has been performed (step S <b> 50). When a predetermined operation for instructing the end of the process is performed (Yes at Step S50), the process ends. On the other hand, when the predetermined operation for instructing the end of the process has not been performed (No at Step S50), the steady non-language operation determining unit 10i waits for a certain period (Step S51). This fixed period is set to an appropriate period in order to express a steady attitude by gesture, and is, for example, about several hundred milliseconds to several seconds.

  The steady non-language motion determining unit 10i derives a cycle for performing a whisper corresponding to the attitude and a cycle for matching the line of sight from the steady motion frequency data 11d (step S52). The non-language operation instruction unit 10k determines whether it is time to perform a whispering operation (step S53). For example, the non-language operation instruction unit 10k determines that it is the timing to perform the whispering operation when the whispering cycle has passed since the previous whispering operation. If it is not time to perform the whispering operation (No at Step S53), the process proceeds to Step S56 described later. On the other hand, when it is time to perform the whispering operation (Yes at Step S53), the steady non-language motion determining unit 10i determines the motion parameter of the whispering motion (Step S54). For example, the steady non-linguistic motion determination unit 10i determines the whirling depth and the whirling speed corresponding to the attitude from the whispering motion data 11c. The non-linguistic operation instruction unit 10k outputs an operation instruction for performing a whirling operation at the determined whispering depth and whirling speed to the motor control unit 40 (step S55).

  Next, the non-language operation instruction unit 10k determines whether or not it is a timing to perform an operation of aligning the line of sight with the user (step S56). For example, the non-language operation instruction unit 10k determines that it is time to perform a line-of-sight operation when a line-of-sight cycle has elapsed since the previous line-of-sight operation. If it is not time to perform the line-of-sight operation (No at Step S56), the process proceeds to Step S58 described later. On the other hand, when it is the timing to perform the operation of aligning the line of sight (Yes at Step S56), the non-language operation instruction unit 10k outputs an operation instruction to move the eyes 7 or the head 2 in the direction of the user to the motor control unit 40 (Step S56). S57).

  Next, the non-language operation instruction unit 10k determines whether or not it is time to perform a behavior such as gesture (step S58). For example, the non-language operation instruction unit 10k determines that the timing at which the attitude determined by the state determination unit 10e has changed is the timing for performing the behavior. Further, the non-language operation instruction unit 10k determines that it is the timing to perform the behavior when a predetermined period has elapsed since the last gesture was expressed. This predetermined period may be a fixed period or may be a random period within a predetermined range. When it is not time to perform the behavior (No at Step S58), the process proceeds to Step S50. On the other hand, when it is time to perform the behavior (Yes at Step S58), the non-language operation instruction unit 10k determines an operation parameter related to the gesture (Step S59). For example, the stationary non-verbal motion determination unit 10i determines the motion and facial expression from the whispering motion data 11c when expressing a facial expression as a gesture. The non-language operation instruction unit 10k outputs an operation instruction for performing the determined operation and facial expression operation to the motor control unit 40 (step S60), and proceeds to step S50.

  As described above, the robot 1 according to the present embodiment detects at least one of the utterance and action of the conversation target person, and evaluates the conversation with the conversation target person based on at least one of the detected utterance and action. Is calculated. Further, the robot 1 according to the present embodiment constantly expresses an attitude to the conversation target person by gesture. And the robot 1 which concerns on a present Example will change the attitude | position expressed to a state with high understanding with respect to a dialog, when the state where the calculated evaluation is high continues. In addition, when the calculated evaluation continues to be in a low state, the robot 1 according to the present embodiment changes the expressed attitude to a state with a low understanding with respect to the dialogue. Thereby, according to the robot 1 according to the present embodiment, it is possible to determine the conversation target person from the attitude that the robot 1 expresses, whether the robot 1 is in a high understanding state or low understanding state. Thereby, according to the robot 1 according to the present embodiment, even if the robot 1 is in a state of low understanding with respect to the dialogue and the utterance becomes unnatural, the conversation target person has a low understanding of the dialogue of the robot 1 in advance. Since it can be determined that the state is a state, it is possible to suppress an unnatural conversation with a person.

  In addition, the robot 1 according to the present embodiment, as an evaluation, increases the degree of understanding of the utterance and the degree of interest that decreases when the state of high understanding continues and decreases when the state of low understanding continues. calculate. And the robot 1 which concerns on a present Example changes the attitude | position expressed based on an understanding degree and the said interest degree. Thereby, according to the robot 1 according to the present embodiment, the attitude is changed using not only the degree of understanding of one utterance from the user but also the degree of interest, so the attitude frequently occurs for each dialogue with the user. The change can be suppressed.

  In addition, the robot 1 according to the present embodiment can express a sense of listening, unknown, and bored with respect to dialogue by gestures, and when the evaluation continues to be low, the attitude expressed is a sense of listening, unknown, bored Sequentially change the attitude. Thereby, according to the robot 1 which concerns on a present Example, when the state where an understanding is low continues, since it shifts in steps to the attitude | position which expresses more that a dialogue cannot be understood, the uncomfortable feeling of a dialogue can be suppressed.

  In addition, the robot 1 according to the present embodiment has a depth of whispering, a speed of whispering, a timing of whispering, a frequency of whispering, a timing for aligning the line of sight with respect to the conversation target, a time for aligning the line of sight, and a frequency of aligning the line of sight The attitude is constantly expressed by any of the facial expressions. Thereby, according to the robot 1 according to the present embodiment, the user can easily determine the state of the robot 1.

  Although the embodiments related to the disclosed apparatus have been described above, the present invention may be implemented in various different forms other than the above-described embodiments. Therefore, another embodiment included in the present invention will be described below.

  For example, in the above-described embodiments, the case where the attitude to be expressed is the three stages of the sense of listening, the unknown, and the bored feeling has been described, but the disclosed apparatus is not limited to this. For example, the attitude to be expressed may be two or four or more. In addition, the attitude of the expression may be gradually changed without providing a stage.

  In the above-described embodiments, the case where the degree of understanding and the degree of interest are calculated as the evaluation regarding the dialogue has been described. However, the disclosed apparatus is not limited thereto. Any value may be used as long as it is a value indicating the evaluation regarding the dialogue. For example, the ratio of words that can be understood or the measurement value obtained by the analysis by the non-linguistic expression analysis unit 10c may be used as the evaluation.

  In the above-described embodiment, the case where the interactive device is realized by the robot 1 has been described. However, the present invention is not limited to this. For example, the interactive device may be realized by a fixed terminal such as a personal computer (PC), a mobile terminal such as a mobile phone, a PHS (Personal Handyphone System), or a PDA (Personal Digital Assistant). For example, when an interactive device is realized by a fixed terminal or a mobile terminal, a face image may be displayed on a display or the like, and the attitude may be expressed by changing the face image.

[Distribution and integration]
In addition, each component of each illustrated apparatus does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution / integration of each device is not limited to that shown in the figure, and all or a part thereof may be functionally or physically distributed or arbitrarily distributed in arbitrary units according to various loads or usage conditions. Can be integrated and configured. For example, the processing units of the control unit 10 may be integrated as appropriate. Further, the processing of each processing unit may be appropriately separated into a plurality of processing units. Further, all or any part of each processing function performed in each processing unit can be realized by a CPU and a program analyzed and executed by the CPU, or can be realized as hardware by wired logic. .

[Dialogue program]
The various processes described in the above embodiments can be realized by executing a prepared program on a computer such as a personal computer or a workstation. Therefore, in the following, an example of a computer that executes a dialogue program having the same function as in the above-described embodiment, displays a face image on the display, and interacts with the user will be described with reference to FIG.

  FIG. 13 is a diagram for describing an example of a computer that executes a dialogue processing program. As illustrated in FIG. 13, the computer 100 includes an operation unit 110a, a speaker 110b, a camera 110c, a microphone 110d, a display 120, and a communication unit 130. Further, the computer 100 includes a CPU 150, a ROM 160, an HDD 170, and a RAM 180. These units 110 to 180 are connected via a bus 140.

  The HDD 170 stores in advance a dialogue program 170a that performs the same function as each processing unit of the control unit 10. The dialogue program 170a may be appropriately integrated or separated as in the case of each component shown in the first embodiment. In other words, all data stored in the HDD 170 need not always be stored in the HDD 170, and only data necessary for processing may be stored in the HDD 170.

  Then, the CPU 150 reads the interactive program 170 a from the HDD 170 and develops it in the RAM 180. Thus, as shown in FIG. 13, the dialogue program 170a functions as a dialogue process 180a. The interactive process 180a expands various data read from the HDD 170 in an area allocated to itself on the RAM 180 as appropriate, and executes various processes based on the expanded data. The dialogue process 180a includes processes executed by the respective processing units of the control unit 10, for example, the processes shown in FIGS. In addition, each processing unit virtually realized on the CPU 150 does not always require that all processing units operate on the CPU 150, and only a processing unit necessary for the processing needs to be virtually realized.

  Note that the dialogue program 170a is not necessarily stored in the HDD 170 or the ROM 160 from the beginning. For example, each program is stored in a “portable physical medium” such as a flexible disk inserted into the computer 100, so-called FD, CD-ROM, DVD disk, magneto-optical disk, or IC card. Then, the computer 100 may acquire and execute each program from these portable physical media. Each program is stored in another computer or server device connected to the computer 100 via a public line, the Internet, a LAN, a WAN, etc., and the computer 100 acquires and executes each program from these. It may be.

DESCRIPTION OF SYMBOLS 1 Robot 2 Head 4R Right arm 4L Left arm 5R Right leg 5L Left leg 6R Right ear 6L Left ear 7 Eye 7A Eyelid 9 Mouth 10 Control part 10d Evaluation calculation part 10e State determination part 41 Camera 42 Microphone 43 Speaker

Claims (5)

A detection unit for detecting the utterances of the conversation target person,
Based on the utterance detected by the detection unit, a calculation unit that calculates an evaluation regarding the dialogue with the dialogue target person;
An expression unit that constantly expresses an attitude according to one or both of the frequency of whispering and the frequency of aligning the line of sight with the subject of dialogue;
If the evaluation calculated by the calculation unit continues to be in a low state, change one or both of the contact frequency and the frequency of matching the line of sight with the person to be talked, and if the evaluation continues to be high, the contact frequency, A change unit that changes one or both of the frequency of matching the line of sight with respect to the conversation target person,
An interactive device having The calculation unit calculates the degree of understanding for the utterance, as the evaluation, rises when the comprehension that has continued high, calculating the interest degree to be reduced when the comprehension that has continued low state ,
The interactive device according to claim 1, wherein the changing unit changes an attitude expressed by the expressing unit based on the degree of interest. Prior Symbol changing unit, when the degree of interest is equal to or less than a predetermined first threshold value, modify nod frequency, the first state less one or both of the frequency of eye contact to the conversation subject, the degree of interest Is greater than the first threshold and the comprehension level is less than or equal to a predetermined second threshold, one or both of the whispering frequency and the frequency of aligning the line of sight with respect to the person to be dialogued are greater than in the first state. If the degree of interest is greater than the first threshold and the degree of understanding is greater than the second threshold, the frequency of whispering and the frequency of aligning the line of sight with the conversation target person The interactive apparatus according to claim 2, wherein one or both of the two are changed to a third state that is greater than the second state. On the computer,
Based on the utterance detected by the detection unit for detecting the utterance of the conversation target person, an evaluation regarding the dialog with the conversation target person is calculated,
If the calculated evaluation continues to be low, the frequency of whispering expressed by the expression unit that constantly expresses the attitude according to one or both of the whispering frequency and the frequency of aligning the line of sight with the subject of conversation, subject of dialogue If one or both of the frequency of matching the line of sight to the person is changed to a low level and the evaluation continues to be high, the frequency of whispering expressed by the display unit, the frequency of matching the line of sight to the conversation target person or An interactive program characterized by executing a process that changes both. Computer
Based on the utterance detected by the detection unit for detecting the utterance of the conversation target person, an evaluation regarding the dialog with the conversation target person is calculated,
If the calculated evaluation continues to be low, the frequency of whispering expressed by the expression unit that constantly expresses the attitude according to one or both of the whispering frequency and the frequency of aligning the line of sight with the subject of conversation, subject of dialogue If one or both of the frequency of matching the line of sight to the person is changed to a low level and the evaluation continues to be high, the frequency of whispering expressed by the display unit, the frequency of matching the line of sight to the conversation target person or An interactive method characterized by executing a process that changes both.

Images