JP2017100221A - Communication robot - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a communication robot that can support health care of a user.SOLUTION: A communication robot 20, which is configured to have a conversation with a person, comprises: a microphone 21; a speaker 22; a voice recognizing portion 23 that recognizes voice inputted from the microphone 21; a voice synthesis portion 25 that outputs voice from the speaker 22 on the basis of a recognized result by the voice recognizing portion 23; and a biological state estimating portion 24 that estimates a biological state of a person producing voice, on the basis of an acoustic property of the voice inputted from the microphone 21.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a communication robot that has a conversation with a person.

  In recent years, various communication robots that have conversations with people have been proposed (for example, Patent Document 1).

  In Patent Document 1, as an example of a communication robot, a pet-type mobile robot including a closeness evaluation unit that determines a closeness with a mobile communication device based on a history of wireless connection with another mobile communication device is proposed. ing. According to this technique, a message corresponding to the familiarity is output, and communication with the user is activated.

International Publication No. 2008/152943

  However, in the technique of Patent Document 1, although communication of a user in a normal health state is activated, there is a problem that when the user is not healthy due to poor physical condition or the like, it can no longer cope with it. If the user is in poor health, the person (caregiver, family, etc.) who manages the user's health knows that as soon as possible, rather than activating communication with the pet-type mobile robot. It is necessary to deal with it. However, the technique of Patent Document 1 has a problem that it cannot cope with such a case.

  Therefore, the present invention has been made in view of the above problems, and an object of the present invention is to provide a communication robot that can support user health management.

  In order to achieve the above object, a communication robot according to an aspect of the present invention is a communication robot that has a conversation with a person, and includes a microphone, a speaker, and a voice recognition unit that recognizes a voice input from the microphone. Based on the recognition result by the voice recognition unit, the voice synthesis unit that outputs the voice from the speaker, and the biological state of the person who emitted the voice is estimated based on the acoustic characteristics of the voice input from the microphone A living body state estimation unit.

  The present invention provides a communication robot that can support the health management of a user.

The block diagram which shows the structure of the communication robot in embodiment of this invention Sequence diagram showing the operation procedure in a scene where the user and the communication robot have a conversation Time chart showing an example of biological state estimation by the biological state estimation unit of a communication robot FIG. 3A is a time chart showing an example of the life log of the biological state shown in FIG. Plot diagram showing another estimation example of the biological state by the biological state estimation unit of the communication robot Plot diagram in which estimation example of biological state shown in FIG. 4A is divided for each user The image figure which shows a mode that the update of the judgment standard in the estimation of the biological state is executed by the cloud

(Knowledge that became the basis of the present invention)
Due to the increasing burden of caregivers due to the aging society and the shortage of caregivers in recent years, the elderly who live alone or are hospitalized at homes, nursing homes, medical facilities, etc. (Hereinafter referred to as “caregiver”) or the surrounding elderly. Elderly emotions are determined by biometric information (blood pressure, pulse, etc.) and the caregiver's own subjectivity. Therefore, an appropriate dialogue is possible if the care period is long, but a new caregiver or a slight care period tends to be insufficient. In addition, the life transition of emotional transition is limited only during the care period, and relies heavily on the caregiver's memory and notes, and lacks objectivity. Furthermore, even in a kind of “closed facility”, even if the undulations (especially depression) of emotions are grasped, the subsequent response is left to the caregiver. Under such circumstances, there arises a problem that the burden on the caregiver increases and a problem that information sharing with the family about the health condition of the elderly is stagnant.

  In addition, the frequency of answering only by children is increasing due to the recent advancement of co-working and women. In connection with this, the case where the talk partner of the child who lives in or is hospitalized in a nursing care facility, a medical facility, etc. is a caregiver or a patient in the same room is increasing. It is difficult to detect the emotion of a child who is answering alone. If the time period for which an answering machine has to be taken is long, communication between parents and children can be adversely affected. In addition, the emotion of a child in a facility is often judged by biometric information (blood pressure, pulse, etc.) and the caregiver's own subjectivity. Longer care periods allow for appropriate dialogue, but newer caregivers and minor care periods tend to be inadequate. In addition, the life transition of emotional transition is limited only during the care period, and relies heavily on the caregiver's memory and notes, and lacks objectivity. Furthermore, even in a kind of “closed facility”, even if the undulations (especially depression) of emotions are grasped, the subsequent response is left to the caregiver. Under such circumstances, there arises a problem that the burden on the caregiver increases and a problem that information sharing with the family about the health condition of the child is stagnant.

  Therefore, the present invention has a problem that the burden on the caregiver increases as described above, a problem that information sharing with the family about the health status of the elderly is stagnant, and information about the health status of the child with the family It was made to solve the problem of stagnation of sharing. That is, an object of the present invention is to provide a communication robot that can support the health management of a user.

  For this purpose, the present invention is a communication robot that has a conversation with a person, a microphone, a speaker, a voice recognition unit that recognizes voice input from the microphone, and a voice from the speaker based on a recognition result by the voice recognition unit. And a biological state estimating unit that estimates the biological state of the person who has emitted the voice based on the acoustic characteristics of the voice input from the microphone.

  As a result, the biological state of the user is estimated based on the voice uttered by the user such as an elderly person or a child. Therefore, by providing a mechanism for notifying caregivers and family members of the estimation results, the burden on caregivers is reduced, sharing of information about the user's health status with the family is promoted, and user health management is supported. .

(Embodiment)
Hereinafter, embodiments of a communication robot according to the present invention will be described in detail with reference to the drawings. Note that each of the embodiments described below shows a specific example of the present invention. Numerical values, shapes, materials, components, arrangement positions and connection forms of components, steps, order of steps, and the like shown in the following embodiments are merely examples, and are not intended to limit the present invention. In addition, among the constituent elements in the following embodiments, constituent elements that are not described in the independent claims indicating the highest concept of the present invention are described as optional constituent elements.

  FIG. 1 is a block diagram showing a configuration of a communication robot 20 in the embodiment of the present invention. Here, a user 10 (here, an elderly person) who has a conversation with the communication robot 20 is also illustrated.

  The communication robot 20 is a robot that has a conversation with a person (user 10), and includes a microphone 21, a speaker 22, a voice recognition unit 23, a biological state estimation unit 24, a voice synthesis unit 25, a conversation engine 26, a bus 27, and a storage unit 28. And a communication unit 29. The communication robot 20 is illustrated as an example of a pet-type mobile robot in the figure, but any type of electronic device may be used as long as it is an electronic device that has a conversation with a person, and includes a robot that does not move. Examples of the robot that does not move include home appliances having a conversation function, or a HEMS (Home Energy Management System) controller having a conversation function installed on a wall of a building or the like. Further, in FIG. 1, illustration of general components not related to the present invention (a visual sensor, a tactile sensor, a mechanism such as an actuator, an operation sensor such as an encoder, etc. provided in a pet-type mobile robot) is omitted.

  The microphone 21 is a device that collects voices uttered by a user and converts them into electrical signals. For example, the microphone 21 is provided in both ears of a pet type mobile robot.

  The speaker 22 is a device that converts an electrical signal into sound and outputs it. For example, the speaker 22 is provided at the mouth of a pet-type mobile robot. In this specification, “voice” output to the user includes not only human voice but also all kinds of sounds such as music and mechanical sounds.

  The speech recognition unit 23 is a processing unit that recognizes speech input from the microphone 21 and is realized by, for example, a speech recognition IC or software and a processor that executes the software. The voice recognition unit 23 includes a voice print personal authentication unit 23a that identifies the user 10 who has made a voice by referring to a previously registered voice print. The voiceprint personal authentication unit 23a stores therein, as voiceprints, feature data obtained by the voice recognition unit 23 for voices input in advance for each registrant. And the user 10 is identified by collating the feature data obtained in the speech recognition part 23 with the voiceprint about the now input speech.

  The biological state estimation unit 24 is a processing unit that estimates the biological state of the user 10 who has emitted the voice based on the acoustic characteristics of the voice input from the microphone 21 (that is, physical characteristics that are not the characteristics of words). For example, it is realized by software and a processor that executes the software. Here, for example, depending on whether the user 10 is nervous or relaxed, the vocal cords contract or relax, so that the acoustic features such as the size of the voice in the voice, the utterance length of the breath, and the like may change. Used.

  Here, the biological state estimation unit 24 estimates a biological state for each person identified by the voiceprint personal authentication unit 23a, and stores information indicating the estimated biological state in the storage unit 28 as a life log. repeat. At this time, the biological state estimation unit 24 stores information indicating the biological state in the storage unit 28 as a life log while updating the determination criterion used when estimating the biological state. In addition, the biological state estimation unit 24 determines whether or not the biological state has changed by analyzing the log stored in the storage unit 28, and when it is determined that the biological state has changed, the biological state has changed. This is notified (in this case, a predetermined notification destination is notified via the communication unit 29). Note that “notification” includes not only a notification to a specific notification destination but also an operation of simply outputting information such as a display or an alarm.

  More specifically, the biological state estimation unit 24 calculates the probability that a human emotion corresponds to each of a plurality of predetermined emotions based on the acoustic characteristics of the voice input from the microphone 21, thereby Estimate the state. Moreover, the biological state estimation part 24 can estimate whether a person is healthy or not healthy as a biological state. Specifically, the biological state estimation unit 24 calculates the degree of each of a plurality of parameters representing human emotion based on the acoustic feature, and the combination of the calculated degrees of the parameters is in the first range. In this case, it is estimated that the person is healthy, and the person is inferior when the combination of the degrees of the plurality of parameters is in the second range.

  The conversation engine 26 is a processing unit that determines a voice to be uttered by the communication robot 20 based on a recognition result by the voice recognition unit 23, and is realized by a conversation database, software, a processor that executes software, and the like. In the present embodiment, the conversation engine 26 is, for example, a conversation bot (conversation bot) that is a program that simulates an intelligent conversation with a person by voice by matching with an internal database.

  The voice synthesis unit 25 is a processing unit that outputs voice from the speaker 22 based on the recognition result by the voice recognition unit 23. In the present embodiment, the speech synthesizer 25 is realized by a speech synthesis IC that outputs speech in accordance with an utterance instruction from the conversation engine 26, or software and a processor that executes the software.

  The bus 27 is a signal path that electrically connects the voice recognition unit 23, the biological state estimation unit 24, the conversation engine 26, the storage unit 28, and the communication unit 29.

  The storage unit 28 is a memory for storing various information including a life log that is information indicating the biological state estimated by the biological state estimation unit 24, and is, for example, a nonvolatile RAM.

  The communication unit 29 is an interface for communicating with other communication devices (such as a portable information terminal and a server device on the Internet), and is a communication interface for wireless communication such as a wireless LAN or wired communication.

  In addition, the communication robot 20 may include an input device such as a display unit that displays an estimation result or the like in the biological state estimation unit 24 or a button that receives an instruction from the user 10 as a component that is not illustrated.

  Next, the operation of the communication robot 20 in the present embodiment configured as described above will be described.

  FIG. 2 is a sequence diagram showing an operation procedure in a scene where the user 10 (here, an elderly person) and the communication robot 20 have a conversation.

  First, communication robot 20 utters "Hello" (S10). Specifically, the speech synthesis unit 25 in accordance with an instruction from the conversation engine 26, the speaker 22 outputs sound "hello".

  In contrast, the user 10, for example, speaks "Hello" (S11). This voice is input to the voice recognition unit 23 via the microphone 21 of the communication robot 20, and the voice is recognized by the voice recognition unit 23. Further, the voiceprint personal authentication unit 23a identifies the user 10 by comparing the feature data obtained by the voice recognition unit 23 with the voiceprint stored in advance for each registrant for the voice that has just been input.

  In the communication robot 20, in parallel with or immediately after the voice recognition, the biological state estimation unit 24 determines the biological state of the user 10 who has emitted the voice based on the acoustic characteristics of the voice input from the microphone 21. Estimate (S12). More specifically, the biological state estimation unit 24 estimates the biological state for each person identified by the voiceprint personal authentication unit 23a, and associates the information indicating the estimated biological state with the identified user 10 as a life log. And stored in the storage unit 28. Here, it is assumed that the biological state estimation unit 24 estimates that the user 10 is in a malfunction.

  Subsequently, in the communication robot 20, the speech synthesizer 25 provides the speaker with the next speech “something seems to be unsatisfactory” determined by the conversation engine 26 according to the estimation result (here, “bad”) in the biological state estimation unit 24. 22 (S13).

  On the other hand, the user 10 utters, for example, “No such thing” (S14). This voice is input to the voice recognition unit 23 via the microphone 21 of the communication robot 20 as in step S11, and the voice recognition unit 23 recognizes the voice. Further, the voiceprint personal authentication unit 23a identifies the user 10 by comparing the feature data obtained by the voice recognition unit 23 with the voiceprint stored in advance for each registrant for the voice that has just been input.

  Further, in the communication robot 20, in parallel with or immediately after the voice recognition, the biological state estimation unit 24 is based on the acoustic characteristics of the voice input from the microphone 21, and the biological body of the user 10 who has emitted the voice. The state is estimated (S15). More specifically, the biological state estimation unit 24 estimates the biological state for each person identified by the voiceprint personal authentication unit 23a, and associates the information indicating the estimated biological state with the identified user 10 as a life log. And stored in the storage unit 28.

  When a series of conversations have been interrupted for more than a predetermined time (silent state has continued), when a predetermined time has elapsed, or when a predetermined number of biological states have been estimated, the biological state estimating unit 24 The case is detected by an internal timer or the like, and the estimation result is notified (S16). For example, the biological state estimation unit 24 determines whether the biological state has changed by analyzing a life log stored in the storage unit 28, and if the biological state has changed, the biological state changes. This is notified to a predetermined notification destination via the communication unit 29. The predetermined notification destination is, for example, a caregiver who takes care of the user 10 or a family of the user 10.

  According to such a processing procedure, the biological state of the user 10 is estimated based on the voice uttered by the user 10, and the estimation result is notified to a caregiver, a family, and the like. Thus, for example, a caregiver or family member who has received a “not good” notification immediately goes to the user 10, while a caregiver or family member who has received the “healthy” notification goes to the user 10. I can refrain from that. Thus, the burden on the caregiver is reduced, information sharing with the family about the user's health condition is promoted, and the health management of the user 10 is supported.

  FIG. 3A is a time chart illustrating an example of estimation of a biological state by the biological state estimation unit 24 of the communication robot 20. Here, the temporal changes of the probabilities (here, percentage (%)) of four psychological states (four kinds of emotions “calm”, “sadness”, “anger”, “joy”) are shown.

  As shown in the figure, the biological state estimating unit 24 uses a plurality of predetermined emotions (“quiet” as the biological state based on the acoustic characteristics of the voice input from the microphone 21 as the biological state. ”,“ Sadness ”,“ anger ”,“ joy ”)). This process is performed at predetermined time intervals during conversation or every predetermined number of utterances from the user 10.

  More specifically, the biological state estimation unit 24 extracts acoustic feature data from the voice input from the microphone 21 based on changes in voice pitch (frequency) and volume. Then, the biological state estimation unit 24 uses the machine learning such as SVM (“Support Vector Machine”) or deep learning for the extracted feature data, and calculates the probability (“Rakusei”) of each of the four types of emotions. Calculate in (%). As a result, the total of the probabilities of the four types of emotions is normalized to be 100 at any time (for example, each probability on the dotted line indicated by the arrow in FIG. 3A).

  Information indicating the biological state estimated by the biological state estimation unit 24 is stored as a life log in a communication device via the storage unit 28 or the communication unit 29 for each person identified by the voiceprint personal authentication unit 23a. The

  FIG. 3B is a time chart showing an example of a life log of the biological state shown in FIG. 3A. Here, as a life log for each user (Mr. A, Mr. B) stored in the storage unit 28 by the biological state estimation unit 24, one psychological state (for example, “sorrow”) selected from four psychological states. The daily change in the probability (“appearance rate (%)”) is shown.

  As shown in the figure, the biological state estimating unit 24 averages the log stored in the storage unit 28 for each user on a daily basis, thereby showing changes in the biological state of the user every day. A curve can be obtained. Then, the biological state estimating unit 24 determines whether or not the biological state has changed by determining whether or not there is a part exceeding the predetermined range or a part that has changed beyond the predetermined width in the curve. Judging. When it is determined that the biological state has changed, the biological state estimation unit 24 notifies the mobile information terminal or the like possessed by a predetermined caregiver or family via the communication unit 29 that the biological state has changed. Notify first.

  Thus, according to the communication robot 20 of the present embodiment, when the biological state of the user 10 changes from the normal time, this is notified to the caregiver who takes care of the user 10 or the family of the user 10. . Thereby, the burden of the caregiver who takes care of the user 10 is reduced, information sharing with the family about the health state of the user 10 is promoted, and the health management of the user 10 is supported.

  In FIG. 3B, the daily change was analyzed for the probability (“appearance rate (%)”) of one psychological state selected from the four psychological states shown in FIG. 3A. The probabilities (“appearance rate (%)”) of two or more psychological states selected from the states may be analyzed. For example, when the psychological state combining “sorrow” and “anger” is set to be analyzed (an instruction to the biological state estimating unit 24 is given in advance), the biological state estimating unit 24 The same analysis is performed for the average value of the probabilities of “sadness” and “anger”.

  FIG. 4A is a plot diagram illustrating another estimation example of the biological state by the biological state estimation unit 24 of the communication robot 20. Here, the coordinate position determined by the combination of the excitement level (horizontal axis) and the stress level (vertical axis) of the user 10 calculated by the biological state estimation unit 24 during a certain period (November 1 to November 8). The diamond points are plotted along with the detected date. Then, it is shown that the user 10 is determined to be “healthy” or the user 10 is determined to be “bad” who is unhealthy as a biological state by the combination of the excitement level and the stress level.

  That is, the biological state estimation unit 24 calculates the degree of excitement and the degree of stress as a plurality of parameters representing human emotions based on the acoustic features input from the microphone 21. As a method of calculating the degree of excitement and the degree of stress, a relationship that is determined in advance from the probabilities of the four psychological states (four types of emotions “calm”, “sadness”, “anger”, “joy”) shown in FIG. It may be calculated using an equation, or may be the same method as the estimation of the four psychological states shown in FIG. 3A. For example, the average value of the probabilities of “sadness”, “anger” and “joy” may be “excitability”, or acoustic feature data is extracted from speech and machine learning such as SVM or deep learning is performed. The degree of excitement and the degree of stress may be newly calculated.

  The biological state estimation unit 24 estimates that the person is healthy when the combination of the calculated degree of excitement and the degree of stress is in the first range, and the combination of the degree of excitement and the degree of stress is in the second range. It is estimated that the person is out of order. The first range is a range in which the degree of excitement is lower than a predetermined first threshold and the degree of stress is lower than a predetermined second threshold. The second range is a range in which the degree of excitement is higher than a predetermined third threshold and the degree of stress is higher than a predetermined fourth threshold.

  Note that the information indicating the biological state estimated in this manner is a caregiver or user 10 who takes care of the user 10 via the communication unit 29 every time it is estimated or every time it is estimated to be “malfunction”. Notification is sent to a notification destination such as a portable information terminal held by a family member. Further, information indicating the biological state estimated in this way is stored as a life log in a communication device via the storage unit 28 or the communication unit 29 for each person identified by the voiceprint personal authentication unit 23a. Used for analysis as shown in.

  As described above, according to the communication robot 20 of the present embodiment, “health” or “unhealthy” is determined as the biological state, and the determination result or the like takes care of the user 10 or the family of the user 10. Will be notified. Thus, for example, a caregiver or family member who has received a “not good” notification immediately goes to the user 10, while a caregiver or family member who has received the “healthy” notification goes to the user 10. I can refrain from that. Thus, the burden on the caregiver is reduced, information sharing with the family about the user's health condition is promoted, and the health management of the user 10 is supported.

  FIG. 4B is a plot diagram in which the estimation example of the biological state shown in FIG. 4A is divided for each user. Here, it is shown that the first range determined as “healthy” and the second range determined as “malfunction” differ for different users. Here, an estimation example of Mr. A's biological state (FIG. 4B (a)) and an estimation example of Mr. B's biological state (FIG. 4B (b)) are shown.

  As shown in this figure, the biological state estimation unit 24 uses the first range and the second range associated with each user to determine “health” and “unwell” from the combination of the degree of excitement and the degree of stress. Judging. At this time, the biological state estimation unit 24 stores information indicating the biological state as a life log while updating the determination standard (first range and second range for each user) used when estimating the biological state. 28 is stored. Regarding the updating of the judgment criteria, for example, supervised machine learning such as SVM or deep learning using teacher data (actual “health” / “strange”) given by the user 10 or without using teacher data. Or it is performed by unsupervised machine learning.

  It should be noted that the estimation of the biological state or the update of the judgment standard in the estimation is not only performed by the communication robot 20. You may perform by the cloud computing (henceforth "cloud") connected with the communication part 29 of the communication robot 20. FIG.

  FIG. 5 is an image diagram illustrating a state in which the determination criteria are updated by the cloud in estimating the biological state. The biological state estimation unit 24 of the communication robot 20 uploads emotion data (time change in probability of four types of emotions) as shown in FIG. 3A to the cloud as a component of big data via the communication unit 29. To do. In the cloud, deep learning is used to update determination criteria (first range and second range for each user) used to determine “health” and “malfunction” using big data. When the update is completed, the biological state estimation unit 24 of the communication robot 20 receives the updated determination criterion from the cloud via the communication unit 29, and uses the received determination criterion to estimate the subsequent biological state. Do.

  In this way, by estimating the biological state or updating the judgment criteria in the estimation using the cloud, the processing time of machine learning for the estimation of the biological state can be shortened, or the biological state can be determined with more accurate judgment criteria. It is possible to estimate the state.

  It should be noted that the switching of the biological state estimation method (the method shown in FIG. 3A or the method shown in FIGS. 4A and 4B) and the use / non-use of the cloud in the communication robot 20 is realized in any form. Also good. For example, information previously set (switching information is stored) in the communication robot 20 may be referred to, or may be determined by a dialog with a user via a display unit and an input device provided in the communication robot 20.

  As described above, the communication robot 20 in the present embodiment is a communication robot 20 that has a conversation with a person, and includes a microphone 21, a speaker 22, and a voice recognition unit 23 that recognizes a voice input from the microphone 21. Based on the recognition result by the voice recognition unit 23, the voice synthesis unit 25 that outputs the voice from the speaker 22, and the biological state of the person who emitted the voice is estimated based on the acoustic characteristics of the voice input from the microphone 21. And a biological state estimation unit 24.

  Thereby, based on the acoustic feature of the voice uttered by the user of the communication robot 20, the biological state of the person who uttered the voice is estimated. Therefore, when the user's caregiver or family member uses the estimation result, communication with the user is maintained, and at the same time, the user's health management is supported.

  In addition, the voice recognition unit 23 includes a voice print personal authentication unit 23a that identifies a person who has made a voice by referring to a previously registered voice print.

  As a result, it is possible to estimate the biological state after distinguishing the users.

  Further, the communication robot 20 includes a storage unit 28, and the biological state estimation unit 24 estimates the biological state for each person identified by the voiceprint personal authentication unit 23a, and uses information indicating the estimated biological state as a log. The processing stored in the storage unit 28 is repeated.

  Thereby, a life log is accumulated for each user, and health management for each user becomes possible.

  In addition, the biological state estimation unit 24 further determines whether the biological state has changed by analyzing the log stored in the storage unit 28, and if it is determined that the biological state has changed, the biological state is Notify that it has changed.

  As a result, by utilizing the life log, a change in the biological state is detected and notified in consideration of the past biological state. By setting the notification destination to the user's caregiver or family, the biological body When the state changes, this is notified to a caregiver or family member. Therefore, a caregiver or family member who has received a notification that the biological condition has changed promptly goes to the user. On the other hand, if there is no notification, the user is expected to be healthy as usual, A caregiver or family member can refrain from visiting the user. As a result, the burden on the caregiver is reduced, information sharing with the family about the user's health condition is promoted, and the user's health management is supported.

  In addition, the biological state estimation unit 24 stores information indicating the biological state in the storage unit 28 as a log while updating the determination criterion used when estimating the biological state.

  As a result, the criterion for estimating the biological state is learned and corrected to a more accurate value, so that the estimation accuracy of the biological state is improved as time passes.

  Moreover, the biological state estimation unit 24 estimates the biological state by calculating a probability that a human emotion corresponds to each of a plurality of predetermined emotions based on the acoustic features.

  As a result, probabilities are calculated for each of a plurality of limited representative emotions as the biological state, so that high-precision estimation is performed in a short processing time by estimation in a narrowed range.

  In addition, the biological state estimation unit 24 estimates whether the human state is healthy or not healthy. More specifically, the biological state estimation unit 24 calculates the degree of each of the plurality of parameters representing the human emotion based on the acoustic feature, and the combination of the calculated degrees of the parameters falls within the first range. In some cases, it is estimated that the person is healthy, and when the combination of the degrees of the plurality of parameters is in the second range, it is estimated that the person is out of order.

  As a result, a plurality of parameters used for judgment of health or malfunction are set as appropriate, and further, the first range and the second range as threshold values are set as appropriate values. By doing so, health management based on a desired judgment standard is realized.

  While the communication robot of the present invention has been described based on the embodiment, the present invention is not limited to this embodiment. Without departing from the gist of the present invention, various modifications conceived by those skilled in the art have been made in the present embodiment, and other forms constructed by combining some components in the embodiment are also within the scope of the present invention. Contained within.

  For example, in the embodiment, the communication robot 20 includes the speech recognition unit 23, the biological state estimation unit 24, the speech synthesis unit 25, and the conversation engine 26. However, the configuration is not limited thereto. At least one of the speech recognition unit 23, the biological state estimation unit 24, the speech synthesis unit 25, and the conversation engine 26 is provided in a device separate from the communication robot 20, for example, a computer device or a cloud connected via a communication path. May be. Furthermore, at least one of the speech recognition unit 23, the biological state estimation unit 24, the speech synthesis unit 25, and the conversation engine 26 is provided in a distributed manner in each of the communication robot 20 and a separate device, and communicates between both devices. You may be comprised so that a function may be demonstrated while fitting.

  In addition, the speech recognition unit 23, the biological state estimation unit 24, the speech synthesis unit 25, and the conversation engine 26 included in the communication robot 20 may be configured as software executed by a common processor when implemented by software. Good. The software may be stored in a computer-readable recording medium such as a memory or a DVD (Digital Versatile Disc).

DESCRIPTION OF SYMBOLS 20 Communication robot 21 Microphone 22 Speaker 23 Voice recognition part 23a Voiceprint personal authentication part 24 Living body state estimation part 25 Speech synthesis part 28 Storage part

Claims (8)

A communication robot that talks to people
A microphone,
Speakers,
A speech recognition unit for recognizing speech input from the microphone;
A voice synthesis unit that outputs voice from the speaker based on a recognition result by the voice recognition unit;
A communication robot comprising: a biological state estimating unit that estimates a biological state of a person who has emitted the voice based on an acoustic feature of the voice input from the microphone. The communication robot according to claim 1, wherein the voice recognition unit includes a voiceprint personal authentication unit that identifies a person who has made the voice by referring to a voiceprint registered in advance. Furthermore, a storage unit is provided,
The biological state estimating unit repeats a process of estimating the biological state for each person identified by the voiceprint personal authentication unit, and storing information indicating the estimated biological state in the storage unit as a log. 2. The communication robot according to 2. The biological state estimation unit further determines whether the biological state has changed by analyzing the log stored in the storage unit, and if the biological state has changed, the biological state changes The communication robot according to claim 3. The communication robot according to claim 3, wherein the biological state estimating unit stores information indicating the biological state in the storage unit as the log while updating a criterion used when estimating the biological state. . The biological state estimation unit estimates the biological state by calculating a probability that the human emotion corresponds to each of a plurality of predetermined emotions based on the acoustic feature. The communication robot according to any one of the above. The communication robot according to claim 1, wherein the biological state estimation unit estimates whether the person is healthy or not healthy as the biological state. The biological state estimation unit calculates a degree of each of a plurality of parameters representing human emotion based on the acoustic feature, and when the combination of the calculated degrees of the parameters is in a first range, The communication robot according to claim 7, wherein the person is estimated to be healthy, and the person is estimated to be out of order when the combination of the degrees of the plurality of parameters is in the second range.

Images