JP2021174204A - Behavior estimation device, method, and program - Google Patents

Abstract

To allow a behavior of an object user to be further accurately estimated regardless of the user's situations without physical and mental burdens on the user.SOLUTION: A behavior estimation device, method, and program causes an interactive device to present question information to an object user and acquires answer information of the object user to the question information from the interactive device to estimate a behavior of the object user. The behavior estimation device, method, and program detect an initial behavior of the object user, determine a cognitive load of the object user during the initial behavior on the basis of the detected initial behavior, generate the question information to the object user on the basis of the determined cognitive load and a behavior scene estimated from the initial behavior, and output the generated question information to the interactive device.SELECTED DRAWING: Figure 3

Description

The present invention relates to, for example, a behavior estimation device, a method, and a program for estimating the behavior of a target user using an interactive robot.

With the development of an aging society, the proportion of living alone in the elderly population aged 65 and over is increasing year by year. Elderly people living alone have few contacts with society and tend to withdraw, so there is a risk that dementia will progress unknowingly.

So far, for the diagnosis of dementia in the elderly, as described in Non-Patent Document 1, for example, a method of testing by a doctor or the like having specialized knowledge according to a predetermined algorithm has been used. However, when receiving a diagnosis, elderly people are often forced to have the physical burden of going to a specialized place such as a hospital and the mental burden of taking a test.

On the other hand, a system for diagnosing dementia based on natural conversation with an elderly person living alone by utilizing a humanoid communication robot has also been developed (see, for example, Non-Patent Document 2). By using this system, doctors can make a diagnosis without imposing a physical burden such as movement on the elderly and without being aware that there is a dementia test as much as possible. Furthermore, it is also possible to notify close relatives such as tribes of the diagnosis result of dementia via social media or the like.

Hasegawa Dementia Scale, Internet <URL: https://info.ninchisho.net/check> Toru Kobayashi, Taishi Miyazaki, Kenichi Arai, "Research on Early Detection System for Dementia by Social Media Mediated Robots", Multimedia, Distributed, Cooperative and Mobile (DICOMO2018) Symposium, July 2018, 7A-3, p.1332

By the way, in order to comprehensively diagnose dementia, it has been found that it is necessary not only to grasp cognitive dysfunction but also to grasp behavioral disorders in daily life of the target user. However, neither of Non-Patent Documents 1 and 2 describes the point of grasping the behavioral disorder of the target user. Further, as a method of grasping the behavioral disorder of the target user, for example, there is a method of hearing with people around the target user. However, this method was difficult to apply to target users living alone in the first place, and it was difficult to grasp behavioral disorders at an early stage.

The present invention has been made by paying attention to the above circumstances, and on one aspect, the behavior of the target user is estimated more accurately without imposing a physical and mental burden on the user and regardless of the user's situation. It is to provide the technology that makes it possible.

In order to achieve the above object, one aspect of the present invention is to make a target user present question information by a dialogue device, and acquire the answer information of the target user to the question information from the dialogue device. A behavior estimation device, method, or program that estimates behavior detects the initial behavior of the target user, and determines the cognitive load of the target user during the initial behavior based on the detected initial behavior. Based on the determined cognitive load and the behavior scene estimated from the initial behavior, the question information for the target user is generated, and the generated question information is output to the dialogue device. It is a thing.

According to one aspect of the present invention, when presenting the question information according to the initial action of the target user, the question information in consideration of the cognitive load of the target user during the initial action is generated and output. Therefore, it is possible to obtain a highly valid answer from the target user, which makes it possible to more accurately estimate the behavior of the target user.

That is, according to one aspect of the present invention, there is provided a technique capable of estimating the behavior of a target user more accurately regardless of the user's situation without imposing a physical and mental burden on the user. can do.

FIG. 1 is a diagram showing an example of an overall configuration of a system including a behavior estimation device according to an embodiment of the present invention. FIG. 2 is a block diagram showing a hardware configuration of the behavior estimation device according to the embodiment of the present invention. FIG. 3 is a block diagram showing a software configuration of the behavior estimation device according to the embodiment of the present invention. FIG. 4 is a flowchart showing a processing procedure and processing contents by the behavior estimation device shown in FIG. FIG. 5 is a flowchart showing the procedure and the processing content of the robot dialogue processing among the processing procedures shown in FIG. FIG. 6 is a diagram showing an example of a registration screen displayed on the management terminal. FIG. 7 is a diagram showing an example of device information stored in the device DB. FIG. 8 is a diagram showing an example of scenario information stored in the scenario DB. FIG. 9 is a diagram showing an example of cognitive load and dialogue group allocation for a scene. FIG. 10 is a diagram showing a list of lines prepared in advance for the cognitive load. FIG. 11 is a diagram showing an example of the structure of data showing the scenario execution result stored in the question / answer DB.

Hereinafter, embodiments according to the present invention will be described with reference to the drawings.
[One Embodiment]
(Configuration example)
(1) System FIG. 1 is a diagram showing an overall configuration of a system including a behavior estimation device according to an embodiment of the present invention.
For example, an interactive robot (hereinafter sometimes referred to as a robot) RB as a dialogue device is installed in the house of the target user US such as an elderly person living alone in order to have a dialogue with the target user US. Further, in the house, a plurality of sensor devices D1 to Dn are installed in order to detect the behavior of the target user US in the house.

The sensor devices D1 to Dn are attached to, for example, a sensor device that detects opening and closing of windows, curtains, doors of toilets and bathrooms, doors of refrigerators and washing machines, FDs of medicine boxes, lids of medicine boxes, and the wrists of the target user US. Includes wearable devices. The former has, for example, an acceleration sensor, and when it detects the opening and closing of each of the doors WD and FD, it transmits sensor data indicating the detection result to the interactive robot RB. The latter has, for example, an acceleration sensor or a voice sensor, detects the number of steps and conversation time of the target user US, and transmits sensor data indicating the detection result to the interactive robot RB. As the sensor data transmission means, a wireless LAN (Local Area Network) or a wireless interface using a low power data communication standard such as Bluetooth (registered trademark) is used.

The system also includes a server computer (hereinafter referred to as a server device) SV that operates as a behavior estimation device. The server device SV is installed, for example, in the cloud, the Web, or the edge. Then, the server device SV is connected to the interactive robot RB, the management terminal MT used by the system administrator, and the close relative terminal PT used by the close relatives of the target user US via the network NW. Each can perform data communication. The server device SV may be installed in the house in a state of being attached to the interactive robot RB or in a state of being built in the interactive robot RB.

The network NW includes, for example, a wide area network centered on the Internet and an access network for accessing the wide area network. As the access network, for example, a wired and wireless public network, a wired and wireless LAN, and a CATV (Cable Television) network are used.

(2) Device FIG. 2 is a block diagram showing the hardware configuration of the server device SV, and FIG. 3 is a block diagram showing the functional configurations of the server device SV, the interactive robot RB, the management terminal MT, and the close relative terminal PT.

(2-1) Management terminal MT and close relative terminal PT
The management terminal MT is composed of, for example, a personal computer, and has a registration setting processing unit 61 as a processing function according to the present invention. The registration setting processing unit 61 is used, for example, for an administrator who operates the system to register device information and scenario information in the server device SV.

The close relative terminal PT is composed of a mobile information terminal such as a notebook type personal computer, a smartphone or a tablet terminal, and has a question / answer transmission / reception unit 71 and a notification information reception / display unit 72 as processing functions according to the present invention. .. The question / answer transmission / reception unit 71 receives the question information transmitted from the server device SV to the close relatives, displays or outputs the voice, and returns the answer information input by the close relatives to the server device SV. do. The notification information reception display unit 72 receives and displays the notification information transmitted from the server device SV. The notification information includes, for example, information representing an estimation result of the degree of behavioral disorder of the target user US.

(2-2) Interactive robot RB
The interactive robot RB has a sensor data transmission / reception unit 51, a voice synthesis processing unit 52, a voice output unit 53, and a voice input unit 54 as functions according to the present invention. The interactive robot RB may be fixedly installed on the floor or table like a general smart speaker, and the face and limbs move in conjunction with the dialogue movement, or the caller moves autonomously. May be provided.

The sensor data transmission / reception unit 51 receives the sensor data transmitted from the sensor devices D1 to Dn, and transfers the received sensor data to the server device SV. The voice synthesis processing unit 52 receives question information composed of, for example, text data sent from the server device SV, and synthesizes voice data corresponding to the question text based on the received text data. The voice output unit 53 outputs the voice corresponding to the voice data output from the voice synthesis processing unit 52 from the speaker.

The voice input unit 54 detects the voice representing the answer content issued by the target user US with a microphone, encodes the detected voice data, and transmits the detected voice data to the server device SV. When the interactive robot RB is provided with a voice recognition processing unit, the voice recognition processing unit recognizes the answer sentence from the above voice data and transmits the information representing the recognized answer sentence to the server device SV. You may send it.

(2-3) Server device SV
The server device SV has the control function of the interactive robot RB, and uses this robot control function to perform an operation of estimating the behavior of the target user US. Therefore, the server device SV includes a control unit 1 having a hardware processor such as a central processing unit (CPU), and the program storage unit 2 and the data storage unit are provided to the control unit 1 via the bus 5. The storage unit having 3 is connected to the communication interface (communication I / F) 4.

The communication I / F4 transmits data between the management terminal MT, the close relative terminal PT, and the interactive robot RB using the communication protocol defined by the network NW under the control of the control unit 1. And, for example, it has an interface for a wired network.

The program storage unit 2 includes, for example, a non-volatile memory such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive) that can be written and read at any time as a storage medium, and a non-volatile memory such as a ROM (Read Only Memory). In addition to middleware such as an OS (Operating System), a program necessary for executing various control processes according to an embodiment of the present invention is stored.

The data storage unit 3 is, for example, a combination of a non-volatile memory such as an HDD or SSD capable of writing and reading at any time and a volatile memory such as a RAM (Random Access Memory) as a storage medium, and is one of the present inventions. A device database (hereinafter, the database is referred to as a DB) 31, a scenario DB 32, a scene DB 33, a dialogue DB 34, and a question / answer DB 35 are provided as main storage units necessary for implementing the embodiment.

The device DB 31 stores device information representing its attributes for each of the sensor devices D1 to Dn. Each device information associates the device type, the installation location, the application period, and the scenario ID with the device ID that identifies the sensor device.

The scenario DB stores a plurality of scenario information for asking a question corresponding to an action scene estimated from the initial action of the target user US. Each scenario information corresponds to the scenario ID with the type and timing of the acquired sensor data, the question content and timing for the target user US, or the question content and timing for a close relative.

The scene DB 33 stores scene information representing the cognitive load and the allocation result of the dialogue group for each of the plurality of action scenes assumed for the target user US. The scene information is associated with the content of the scene, the installation location of the sensor device, the cognitive load, and the dialogue group ID according to the scene ID, for example, for each application period of morning and evening.

The dialogue DB 34 stores data indicating the contents of a plurality of dialogues included in the dialogue group for each dialogue group in association with the dialogue ID. Hereinafter, the data indicating the content of the dialogue is also referred to as the dialogue information.

The question / answer DB 35 includes sensor data indicating the initial behavior of the target user US, question information issued to the target user US or close relatives according to the above scenario information, and the target user US or close relatives responding to the question information. The answer information is stored in association with each other. As the storage form, for example, Resource Description Framework (RDF) representing the relationship between a subject, a predicate, and an object is used.

An example of the information stored in the device DB 31, the scenario DB 32, the scene DB 33, the dialogue DB 34, and the question / answer DB 35 will be described later.

The control unit 1 includes a registration reception processing unit 11, a sensor data acquisition processing unit 12, a scenario selection processing unit 13, a cognitive load / dialogue group determination processing unit 14, and a processing function according to an embodiment of the present invention. It includes a dialogue selection processing unit 15, a question sentence transmission / storage processing unit 16, an answer sentence reception / storage processing unit 17, a sign determination processing unit 18, and a notification information transmission processing unit 19. Each of these processing units 11 to 19 is realized by causing the hardware processor of the control unit 1 to execute the program stored in the program storage unit 2.

The registration acceptance processing unit 11 performs a process of accepting registration of device information and scenario information. For example, the template data for registration is transmitted in response to the registration request from the management terminal MT, and the device information and the scenario information input in the management terminal MT are received according to the registration template data, respectively, and the device DB 31 and the device DB 31 and the scenario information are received. The process of storing in the scenario DB 32 is performed. The registration acceptance processing unit 11 may have a processing function of accepting the registration of the scene information and the dialogue information and storing them in the scene DB 33 and the dialogue DB 34.

The sensor data acquisition processing unit 12 receives the sensor data output from the sensor devices D1 to Dn via the interactive robot RB, and asks / answers the received sensor data as information representing the initial action of the target user US. In addition to being stored in the DB 35, a process of passing the data to the scenario selection processing unit 13 and the cognitive load / dialogue group determination processing unit 14 is performed.

Based on the sensor data received by the sensor data acquisition processing unit 12, the scenario selection processing unit 13 reads the scenario ID associated with the device information of the sensor device that is the source of the sensor data from the device DB 31. Then, the scenario selection processing unit 13 performs a process of reading the corresponding scenario information from the scenario DB 32 based on the read scenario ID.

The cognitive load / dialogue group determination processing unit 14 is in the initial action corresponding to the sensor data received by the sensor data acquisition processing unit 12 when the question partner defined in the selected scenario information is the target user US. Performs a process of determining the cognitive load of the target user US and the dialogue group corresponding to the cognitive load. For example, based on the type of the sensor data and its occurrence timing, the action scene of the target user US is specified from the scene DB 33, and the cognitive load and the dialogue group associated with the specified action scene are read out.

The dialogue selection processing unit 15 performs a process of selecting the corresponding dialogue content, that is, a question sentence from the dialogue DB 34, based on the cognitive load and the dialogue group read from the scene DB 33.

When the question text transmission / storage processing unit 16 receives the question text from the dialogue selection processing unit 15, for example, text data representing the question text is transmitted from the communication I / F4 to the interactive robot RB, and the question text is transmitted. Is stored in the question / answer DB35.

In addition, when the question partner defined in the scenario information selected by the scenario selection processing unit 13 is a close relative, the question text transmission / storage processing unit 16 asks a question for the close relative described in the scenario information. The text data representing the sentence is transmitted from the communication I / F4 to the close relative terminal PT, and the above question sentence is stored in the question / answer DB 35.

When the response sentence reception / storage processing unit 17 receives the voice data of the target user US from the interactive robot RB after transmitting the above question sentence, the response sentence reception / storage processing unit 17 performs voice recognition processing on the received voice data and reproduces the answer sentence. .. Then, the reproduced text data of the answer sentence is associated with the sensor data representing the initial action and the transmitted text data of the question sentence, and a process of storing the text data in the question / answer DB 35 is performed.

Further, when the answer sentence reception / storage processing unit 17 receives the answer information of the close relative from the close relative terminal PT after the transmission of the question sentence, the received answer information is transmitted to the sensor data representing the initial action and the transmission. The process of storing the question / answer DB 35 in association with the text data of the above question sentence is performed.

The sign determination processing unit 18 evaluates the degree of behavioral disorder of the target user US based on each data stored in the above question / answer DB 35, compares the evaluation result with the threshold value, and needs to notify close relatives. Performs a process to determine whether or not.

The notification information transmission processing unit 19 performs a process of transmitting notification information including the estimation result of the dementia sign from the communication I / F4 to the close relative terminal PT when the sign determination processing unit 18 determines that the notification is necessary. conduct.

(Operation example)
Next, an operation example of the server device SV configured as described above will be described together with the operations of the interactive robot RB, the management terminal MT, and the close relative terminal PT.
FIG. 4 is a flowchart showing an example of the processing procedure and the processing content related to the action estimation operation by the server device SV.

(1) Registration of device information and scenario information The registration process of device information and scenario information is performed as follows, for example, according to the operation of the service operator.

That is, when the sensor devices D1 to Dn are installed at a predetermined location in the home of the target user US (including the wrist of the target user US), for example, when the service operator performs a registration request operation on the management terminal MT. , The registration request is transmitted from the management terminal MT to the server device SV.

Upon receiving the registration request, the control unit 1 of the server device SV first transmits the template data for device information registration in step ST10 under the control of the registration reception processing unit 11. Then, when the device information is input in the management terminal MT according to the template data, the registration reception processing unit 11 receives the input device information from the management terminal MT via the communication I / F4 and receives it. The generated device information is associated with the device ID and stored in the device DB 31.

FIG. 6 shows an example of a registration screen using template data displayed on the administrator terminal MT, and FIG. 7 shows an example of registered device information. As shown in FIG. 7, the data registered as the device information includes each data indicating the device type, the installation location, the applicable period indicating the period during which the sensor device is valid, and the scenario ID, and these data. Is stored in a state associated with the device ID. The scenario ID is used for each of the sensor devices D1 to Dn to specify a scenario corresponding to the initial action of the target user US detected by the sensor device.

The registration reception processing unit 11 subsequently transmits template data for scenario information registration. In this state, when the scenario information is input in the management terminal MT according to the template data, the registration reception processing unit 11 receives the scenario information from the management terminal MT via the communication I / F4 and sends it to the scenario DB 32. Remember.

FIG. 8 shows an example of the registered scenario information. As shown in FIG. 8, the data registered as the scenario information shows the type and acquisition timing of the acquired sensor data, the question content and transmission timing to the target user US, or the question content and transmission timing to the close relatives. Data is included, and these data are stored in a state associated with the scenario ID.

On the other hand, the scene information and the dialogue information indicating the question contents are registered in the scene DB 33 and the dialogue DB 34 in advance by, for example, the system administrator.
FIG. 9 shows an example of scene information stored in the scene DB 33. As shown in FIG. 9, the scene information is the initial stage of the target user US when executing the action scene for each of a plurality of action scenes expected to be executed by the target user US in each of the morning and evening time zones. Define the correspondence with the types of actions (1) to (6). Further, the scene information includes cognitive load (high, medium, low) and dialogue groups G1 to G4 assigned to each of the above action scenes, and these data are associated with the scene ID in the scene. It is stored in the DB 33.

FIG. 10 shows an example of dialogue information stored in the dialogue DB 34. As shown in FIG. 10, the serif information is stored for each of the serif groups G1 to G4 in a state in which a plurality of serif contents (question sentences) included in the group are associated with the serif ID. The content of the dialogue consists of, for example, text data.
The process of registering the scene information and the dialogue information in the scene DB 33 and the dialogue DB 34 may be performed by the operation of the service operator under the control of the registration reception processing unit 11.

(2) Behavior estimation of the target user US (2-1) Acquisition of sensor data For example, when the target user US performs an opening / closing operation of the refrigerator door FD in the house, the opening / closing operation is detected by the sensor device D2. Similarly, when the windows and curtains WD are opened and closed, the door of the washing machine is opened and closed, the lid of the medicine box is opened and closed, and the sofa, table or kotatsu is seated or removed, the respective sensors are similarly opened and closed. Detected by the device. Further, even when the target user US goes out or returns home, the behavior is detected by the wearable terminal Dn.

The sensor data output from the sensor devices D1 to Dn is sent to the interactive robot RB via a wireless interface such as wireless LAN or Bluetooth (registered trademark). Then, the sensor data is relayed by the sensor data transmission / reception unit 51 of the interactive robot RB and transmitted to the server device SV. The wearable terminal Dn attached to the target user US among the sensor devices D1 to Dn is stored in a data terminal representing, for example, the number of steps and the conversation time detected while the target user US is out. Then, the stored data representing the number of steps and the conversation time is read out after returning home and transmitted to the interactive robot RB. The interactive robot RB transfers data representing the number of steps and conversation time to the server device SV.

The server device SV receives the sensor data transmitted from the interactive robot RB under the control of the sensor data acquisition processing unit 12 via the communication I / F4 in steps ST11 and ST12, and the received sensor data. Is stored in the question / answer DB35. The sensor data includes identification information (device ID) of the source sensor devices D1 to Dn.

(2-2) Selection of scenario information When the control unit 1 of the server device SV receives the sensor data, the device included in the sensor data is included in the sensor data in step ST13 under the control of the scenario selection processing unit 13. The corresponding device information is read from the device DB 31 based on the ID. For example, if the sensor data is generated from the sensor device D2 attached to the refrigerator door FD, the device information associated with the device ID (D02) is read out. This device information includes a scenario ID (S02) as shown in FIG. 7, for example. Then, the scenario selection processing unit 13 searches the scenario DB 32 based on the scenario ID (S02), and reads out the corresponding scenario information from the scenario DB 32.

Subsequently, in step ST14, the scenario selection processing unit 13 determines whether or not the question is described in the scenario information. Then, if a question is described, in step ST15, the scenario information to be executed first within the applicable period defined in the corresponding device information is selected, and the question partner is asked based on the selected scenario information. It is determined in step ST16 whether is the target user US or a close relative. As a result of this determination, if the question partner is the target user US, the process proceeds to the robot dialogue process according to step ST17.

On the other hand, if the question is not described in the scenario information, the scenario selection processing unit 13 shifts to the sign determination processing in step ST20. Further, if the question is described and the other party is a close relative, the scenario selection processing unit 13 shifts to the close relative dialogue process according to step ST18.

(2-3) Robot Dialogue Processing The control unit 1 of the server device SV executes robot dialogue processing as follows.
FIG. 5 is a flowchart showing a processing procedure and processing contents of the robot dialogue processing. That is, first, under the control of the cognitive load / dialogue group determination processing unit 14, in steps ST171 and ST172, the action type and the application period are determined based on the searched device information and the scenario information, and each determination result is determined. Based on this, the corresponding action scene is selected from the scene DB 33 in step ST173. Then, based on the selected action scene, in step ST174, the cognitive load of the target user US corresponding to the selected action scene and the dialogue group corresponding to the cognitive load are determined.

For example, if the sensor device that has generated the sensor data is the sensor device D2 installed on the door FD of the refrigerator and the time when the sensor data is generated is 7:00 in the morning, it is shown as an action scene, for example, FIG. The scene information associated with the scene ID (S102) is selected as described above. Then, from this scene information, the cognitive load is determined to be "high", and the corresponding dialogue group is determined to be G1.

Subsequently, under the control of the dialogue selection processing unit 15, the control unit 1 of the server device SV corresponds to the dialogue group and changes to the action scene from among a large number of dialogue information stored in the dialogue DB 34 in step ST175. Select the matching dialogue content. For example, as shown in FIG. 9, if the dialogue group determined from the scene information is G1 and the action scene is "preparation for breakfast", the dialogue DB34 shown in FIG. 10 is linked to the dialogue ID (M10). The content of the dialogue "What is the breakfast menu?" Is selected.

Next, the control unit 1 of the server device SV generates the text data representing the selected dialogue content, that is, the question sentence in step ST176 under the control of the question sentence transmission / storage processing unit 16, and generates the generated question. The text data of the sentence is transmitted from the communication I / F4 to the interactive robot RB in step ST177.

When the interactive robot RB receives the text data of the question sentence, the voice synthesis processing unit 52 performs a voice synthesis process based on the text data to generate a voice corresponding to the question sentence. The voice is output to the target user US from, for example, the voice output unit 53 having a speaker. As a result, for example, the interactive robot RB asks the target user US "What is the breakfast menu?" As a question.
The voice synthesis process of the above question sentence may be performed by the server device SV instead of the interactive robot RB.

Similarly, when the sensor device detects that the target user US is sitting on the sofa or kotatsu, the server device SV first selects, for example, a scenario ID (S03) from the device information based on the device ID of the sensor device. do. Subsequently, it is determined whether the question partner is the target user US or a close relative based on the scenario information corresponding to the selected scenario ID. Then, if the question partner is determined to be the target user US, the action scene (S105) of the target user US is determined from the device ID from which the sensor data is generated and the time of occurrence. Then, the cognitive load "medium" and the dialogue group G2 are determined from the determined behavior scene (S105), and one dialogue content is asked from the plurality of dialogue contents M011 to M020 included in the dialogue group G2. And send the data of this question to the interactive robot RB. As a result, the interactive robot RB emits the voice of the above question to the target user US.

Further, when the target user US returns home from going out and the wearable terminal Dn sends the step count data or the data representing the conversation time while going out to the server device SV via the interactive robot RB, the server device SV first describes the above. For example, a scenario ID (S06 or S07) is selected from the device information based on the device ID of the wearable terminal Dn.

The target user US returns home by using, for example, the difference between the position information of the wearable terminal Dn and the position information of the home registered in advance, or wireless communication between the wearable terminal Dn and the interactive robot RB. Can be determined by detecting that is possible. Further, after the interactive robot RB detects the approach of the target user US and identifies the person by face recognition, the interactive robot RB acquires the step count data or the data representing the conversation time from the wearable terminal Dn, and the server device SV. You may want to transfer to.

Subsequently, the server device SV determines whether the question partner is the target user US or a close relative based on the scenario information corresponding to the selected scenario ID (S06 or S07). Then, if the question partner is determined to be the target user US, the action scene of the target user US is determined based on the device ID of the wearable terminal Dn and the time to return home. Then, the cognitive load (for example, "medium") set for the determined action scene and the dialogue group G2 are determined from the scene information, and the dialogue is further selected from the plurality of dialogue contents M011 to M020 included in the dialogue group G2. Select one of the contents as a question and send it to the interactive robot RB.

As a result, the voice of the above question sentence is emitted from the interactive robot RB to the target user US. For example, the question texts are "Where did you go today?" And "Who did you meet today?"

When the transmission of the question information is completed as described above, the question text transmission / storage processing unit 16 indicates the initial action of the target user US that triggered the question by displaying the transmitted data of the question text in step ST19. It is associated with the sensor data and stored in the question / answer DB35.

(2-4) Dialogue processing with close relatives For example, suppose that a medicine box is opened and sensor data indicating the opening and closing of the medicine box is output from a sensor device installed on the lid. In this case, when the server device SV receives the sensor data from the interactive robot RB, the server device SV searches for device information based on the device ID (D05) of the sensor device from which the sensor data is generated, and the scenario ID (S05) is searched from this device information. To judge. Then, based on the scenario information corresponding to this scenario ID (S05), if it is determined in step ST16 that the question partner is a close relative (for example, a care manager), the process proceeds to step ST18.

In step ST18, the control unit 1 of the server device SV first determines whether or not the current time corresponds to the question timing for close relatives described in the above scenario information under the control of the question text transmission / storage processing unit 16. judge. As a result of this determination, for example, if the medicine box has not been opened or closed for the past three days, the question text transmission / storage processing unit 16 reads out the question content described in the scenario information and communicates the text data representing the question content. It is transmitted from the I / F4 to the close relative terminal PT registered in advance in association with the target user US.

As a result, in the close relative terminal PT possessed by the care manager or the like, the question / answer transmission / reception unit 71 receives and displays the text data representing the above question content. For example, "Is the drug prescribed?" Shown in FIG. 8 is displayed. In this case, a question such as "I forgot to take the medicine. Please check with the target user." May be selected, and the question content can be arbitrarily selected.

(2-5) Receiving and storing answer information It is assumed that the interactive robot RB issues the above-mentioned voice question, and the target user US responds by voice. The voice of the answer content is detected by the voice input unit 54 having a speaker in the interactive robot RB, converted into voice data, and then transmitted to the server device SV.

When the control unit 1 of the server device SV receives the voice data of the response content transmitted from the interactive robot RB via the communication I / F4 under the control of the response sentence reception / storage processing unit 17, it is received. Voice recognition processing is performed on the voice data to generate text data of the answer contents. Then, in step ST19, the generated text data representing the answer content is stored in the question / answer DB 35 in association with the text data of the previously stored question.

The interactive robot RB is provided with a voice recognition processing function, and the voice of the response content input by the target user US is voice-recognized by the interactive robot RB to generate text data representing the response content, and the generated response. Text data representing the contents may be transmitted to the server device SV.

Further, when a close relative inputs an answer to the received question content in the close relative terminal PT, the close relative terminal PT uses the question / answer transmission / reception unit 71 to send data representing the input answer content to the server. Send to the device SV. The question information and the answer information are transmitted and received between the close relative terminal PT and the server device SV by mail or SNS (Social Network Service), but voice data communication may also be used.

Under the control of the response reception / storage processing unit 17, the control unit 1 of the server device SV receives and receives the text data representing the response content transmitted from the close relative terminal PT via the communication I / F4. The text data representing the answer content is stored in the question / answer DB 35 in association with the text data of the question stored earlier.

By the way, in the question / answer DB 35, the question information and the answer information are stored in association with the information representing the initial behavior of the target user US detected by the sensor devices D1 to Dn. As the data structure, for example, Resource Description Framework (RDF) representing the relationship between a subject, a predicate, and an object is used.

FIG. 11 shows an example thereof. In this example, "target user US" is the subject, where the device is installed is the predicate, and the device ID is the object. Further, the device ID becomes the subject, the date when the data is acquired becomes the predicate, and the result of estimating the detection time by the device and the usage time of the device in which the device is installed based on the detection time becomes the object. .. Then, the question asked to the target user US or a close relative and the content of the answer are stored as a new object. For example, in the example of FIG. 11, "opening of horse mackerel" which is an answer to a question to the target user US and "with a drug prescription" which is an answer to a question to a close relative are stored.

In this way, by using RDF as the data structure of questions and answers stored in the question / answer DB35, the questions and answers can be flexibly stored even if a new device is added or a new scenario is added. Is possible.

(2-6) Estimating behavior when no question / answer is used Scenario information also defines scenarios in which no question is asked, as shown in FIG. 8, and even if these scenarios are used, the behavior of the target user Can be estimated. For example, the fact that the sensor device is installed on the toilet door is registered in the device information, and when the opening / closing of the toilet door is detected by this sensor device, the scenario S01 that stores the number of times the toilet is used and the usage time is registered. I'll keep it. In this way, the behavior of using the toilet in the home of the target user US can be estimated.

Similarly, a scenario in which the fact that a sensor device is installed on the door of the washing machine is registered in the device information, and when the opening / closing of the door of the washing machine is detected by this sensor device, the number of times the washing machine is used and the usage time are stored. Register S04. In this way, the behavior of the target user US using the washing machine can be estimated.

(3) The control unit 1 of the behavioral disorder determination server device SV, under the control of the sign determination processing unit 18, in step ST20, the initial behavior of the target user US stored in the question / answer DB 35 and the target user US itself. Alternatively, the degree of behavioral disorder of the target user US is evaluated based on the relationship with the answers of close relatives. For example, if the refrigerator door FD is opened but there is no response to cook (or have) breakfast, a negative evaluation value is calculated in terms of breakfast preparation behavior.

As an example, when the refrigerator door FD is opened, the interactive robot RB asks the target user US whether or not to prepare breakfast. Then, when the target user US answers "make" to this question, +5 points are given as an evaluation point, and when the target user US answers "do not make", +3 points are given respectively. Also, if the same question is asked when the operation to open the refrigerator door FD is not performed and the target user US answers "make" to this question, +2 points are given to this question, and "do not make". If you answer, you will be given 0 points.

The sign determination processing unit 18 compares, for example, the total value of the evaluation points of the target user in a certain period with a preset threshold value. Then, as a result of this comparison, when the total value of the evaluation points exceeds the threshold value, it is determined that the degree of behavioral disorder is high.

In step ST21, the notification information transmission processing unit 19 generates notification information to that effect based on the determination result of the evaluation points. Then, the generated notification information is transmitted from the communication I / F4 to the close relative terminal PT possessed by a family member or a care manager who does not live together. As the transmission medium, for example, e-mail or SNS is used.

On the other hand, in the close relative terminal PT, when the notification information is sent from the server device SV, the notification information reception display unit 72 receives and displays the notification information. Therefore, by confirming the notification information regarding the above-mentioned behavioral disorder, a close relative such as a care manager can be aware of the suspicion of a sign of dementia of the target user at an early stage, which can lead to an early consultation. ..

(Action / effect)
As described above, according to one embodiment, device information including the type and installation location of the sensor devices D1 to Dn and scenario information corresponding to the initial action detected by the sensor devices D1 to Dn for the target user are provided in advance. When the initial behavior of the target user is detected by the sensor devices D1 to Dn, the corresponding scenario information is selected, a question is asked to the target user or a close relative according to the scenario information, and the answer is obtained. By doing so, the behavior of the target user is estimated. Therefore, various behaviors of the target user can be appropriately estimated according to the scenario according to the initial behavior of the target user.

In addition, when asking a question to the target user, the cognitive load corresponding to the behavior scene assumed from the initial behavior of the target user is determined, and the behavior scene is selected from the dialogue groups corresponding to the determined cognitive load. The line corresponding to is selected, and the selected line is output as question information. Therefore, an appropriate question is asked for each behavior scene of the target user in consideration of the cognitive load of the target user at that time, and it is possible to obtain a highly valid answer from the target user. Therefore, the behavior of the target user can be estimated more accurately.

[Other Embodiments]
(1) For the morning and evening setting items in the scene information, the time or time zone may be set. In addition, although the cognitive load level is set to two levels of "high" and "medium / small", more levels may be set.

(2) If the answer is not valid or if there is no answer, the number of answers is set in advance (for example, 3 times), or the specified number (for example, half) of the answers obtained during the specified period (for example, 1 week). If it is determined that the above is not appropriate, the level of cognitive load corresponding to the behavior scene is considered to be high for the target user, and the level of cognitive load corresponding to the behavior scene is lowered by one level. Control may be performed to change the setting of the cognitive load level for the action scene. This control can be realized, for example, by storing a collection of valid answers to the question in advance for each question and counting the number of answers not included in this collection of answers.

(3) The dialogue for the cognitive load level may be changed and set in consideration of user attributes such as the hobby of the target user and daily activities. For example, in FIG. 10, in the dialogue M017 included in the dialogue group G2, "Why don't you do brain teaser? Please list three foods with" ki "", the cognitive load level of the target user US is "small / It is set in the question item when "Medium". However, for example, if the hobby of the target user US is flower (growing) and it is assumed that the target user US is familiar with the name of the flower, there is a high possibility that the target user US can easily answer even when the cognitive load level is "high". ..

Therefore, for this target user US, if the above line M017 is included in the line group with a cognitive load of "high", or if, for example, "flowering" is prepared as an action scene of the target user US, the relevant line is applicable. Set the cognitive load level for the behavior scene to "Small / Medium". For example, the system operator obtains a declaration that the target user's hobby is a flower from the target user or a close relative in advance, and sets scene information or dialogue information in consideration of the acquired hobby information. It can be realized by doing.

By setting the content (line) of the question issued from the interactive robot RB in accordance with the attribute information of the target user US at that time, it is possible to ask a more appropriate question to the target user US. It becomes possible to more accurately grasp the behavior of the target user US.

(4) In one embodiment, a case where the device DB 31, the scenario DB 32, the scene DB 33, the dialogue DB 34, and the question / answer DB 35 are provided in the data storage unit 3 of the server device SV has been described as an example. However, the present invention is not limited to this, and each of the DBs 31 to 35 is provided in, for example, another database server so that the server device SV accesses each of the DBs 31 to 35 of the database server via a network. It may be configured.

(5) In one embodiment, the control unit 1 of the server device SV is a registration reception processing unit 11, a sensor data acquisition processing unit 12, a scenario selection processing unit 13, a cognitive load / dialogue group determination processing unit 14, and a dialogue selection processing unit. 15. A case where a program for executing each process of the question sentence transmission / storage processing unit 16, the answer sentence reception / storage processing unit 17, the sign determination processing unit 18, and the notification information transmission processing unit 19 is provided has been described as an example. However, the present invention is not limited to this, and a program for executing the processes of each of the above parts 11 to 19 is provided in another cloud or a server device on the Web, and the above program is downloaded from the server device and used as needed. It may be configured as follows.

(6) The type and installation location of the sensor device may be set in any way according to the initial behavior of the target user to be detected. Further, the scenario information, the scene information, and the dialogue information may be set in any way according to the type of initial action of the assumed target user, the action scene, the user attribute, and the like. In addition, the type and configuration of the interactive robot, the type and configuration of the server device, the processing procedure and processing content thereof, and the like can be variously modified and implemented without departing from the gist of the present invention.

In short, the present invention is not limited to the above-described embodiment as it is, and at the implementation stage, the components can be modified and embodied within a range that does not deviate from the gist thereof. In addition, various inventions can be formed by an appropriate combination of the plurality of components disclosed in the above-described embodiment. For example, some components may be removed from all the components shown in the embodiments. In addition, components from different embodiments may be combined as appropriate.

SV ... Server device US ... Target user RB ... Interactive robot FD, WD ... Monitoring target D1, D2 to Dn ... Sensor device MT ... Management terminal PT ... Relatives terminal NW ... Network 1 ... Control unit 2 ... Program storage unit 3 ... Data storage unit 4 ... Communication I / F
5 ... Bus 11 ... Registration reception processing unit 12 ... Sensor data acquisition processing unit 13 ... Scenario selection processing unit 14 ... Cognitive load / dialogue group judgment processing unit 15 ... Dialogue selection processing unit 16 ... Question text transmission / storage processing unit 17 ... Answer Sentence reception / storage processing unit 18 ... Predictive judgment processing unit 19 ... Notification information transmission processing unit 31 ... Device DB
32 ... Scenario DB
33 ... Scene DB
34 ... Dialogue DB
35 ... Question / Answer DB
51 ... Sensor data transmission / reception unit 52 ... Voice synthesis processing unit 53 ... Voice output unit 54 ... Voice input unit 61 ... Registration setting processing unit 71 ... Question / answer transmission / reception unit 72 ... Notification information reception display unit

Claims (7)

An action estimation device that estimates the behavior of the target user by having the target user present question information by a dialogue device and acquiring the response information of the target user to the question information from the dialogue device.
A detection unit that detects the initial behavior of the target user,
Based on the detected initial action, a determination unit that determines the cognitive load of the target user during the initial action, and a determination unit.
A generation unit that generates the question information for the target user based on the determined cognitive load and the behavior scene estimated from the initial behavior.
A behavior estimation device including an output unit that outputs the generated question information to the dialogue device. The determination unit includes a first storage medium for storing the cognitive load preset corresponding to each of the plurality of initial actions, and from the first storage medium based on the detected initial action. The behavior estimation device according to claim 1, wherein the corresponding cognitive load is selected. When the cognitive load is defined by multiple levels,
The generation unit includes a second storage medium for storing a plurality of the question information prepared in advance corresponding to a combination of the plurality of behavior scenes and the plurality of levels of the cognitive load, and is selected by the determination unit. The behavior estimation device according to claim 2, wherein the corresponding question information is selected from the second storage medium based on the cognitive load level and the behavior scene estimated from the initial behavior. When the cognitive load is defined by multiple levels,
The validity of the answer information of the target user with respect to the question information is determined, and when the determined validity does not satisfy the predetermined determination condition, it corresponds to the initial action stored in the first storage medium. The behavior estimation device according to claim 2, further comprising a changing unit for changing the level of the cognitive load. A third storage medium for storing the detected information representing the initial action, the question information output in response to the initial action, and the answer information of the target user to the question information is provided and stored. The behavior estimation device according to claim 1, further comprising a behavioral disorder determination unit that determines the degree of behavioral disorder of the target user based on the respective information. A behavior estimation method executed by a behavior estimation device that estimates the behavior of the target user by having the target user present question information by the dialogue device and acquiring the response information of the target user to the question information from the dialogue device. There,
The process of detecting the initial behavior of the target user and
Based on the detected initial behavior, the process of determining the cognitive load of the target user during the initial behavior and
A process of generating the question information for the target user based on the determined cognitive load and the behavior scene estimated from the initial behavior, and
A behavior estimation method including a process of outputting the generated question information to the dialogue device. A program for causing a processor included in the behavior estimation device to execute the processing of each part included in the behavior estimation device according to any one of claims 1 to 5.

Images