JP2004195636A - Interaction control device and method and robot device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an entertainment robot capable of performing an interaction specialized in a user. <P>SOLUTION: This interaction control device stores various kinds of information attached to an object as a value of an item corresponding to the object, selects a topic related to a topic used in a conversation just before, and generates an acquiring conversation for acquiring a value of an item in the selected topic or a utilizing conversation utilizing a value of an item in an already stored topic as a next conversation. At this time, a value acquired by the acquiring conversation is stored as a value of a corresponding item. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

  The present invention relates to a dialog control device and method, and a robot device, and is suitably applied to, for example, an entertainment robot.

  In recent years, many entertainment robots for general households have been commercialized. In such an entertainment robot, a task given by the user, for example, a task of "kicking a ball" when there is an instruction of "kicking a ball", when assuming a scene of living with the user at home. In addition to the action to achieve the goal, it is necessary to avoid a one-sided communication by spontaneously performing a variety of actions, and not to get tired of interacting with the entertainment robot every day.

  Most of conventional voice dialogue systems aim at accomplishing a certain task such as video recording reservation and telephone number guidance (for example, see Non-Patent Document 1). In addition, there is a simple response sentence generation system typified by an Eliza as an algorithm for generating a response to a dialogue (for example, see Non-Patent Document 2).

  However, in order to spontaneously perform a variety of actions, it is necessary to provide the entertainment robot with a variety of knowledge and data on various action patterns for various interactions in advance. As a problem, there is a limit to the amount of data that can be provided to an entertainment robot.

  Also, when creating a robot, creating data that is tailored to the individual user of the robot requires a great deal of effort, so that any robot will be given the same knowledge in advance. It is difficult to feel the intimacy of "a robot only." However, robots living with humans in the home are required to be able to perform personalized actions such as being interesting for the user.

  Thus, information such as a user's name, birthday, gender, likes or dislikes can be actively or passively acquired through interaction with the user, and the information can be used to converse with the user. If such a robot can be constructed, it is considered that the satisfaction of the user can be obtained without causing the above-described problem.

Also, at this time, it is possible to share the learning experience by showing the user the process that the robot learns through the conversation with the user, and it is also expected to have the effect of raising the feeling of familiarity with the robot by talking what the robot taught it can.
IPSJ SIG Report, Spoken Language Information Processing 22-8 (July 24, 1998) p. 41-42 Physicality and Computer, Kyoritsu Shuppan, p. 258-268

  However, as an example of a robot having a conversation using information about a user, information associated with a person or an object (object), such as a user's name, birthday, gender, a favorite or a dislike, is actively used. There are a number of problems when considering acquiring them passively or passively, storing them in the form of a memory, and interacting with the memory.

  First, there is a problem of how to acquire a memory specialized for a user. Since the robot has a limited storage capacity and a predetermined framework for storing it, it is not possible to store anything that comes out during a conversation. Further, it is difficult with the current technology to appropriately process an arbitrary utterance from the user at an arbitrary timing and store the utterance in a storage device.

  However, the problem is that the robot speaks to get the memory like "Tell me the name of your friend!" It can be solved by adopting a method of disclosing and having the user tell the value, and by acquiring the memory by acting from the robot in this way, it is possible to encourage the user to gather information easily .

  Second, there is the problem of how to use the acquired memory as speech. If the utterance is made using the acquired memory at random, a topic transition having no context occurs, which gives the user embarrassment. For this reason, a mechanism (association) for using a related one of the memories acquired by the robot for the next utterance is required. Further, if the acquired memory is simply output as it is, only one-to-one correspondence can be made between the utterances of one stored content, so that the variation of the interaction is limited.

  Third, there is the problem of when to acquire memory and use it for conversation. First, in the situation where there is no information attached to the object, an action trying to use memory appears, or an action trying to acquire memory appears in a situation where all information is acquired. It is necessary to avoid the state, and a mechanism for that is necessary. In addition, if a user acted using the memory immediately after acquiring a certain memory, the user would not see the robot as if he / she remembered it with the storage device, and would not look intelligent as the behavior of the robot. However, there is also a problem of interest for that.

  The present invention has been made in view of the above points, and an object of the present invention is to propose a dialogue control device and method, and a robotic device capable of improving entertainment.

  In order to solve such a problem, in the present invention, in the dialogue control device, storage means for storing various information associated with an object as a value of a corresponding item of the object, and the item of the object as a topic are defined. Sometimes, a topic related to the topic used in the immediately preceding conversation is selected, and an acquired conversation for acquiring the value of the item in the selected topic or the value of the item in the topic already stored in the storage means is obtained. Conversation generating means for generating the used conversation as the next conversation is provided, and the conversation generating means stores the value acquired by the acquired conversation as the value of the corresponding item in the storage means.

  As a result, the dialogue control device can have a conversation with the user as the target object.

  Further, in the present invention, in the interaction control method, a first step of storing various information associated with an object as a value of a corresponding item of the object, and defining the item of the object as a topic, Select a topic that is related to the topic used in the previous conversation, and use the acquired conversation to acquire the value of the item in the selected topic or the use conversation that has already been stored using the item value in the topic, And a second step for generating a next conversation. In the second step, a value acquired by the acquired conversation is stored as a value of a corresponding item.

  As a result, according to this interaction control method, a conversation specialized for the user as the target object can be performed with the user.

  Further, in the present invention, in the robot apparatus, a storage means for storing various information associated with the object as a value of a corresponding item of the object, and a dialog for immediately preceding conversation when the item of the object is defined as a topic. A topic related to the topic used in is selected, and an acquisition conversation for acquiring the value of the item in the selected topic or a use conversation using the value of the item in the topic already stored in the storage unit is performed. A conversation generating means for generating the next conversation is provided, and the conversation generating means stores the value acquired by the acquired conversation as the value of the corresponding item in the storage means.

  As a result, the robot apparatus can have a conversation with the user as a target object.

  As described above, according to the present invention, in a dialogue control device, a storage unit that stores various information associated with an object as a value of a corresponding item of the object, and a case where the item of the object is defined as a topic Then, a topic related to the topic used in the immediately preceding conversation is selected, and an acquired conversation for acquiring the value of the item in the selected topic or the value of the item in the topic already stored in the storage means is used. Conversation generating means for generating the used conversation as the next conversation, and the conversation generating means stores the value acquired by the acquired conversation in the storage means as the value of the corresponding item, thereby making the object In this way, it is possible to realize a conversation control device capable of performing a conversation specialized for the user to be performed with the user, and thus improving the entertainment property.

  According to the present invention, in the dialogue control method, a first step of storing various information associated with an object as a value of a corresponding item of the object, and a step of defining the item of the object as a topic , Selecting a topic related to the topic used in the immediately preceding conversation, and acquiring conversation for acquiring the value of the item in the selected topic or using conversation using the item value in the topic already stored. And a second step for generating a next conversation. In the second step, the value acquired by the acquired conversation is stored as the value of the corresponding item, so that the user who is the target Thus, a conversation can be performed with the user, and thus a conversation control method capable of improving the entertainment property can be realized.

  Further, according to the present invention, in the robot apparatus, storage means for storing various information associated with the object as a value of a corresponding item of the object, and when the item of the object is defined as a topic, A topic related to the topic used in the conversation is selected, and an acquired conversation for acquiring the value of the item in the selected topic or a use conversation using the value of the item in the topic already stored in the storage unit is performed. A conversation generating means for generating the next conversation, and the conversation generating means stores the value acquired by the acquired conversation in the storage means as the value of the corresponding item, thereby providing A specialized conversation can be performed with the user, and thus a robot device that can improve entertainment can be realized.

  Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

(1) Configuration of Robot 1 According to the Present Embodiment (1-1) Hardware Configuration of Robot 1 FIG. 1 schematically illustrates a functional configuration of the robot 1 according to the present embodiment. As shown in FIG. 1, the robot 1 includes a control unit 2 that performs overall control of the entire operation and other data processing, an input / output unit 3, a driving unit 4, and a power supply unit 5. .

  The input / output unit 3 is provided as an input unit in a CCD (Charge Coupled Device) camera 10 corresponding to the eyes of the robot 1, a microphone 11 corresponding to an ear, and a part such as a head and a back to detect a user's contact. Touch sensor 12 or other various sensors corresponding to the five senses. The output unit is provided with a speaker 13 corresponding to a mouth, an LED indicator (eye lamp) 14 for forming a facial expression by a combination of blinking and lighting timing. These output units can express the user feedback from the robot 1 in a form other than the mechanical movement pattern by the legs or the like, such as voice or blinking of a lamp.

  The drive unit 4 is a functional block that implements the body operation of the robot 1 according to a predetermined movement pattern instructed by the control unit 2, and is a control target by behavior control. The drive unit 4 is a functional module for realizing a degree of freedom at each joint of the robot 1, and includes a plurality of drive units 151 to 15n provided for each axis such as roll, pitch, and yaw at each joint. You. Each of the drive units 151 to 15n includes a motor 161 to 16n for performing a rotation operation around a predetermined axis, an encoder 171 to 17n for detecting a rotational position of the motor 161 to 16n, and a motor 161, based on an output of the encoder 171 to 17n. It is composed of a combination of drivers 181 to 18n that adaptively control the rotation position and rotation speed of 16n.

  Depending on how the drive units are combined, the robot 1 can be configured as a legged mobile robot, for example, for bipedal walking or quadrupedal walking.

  The power supply unit 5 is a functional module that supplies power to each electric circuit and the like in the robot 1 as the name implies. The robot 1 according to the present embodiment is of an autonomous driving type using a battery, and the power supply unit 5 includes a charge battery 19 and a charge / discharge control unit 31 that manages a charge / discharge state of the charge battery 20.

  The charging battery 19 is configured, for example, in the form of a “battery pack” in which a plurality of lithium ion secondary battery cells are packaged in a cartridge type.

  Further, the charge / discharge control unit 20 grasps the remaining capacity of the battery 19 by measuring the terminal voltage and the amount of charge / discharge current of the battery 19, the ambient temperature of the battery 19, and determines the start time and end time of charging. decide. The start and end timings of charging determined by the charge / discharge control unit 20 are notified to the control unit 2 and serve as a trigger for the robot 1 to start and end the charging operation.

  The control unit 2 corresponds to a “brain”, and is mounted on, for example, the head or torso of the robot 1.

  As shown in FIG. 2, the control unit 2 has a configuration in which a CPU (Central Processing Unit) 21 as a main controller is bus-connected to a memory, other circuit components, and peripheral devices. The bus 27 is a common signal transmission path including a data bus, an address bus, a control bus, and the like. Each device on the bus 27 is assigned a unique address (memory address or I / O address). The CPU 21 can communicate with a specific device on the bus 27 by specifying an address.

  A RAM (Read Access Memory) 22 is a writable memory composed of a volatile memory such as a DRAM (Dynamic RAM), and loads a program code to be executed by the CPU 21 and temporarily stores work data by the execution program. Used for preservation.

  The ROM (Read Only Memory) 23 is a read-only memory that stores programs and data permanently. The program code stored in the ROM 23 includes a self-diagnosis test program executed when the power of the robot 1 is turned on, a control program for defining the operation of the robot 1, and the like.

  The control program of the robot 1 includes a “sensor input / recognition processing program” that processes sensor inputs such as the CCD camera 10 and the microphone 11 and recognizes the symbols as symbols. A “behavior control program” that controls the behavior of the robot 1 based on a predetermined behavior control model, a “drive control program” that controls the drive of each joint motor, the audio output of the speaker 12, and the like according to the behavior control model are included. .

The non-volatile memory 24 is, for example, an EEPROM (Electrically
It is composed of a memory element that can be electrically erased and rewritten, such as an Erasable and Programmable ROM, and is used to hold data to be sequentially updated in a nonvolatile manner. The data to be sequentially updated includes a memory key and other security information, a device control program to be installed after shipment, and the like.

  The interface 25 is a device for interconnecting with devices outside the control unit 2 and enabling data exchange. The interface 25 performs data input / output with the camera 10, the microphone 11, and the speaker 12 in the input / output unit 3, for example. The interface 25 inputs and outputs data and commands to and from each of the drivers 181 to 18n in the drive unit 4.

Further, the interface 25 is provided with an RS (Recommended
Standard) -232C, serial interface such as IEEE (Institute of Electrical and Electronics Engineers) 1284, USB (Universal Serial Bus) interface, i-Link (IEEE1394) interface, SCSI (Small)
It has a general-purpose interface for connecting computer peripheral devices, such as a computer system interface), a memory card interface (card slot) that accepts PC cards and memory sticks, etc., and can be connected to external devices that are locally connected. A program or data may be moved.

  As another example of the interface 25, an infrared communication (IrDA) interface may be provided to perform wireless communication with an external device.

  Further, the control unit 2 includes a wireless communication interface 26, a network interface card (NIC) 28, and the like, and performs near field wireless data communication such as Bluetooth, a wireless network such as IEEE802.11b, or a high frequency band such as the Internet. Data communication can be performed with various external host computers via the network.

  By such data communication between the robot 1 and the host computer, complicated operation control of the robot 1 can be calculated or remotely controlled using a remote computer resource.

(1-2) Software Configuration of Robot 1 FIG. 3 schematically illustrates a functional configuration of the behavior control system 30 of the robot 1 configured by a control program group stored in the ROM 23. The robot 1 can perform behavior control according to the recognition result of the external stimulus and a change in the internal state. Furthermore, by providing a long-term memory function and associatively storing the change in the internal state from the external stimulus, the behavior control can be performed according to the recognition result of the external stimulus and the change in the internal state.

  This behavior control system 30 is implemented using object-oriented programming. In this case, each software is handled in a module unit called an “object” that integrates data and a processing procedure for the data. Further, each object can perform data transfer and Invoke by a message communication and an inter-object communication method using a shared memory.

  The behavior control system 30 includes a visual recognition function unit 31, a hearing recognition function unit 32, and a visual recognition function unit 31 for recognizing an external environment based on sensor outputs of the CCD camera 10, the microphone 11, and the touch sensor 13 in the input / output unit 3. The contact recognition unit 33 is provided.

  The visual recognition function unit 31 executes image recognition processing such as face recognition and color recognition and feature extraction based on an image signal formed by the sensor output of the CCD camera 10. Then, the visual recognition function unit 31 obtains information such as a face ID (identifier) unique to the person as the face recognition result, the position and size of the face image area, and the position and size of the color area as the color recognition result Then, information such as the feature amount is output.

  The auditory recognition function unit 32 performs recognition processing on various sounds, such as voice recognition and speaker recognition, based on a voice signal that is a sensor output of the microphone 11. Then, the auditory recognition function unit 32 converts the character string information of the recognized word, which is the result of the speech recognition, and the speaker ID information unique to the speaker, which is the result of the speaker recognition processing based on the acoustic features and the like. Output.

  The contact recognition function unit 33 recognizes an external stimulus such as “patched” or “hit” based on a pressure detection signal that is a sensor output of the touch sensor 13 and outputs a recognition result.

  The internal state management unit 34 manages several types of emotions, such as instinct and emotion, by mathematical modeling and manages the emotions by the visual recognition function unit 31, the auditory recognition function unit 32, and the contact recognition function unit 33. The internal state such as the instinct and emotion of the robot 1 is managed accordingly.

  On the other hand, the behavior control system 30 compares information with a short-term storage unit 35 that performs short-term storage lost over time in order to perform behavior control according to the recognition result of an external stimulus or a change in an internal state. A long-term storage unit 36 is provided for long-term storage. The classification of short-term memory and long-term memory depends on neuropsychology.

  The short-term storage unit 35 is a functional module that holds a target or an event recognized by the visual recognition function unit 31, the auditory recognition function unit 32, and the contact recognition function unit 33 for a short time. For example, the input image from the CCD camera 10 is stored for a short period of about 15 seconds.

  The long-term storage unit 36 is used to hold information obtained by learning, such as the name of an object, for a long time, and the RAM 22 and the nonvolatile memory 24 in the control unit 2 (FIG. 2) are used. .

  Furthermore, the behavior of the robot 1 generated by the behavior control system 30 includes “reflex behavior” realized by the reflex behavior unit 39, “situation-dependent behavior” realized by the context-dependent behavior hierarchy 38, and It is broadly divided into "consideration actions" that are realized.

  The reflexive action unit 39 is a functional module that implements a reflexive body operation in response to an external stimulus recognized by the visual recognition function 31, the auditory recognition function unit 32, and the positive recognition function unit 33.

  The reflex action is basically an action of directly receiving a recognition result of external information input from a sensor, classifying the recognition result, and directly determining an output action. For example, it is preferable to implement a behavior such as chasing or nodding a human face as a reflex action.

  The situation-dependent behavior hierarchy 38 performs an action immediately corresponding to the situation where the robot 1 is currently placed, based on the storage contents of the short-term storage unit 35 and the long-term storage unit 36 and the internal state managed by the internal state management unit 34. Control.

  The situation-dependent action hierarchy 38 prepares a state machine for each action, classifies the recognition result of the external information input from the sensor depending on the action or situation before that, and expresses the action on the aircraft. I do. The context-dependent behavior hierarchy 38 also implements an action for keeping the internal state within a certain range (also referred to as “homeostasis action”). When the internal state exceeds a specified range, the internal state is regarded as the relevant state. The action is activated so that the action for returning to the range easily appears (actually, the action is selected in consideration of both the internal state and the external environment). Situation-dependent behavior has a slower reaction time than reflex behavior.

  The reflection behavior hierarchy 37 performs a relatively long-term action plan of the robot 1 based on the storage contents of the short-term storage unit 35 and the long-term storage unit 36.

  Reflection behavior is behavior that is performed based on a given situation or a command from a human, based on reasoning and planning to realize it. For example, searching for a route from the position of the robot and the position of the target is equivalent to deliberate behavior. Such inferences and plans may require more processing time and computational load (that is, more processing time) than reaction time for the robot 1 to maintain interaction, and such reflexive behavior and situation-dependent behavior may be performed in real time. While responding, the contemplation conducts reasoning and planning.

  The reflection behavior hierarchy 37, the situation-dependent behavior hierarchy 38, and the reflex behavior unit 39 can be described as a higher-level application program independent of the hardware configuration of the robot 1. In response to this, the hardware-dependent layer control unit 40 outputs sound through the speaker 12 of the input / output unit 3, blinks the LED 14 in a predetermined pattern, The corresponding drive units 151 to 15n of the unit 4 are driven.

(1-3) Configuration of Internal State Management Unit 34 Here, the configuration of the internal state management unit 34, which is directly related to a later-described interaction control function, among the components of the behavior control system 30 will be described.

  The internal state management unit 34 manages the instinct and emotion modeled as described above, and the state of the instinct and emotion is determined by the visual recognition function unit 31, the auditory recognition function unit 32, and the contact recognition function unit 33. It changes according to the recognized external stimulus.

  In this case, instinct elements that constitute such instinct include fatigue, heat or temperature, pain, appetite or hunger, thirst, affection, and curiosity. There are nine instinct elements such as heart, elimination, and sexuality, and the emotional elements that make up such feelings are happiness, sadness, anger, and anger. Surprise, disgust, fear, frustration, boredom, sleep (somnolence), sociability (gregariousness), patience, tension, relaxed There are 18 emotional elements such as, alertness, guilt, spite, loyalty, submission, and jealousy.

  The internal state management unit 34 holds the instinct element and the emotion element as parameters representing the strength of the element, and stores the parameter value of each element in the visual recognition function unit 31 and the auditory recognition function. The instinct and emotion of the robot 1 are changed every moment by periodically updating based on the recognition result of the unit 32 and the contact recognition function unit 33, the elapsed time, and the like.

  Specifically, the internal state management unit 34 determines a predetermined arithmetic expression for each instinct element based on the recognition results of the visual recognition function unit 31, the auditory recognition function unit 32, and the contact recognition function unit 33, the elapsed time, and the like. The variation amount of the instinct element at the time calculated by ΔI [k], the current parameter value of the instinct element is I [k], and the coefficient representing the sensitivity of the instinct element is ki And the following equation

Is used to calculate the parameter value I [k + 1] of the instinct element in the next cycle, and replace the result of this operation with the parameter value I [k] of the present instinct element so that the parameter of the instinct element Update the value.

  The internal state management unit 34 also determines, for each emotional element, the recognition results of the visual recognition function unit 31, the auditory recognition function unit 32, and the contact recognition function unit 33, the behavior of the robot 1 at that time, ΔE [t], the current parameter value of the emotion element is E [t], which is calculated by a predetermined arithmetic expression based on the elapsed time of the emotion element and the like. Let ke be a coefficient representing the sensitivity of

Is used to calculate the parameter value E [t + 1] of the emotional element in the next cycle, and replace this with the current parameter value of the emotional element to update the parameter value of the emotional element.

  Note that how the recognition results of the visual recognition function unit 31, the auditory recognition function unit 32, and the contact recognition function unit 33 affect each instinct element and each emotion element is determined in advance. The recognition result of “stroke” by the recognition function unit 33 is based on the variation ΔI [k] of the parameter value of “affection” of the instinct element and the variation ΔE [k] of the parameter value of “happy” of the emotional element. t].

(2) Dialogue Control Function in Robot 1 (2-1) Dialogue Control Function in Robot 1 Next, a dialogue control function mounted on the robot 1 will be described.

  The robot 1 has a predetermined number of items such as the name, birthday, and favorite items of the user and other objects (hereinafter collectively referred to as a user) through dialogue with the user. Information (hereinafter referred to as the value of the item) is obtained and stored in, for example, a predetermined format shown in FIG. 4, and the user interacts with the user while using the stored value of each item. It is equipped with a dialogue control function to obtain.

  In FIG. 4, columns include predetermined items (“Name”, “Kind”, “Face ID (FaceID)”, “Speaker ID (SpeakerID)”, “Birthday” (BIRTHDAY) "," Favorite "and" Friend ") (" Yukiko "," Human "," 0 "," 1 "," 73/5/2 "," Tea " , "Kazumi" etc.), and the row indicates the value of each item obtained for one object. The first number in each row indicates the ID (object ID) of the object given when the object is first registered.

  FIG. 4 shows a state after the values of all items for each object have already been acquired. Among the items, “Name”, “Speaker ID”, “Birth” "BIRTHDAY", "Favorite" and "Friend" are obtained by various voice recognition processes of the hearing recognition function unit 32 at the time of dialogue with the user, and the "face ID (FaceID)" is obtained. "And" Kind "are obtained and stored by various image recognition processes of the visual recognition function unit 31 at the time of dialogue with the user, and are stored.

  In FIG. 4, the numerical value described to the left of each value represents the degree of impression for that value. This impression level serves as a measure of whether or not the robot 1 may use the topic for conversation in the future. For example, in the example of FIG. No), the lower the value, the worse the impression (not wanting to talk).

  In the present embodiment, the impression degree is determined by the parameter value of “affection” in the internal state management unit 34 immediately before acquiring the value of the item and the parameter value of “affection” immediately after acquiring the value of the item. And the difference is given.

  Here, the interaction control function is mainly performed by the processing of the situation-dependent behavior hierarchy 38 in the behavior control system 30 described above with reference to FIG. Then, when the processing contents of the context-dependent behavior hierarchy 38 relating to the dialog control function are functionally classified, as shown in FIG. 5, a conversation for acquiring the value of each item for the user or the like (hereinafter referred to as an acquired conversation) A storage-based conversation generation unit 50 that generates a conversation (hereinafter, referred to as a usage conversation) using the acquired values of the respective items for the user and the like; It can be divided into a situation determination unit 52 that controls the storage acquisition conversation generation unit 50 and the storage use conversation generation unit 51.

  In this case, based on the recognition result of the visual recognition function unit 31 and the recognition result of the auditory recognition function unit 32 obtained via the short-term storage unit 35 (FIG. 3), the situation determination unit 52 determines the presence of a user who can be a conversation partner. As shown in FIG. 6, at this time, the user obtains the face ID and the speaker ID of the user obtained as the recognition results of the visual recognition function unit 31 and the auditory recognition function unit 32, and obtains the conversation ID. In addition to sending the generated conversation to the generation unit 51, the storage acquisition conversation generation unit 50 and the storage use conversation generation unit 51 are inquired as to whether an acquisition conversation or a use conversation can be generated (step SP1A, step SP1B).

  At this time, the memory acquisition conversation generation unit 50 and the memory use conversation generation unit 51 generate a topic and how to use the generated topic in accordance with a predetermined topic generation rule and topic usage method determination rule. A conversation generation process for determining a topic usage method of generating an acquired conversation or a use conversation is executed (step SP2A, step SP2B).

  Here, in the case of the present embodiment, there are six rules as such topic generation rules.

  The first topic generation rule is a rule for selecting another arbitrary item of the same object as the topic used in the immediately preceding conversation as the next topic. In the present embodiment, the first topic generation rule starts with the item having the highest impression degree. Select in order. For example, in FIG. 4, when the topic of the immediately preceding conversation is “BIRTHDAY” of “Object ID1”, the item having the highest impression degree among the other items of the same “Object ID1” is next. This is the case for the topic of conversation.

  The second topic generation rule is a rule for selecting, as the next topic, an item that is related to the same object as the topic used in the immediately preceding conversation. For example, in FIG. , "Favorite" in "", and "Dislike" in "Object ID1" is the topic of the next conversation.

  Further, the third topic generation rule is a rule for selecting another item of the object that can be specified from the value of the topic item used in the immediately preceding conversation as the next topic. In the present embodiment, The next object is set in order from the object having the highest impression degree. For example, in FIG. 4, when the topic of the immediately preceding conversation is “Friend” of “Object ID1”, one of the items of “Object ID2” that is “Friend” is changed to the following. This is the case for conversation topics.

  On the other hand, the fourth topic generation rule is a rule for selecting the same item of the same object as the topic used in the immediately preceding conversation as the next topic. For example, in FIG. 4, the topic of the immediately preceding conversation is “object ID 1”. Is "Birthday", the case where the same "Birthday" of the same "Object ID1" is set as the topic of the next conversation corresponds to the case.

  The fifth topic generation rule is a rule for selecting the same item of another object having the same value as the value of the item in the topic used in the immediately preceding conversation as the next topic. For example, in FIG. Is "Favorite" of "Object ID1", and "Favorite" of "Object ID3" with the same "Favorite" value of "Tea" (Favorite)] is the topic of the next conversation.

  Further, the sixth topic generation rule is a rule for selecting an item of another object having a value related to the above-mentioned value of the item in the topic used in the immediately preceding conversation as the next topic. If the conversation topic is “Favorite” of “Object ID1”, “horse racing” is changed from “Favorite” that is “Favorite” to “Favorite”. )) Is selected, and "Favorite" of the "object ID 6" is set as the topic of the next conversation.

  Further, the seventh topic generation rule is a rule for selecting another object as the next topic with the same item as the topic used in the immediately preceding conversation. For example, in FIG. 4, the topic of the immediately preceding conversation is “object ID 1”. , "Favorite" of "object ID2" is the topic of the next conversation.

  Note that among these first to seventh topic generation rules, the first to third topic generation rules can be used both when generating the acquired conversation and when generating the use conversation. Topic generation rules cannot be used when generating acquired conversations, but are effective when generating used conversations.

  Therefore, the memory-acquired conversation generation unit 50 randomly (arbitrarily) selects one of the first to third topic generation rules, and the memory-based conversation generation unit 51 selects one of the first to seventh topic generation rules. One of the generation rules is randomly selected, and according to the selected topic generation rule, for example, when the topic of the immediately preceding conversation is “Favorite object ID 1 (Favorite)”, "Friend of ID1 (Friend)", "Dislike of object ID1 (Dislike)", "Birthday of object ID2 (BIRTHDAY)", "Birthday of object ID1 (BIRTHDAY)", "Object ID3 The topic of the next conversation is generated in the form of "Favorite" or "Favorite of the object ID6".

  On the other hand, in the present embodiment, there are three rules as topic usage method determination rules.

  In this case, the first topic use method determination rule is a rule that generates an utterance using the value of the corresponding item stored in the long-term storage unit 36 as it is. When "Favorite" is generated, "Yukiko's favorite thing is tea. And "What is Yukiko's favorite thing?" ] Is generated.

  The second topic use method determination rule is a rule for searching the database from the value of the corresponding item stored in the long-term storage unit 36, reading out related items, and using this to generate an utterance, For example, if "Birthday of object ID1" is generated as a topic, the anniversary database is searched using "73/5/2", which is the value of "Birthday", as a keyword. Using the obtained "Travel Ad Day", "May 2 is a traffic ad day. ] Is generated.

  Further, the third topic use method determination rule reads the database with a related value calculated or associated from the value of the corresponding item stored in the long-term storage unit 36, reads out a related item, and reads this. This rule is used to generate an utterance. For example, when “birthday of object ID1 (BIRTHDAY)” is generated as a topic, the value of “birthday (BIRTHDAY)” is “73/5/2”. Searching the horoscope database using "Taurus" as a keyword calculated from "" and using the obtained "Patience" item, "Taurus is a patient. ] Is generated.

  Note that, of these first to third topic use method determination rules, the first topic use method determination rule can be used both at the time of generation of the acquired conversation and at the time of generation of the use conversation. Although the topic usage method determination rule cannot be used when generating an acquired conversation, it is effective when generating a used conversation.

  Therefore, the memory acquisition conversation generation unit 50 selects the first topic use method determination rule, and the memory use conversation generation unit 51 randomly selects one from the first to third topic use method determination rules. Then, an utterance for acquiring information or an utterance using the acquired information is generated according to the selected topic use method determination rule.

  When the memory generation conversation generation unit 50 and the memory use conversation generation unit 51 can generate a topic and determine a method of using the topic by such a topic generation process, the fact and the topic and the method of use are sent to the situation determination unit 52. On the other hand, if such a topic cannot be generated, the status determination unit 52 is notified of the fact (step SP3A, step SP3B).

  On the other hand, when both of the memory acquisition conversation generation unit 50 and the memory use conversation generation unit 51 are notified that the topic has been generated and the method of using the topic has been determined, the situation determination unit 52 performs the dialogue at that time. A first degree, which is the degree of the number of items that have not yet obtained a value with respect to the total number of items of the partner, and a second degree, which is the degree of the number of items that have already obtained a value with respect to the total number of items of the dialogue partner. Based on this, one of the memory acquisition conversation generation unit 50 and the memory use conversation generation unit 51 is selected (step SP4).

  More specifically, the situation determination unit 52 selects the memory acquisition conversation generation unit 50 when the first degree is equal to or higher than the second degree, and when the first degree is smaller than the second degree, the storage use conversation generation unit 50 is selected. The generation unit 51 is selected.

  Then, when the situation determination unit 52 selects one of them in this manner, the selected storage acquisition conversation generation unit 50 or the storage use conversation generation unit 51 receives the acquisition conversation or the storage use conversation generation unit 51 based on the topic and the use method. An instruction is given to start a process of generating a use conversation (hereinafter, this process is referred to as a conversation generation process) (step SP5).

  Also, when a notification of a topic and its use method is given from only one of the memory acquisition conversation generation unit 50 and the storage use conversation generation unit 51, the situation determination unit 52 or the storage acquisition conversation generation unit 50 or the storage The use conversation generation unit 51 is instructed to start a conversation generation process based on the topic and the method of use (step SP5).

  Thus, the memory acquisition conversation generation unit 50 or the storage use conversation generation unit 51 to which the instruction is given starts the conversation generation process, and first accesses the internal state management unit 34 to access the internal state management unit 34 for “love” of the instinct element. A parameter value is acquired, and thereafter, an utterance for acquiring information (hereinafter, referred to as an acquired utterance) or the acquired information is generated based on the topic determined as described above and the method of using the topic. The character string data D1 of the contents of each utterance for performing an acquired conversation or a series of utterances including utterances (hereinafter, referred to as usage utterances) is sent to the speech synthesis unit 53 in the hardware-dependent behavior control unit 40. Send out sequentially.

  As a result, an audio signal S1 is generated in the audio synthesizing section 53 based on the character string data D1, and is supplied to the speaker 12 (FIG. 1), so that, for example, "What is Yukiko's favorite thing?" ), And form a usage conversation consisting of a series of utterances that include a series of utterances that form a series of utterances that include the utterance utterances such as "Ukiko's favorite thing is tea!" The voice of each uttered content is output from the speaker 12 (step SP6).

  At this time, the user's response to the topic is collected by the microphone 11 and given to the auditory recognition function unit 32 of the behavior control system 30 (FIG. 3), and the voice is recognized by the auditory recognition function unit 32.

  Thus, when the memory acquisition conversation generation unit 50 is selected in step SP4, the memory acquisition conversation generation unit 50 responds to the acquired utterance by the user based on the audio recognition result of the auditory recognition function unit 32. Then, the answer to the question based on the acquired utterance (that is, the value of the item that was about to be acquired at that time) is extracted and stored in the long-term storage unit 36 in the format described above with reference to FIG. 4 (step SP6).

  At the same time, the memory acquisition conversation generation unit 50 accesses the internal state management unit 34 to acquire the parameter value of “affection” at this time, and the acquired parameter value and the same “affection” acquired immediately before the conversation. Is calculated, and the result of the calculation is stored in the long-term storage unit 34 in association with the calculated value as an impression degree of the value of the item obtained by the user or the like as described above (step SP6). .

  On the other hand, when the acquired conversation or the use conversation ends, the storage acquisition conversation generation unit 50 or the storage use conversation generation unit 51 sends a notification notifying this to the situation determination unit 52 (step SP7). Then, upon receiving the notification, the situation determination unit 52 stores the topic notified in step SP3A or step SP3B from the storage acquisition conversation generation unit 50 or the storage use conversation generation unit 51 selected in step SP4 and the method of using the topic. The notification is sent to the generation unit 50 and the storage use conversation generation unit 51 (step SP8A, step SP8B).

  Thus, the memory acquisition conversation generation unit 50 and the memory use conversation generation unit 51 store the notified topic and its usage as a conversation history (step SP9A, step SP9B), and thereafter perform a single conversation with the same conversation partner. In step (1), acquired conversations or use conversations are sequentially generated in the same manner as described above until the conversation is completed so as not to use the same topic (step SP10A-step SP1A-step SP10A, step SP10B-step SP1B-step SP10B).

  In this way, in this robot, while sequentially acquiring various kinds of information about the user in a natural manner and utilizing the information, while sequentially transiting the topics in a natural flow, the conversation of the topic specialized for the user is performed. Has been made possible.

(2-2) Specific Processing Contents of Each Conversation Generation Unit at the Time of Topic Generation Next, specific processing contents of the memory acquisition conversation generation unit 50 and the memory use conversation generation unit 51 at the time of topic generation will be described.

  The storage acquisition conversation generation unit 50 and the storage use conversation generation unit 51 determine the generation of a topic and the method of using the topic in accordance with the topic generation processing procedure RT1 shown in FIG.

  That is, the memory acquisition conversation generation unit 50 and the storage use conversation generation unit 51, when inquiring from the situation determination unit 52 in step SP1A and step SP1B of FIG. The generation processing procedure RT1 is started in step SP20, and in the following step SP21, one topic generation rule is randomly selected from the above-described first to third or first to sixth topic generation rules.

  Next, the memory acquisition conversation generation unit 50 and the storage use conversation generation unit 51 proceed to step SP22, and generate a topic according to the topic use rule selected in step SP1. Thus, for example, when the topic of the immediately preceding conversation is “Favorite of the object ID 1”, “Friend of the object ID 1 (Friend)” and “Dislike of the object ID 1 (Dislike) )] ”And other topics are generated.

  Subsequently, the storage acquisition conversation generation unit 50 and the storage use conversation generation unit 51 proceed to step SP23, and the data of the storage area associated with the “Name” column of the object ID in the long-term storage unit 36 And the data of the storage area associated with the column of the item corresponding to the object ID. For example, if the topic selected in step SP21 is “Friend of the object ID1”, the long-term storage unit associated with the “Name” column of the “object ID1” in FIG. The data in the storage area in the long-term storage unit 36 associated with the “Friend” column of the “object ID1” is read out.

  However, in this case, the value of the “Name” of the object ID and the value of the corresponding item of the object ID have not been obtained yet, and the data is stored in the read storage area. Not always.

  Therefore, in the succeeding step SP24, the storage acquisition conversation generation unit 50 and the storage use conversation generation unit 51 determine whether or not the acquisition conversation or the use conversation can be generated based on each data read in step SP23, that is, a desired item. It is determined whether or not the value of has been acquired.

  When the memory acquisition conversation generation unit 50 and the storage use conversation generation unit 51 obtain a negative result in step SP24, they return to step SP21, and thereafter repeat steps SP21 to SP24 until a positive result is obtained in step SP24.

  On the other hand, if the storage acquisition conversation generation unit 50 and the storage use conversation generation unit 51 obtain a positive result in step SP24, the process proceeds to step SP25, where the storage acquisition conversation generation unit 50 sets the topic use method determination rule as The above-described first topic use method determination rule is selected, and the storage use conversation generation unit 51 randomly selects one topic use method determination rule from the above-described first to third topic use method determination rules. .

  Further, the memory acquisition conversation generation unit 50 and the storage use conversation generation unit 51 thereafter proceed to step SP26, check the conversation history stored at that time, and perform a series of conversations with the user up to that point. Then, it is determined whether or not the conversation of the same combination as the combination of the topic selected in step SP21 and step SP25 in this time has already been performed.

  When the storage acquisition conversation generation unit 50 and the storage use conversation generation unit 51 obtain a negative result in step SP26, the process proceeds to step SP27, ends the topic generation processing procedure RT1, and thereafter, the situation determination unit 52 Notification that the topic has been generated and the method of using the topic has been determined (steps SP3A and SP3B in FIG. 6).

  On the other hand, if the storage acquisition conversation generation unit 50 and the storage use conversation generation unit 51 obtain a positive result in step SP26, they return to step SP25, select a new topic use method determination rule, and thereafter, On the other hand, the loop of steps SP25-SP26-SP25 is repeated until all topic use method determination rules are combined or a positive result is obtained in step SP26.

  Then, the memory acquisition conversation generation unit 50 and the storage use conversation generation unit 51 complete the combination of all topic use method determination rules for the topic by this processing, but still do not obtain a negative result in step SP27. Accordingly, when it is confirmed that there is no selectable topic use method determination rule, the process returns to step SP21 to generate a new topic, and thereafter, steps SP2 and thereafter are processed in the same manner as described above.

  Furthermore, the memory acquisition conversation generation unit 50 and the storage use conversation generation unit 51 cannot generate a conversation even by a combination of all topics and all use methods by repeatedly performing the processing from step SP21 to step SP26 (step SP27). Is reached, the process proceeds to step SP28, the topic generation processing procedure RT1 is terminated, and after that, the situation determination unit 52 is informed that generation of a topic and determination of a method of using the topic could not be performed. Notify (step SP3A, step SP3B in FIG. 6).

  In this way, the memory acquisition conversation generation unit 50 and the memory use conversation generation unit 51 generate topics and determine how to use them.

(2-3) Specific Processing of Each Conversation Generating Unit During Conversation Generation Processing Next, specific processing contents of the storage acquisition conversation generation unit 50 and the storage use conversation generation unit 51 during the conversation generation processing will be described.

  The memory acquisition conversation generation unit 50 and the memory use conversation generation unit 51 correspond to each item described above with reference to FIG. 4, respectively, as shown in FIG. A template for generating a use utterance using the value of the item is provided in advance.

  At the time of the conversation generation processing, the memory acquisition conversation generation unit 50 and the memory use conversation generation unit 51 add the “Name” of the object that is the topic at that time to the corresponding part of the template of the corresponding item. A character string data D1 of an acquired utterance for acquiring the value of the item or a use utterance for using the value of the item by applying the value or the value of “Favorite” is applied. It has been made.

  In addition, the memory acquisition conversation generation unit 50 and the memory use conversation generation unit 51 perform an acquisition conversation consisting of a series of utterances including an acquisition utterance or a use utterance based on the topic determined in step SP2A or step SP2B in FIG. In order to generate a use conversation, a state transition model in which what utterance and processing are to be performed and at what timing is used for each method of use, that is, for each of the above-described first to third topic use method determination rules. are doing.

  In this state transition model, for example, as shown in FIG. 9, each state called a node (Node1 to Node6) is connected by a path called an arc (Ac1 to Ac6), The processing contents to be performed at that time are associated with each other.

  Then, at the time of the conversation generation process, that is, at the time of generation of the acquired conversation or the used conversation, the storage acquisition conversation generation unit 50 and the storage use conversation generation unit 51 sequentially transition the states (nodes Node1 to Node6) according to the corresponding state transition model. Meanwhile, by transmitting the character string data D1 of the utterances respectively associated with the paths (arcs Ar1 to Ar7) between the states that pass at that time to the sound synthesizing unit 53 (FIG. 5), the utterance sound of a predetermined content is output. The signals are sequentially output from the speakers 11 in a predetermined order.

  At this time, the acquired utterance and the use utterance are also associated with one of the paths in each state transition model, and the stored acquisition conversation generation unit 50 and the storage use conversation generation unit 51 perform the processing as described above when passing through the path. By generating the acquired utterance or the use utterance, and transmitting the character string data D1 to the voice synthesizing unit 53, the voice of the obtained utterance or the use utterance is output from the speaker 11.

  In this way, the memory acquisition conversation generation unit 50 and the memory use conversation generation unit 51 are composed of a series of utterances including the acquisition utterance or the use utterance according to the topic determined in step SP2A and step SP2B in FIG. An acquisition conversation and a use conversation can be generated.

  8 and 9 show a specific example of a conversation (acquired conversation) with a user for acquiring a value of a certain item under the control of the memory acquisition conversation generating unit 50. In this specific example, it is assumed that the topic is “Favorite of the object ID 1 (Favorite)” and the first topic use method determination rule is selected as the use method.

  In this case, the memory acquisition conversation generation unit 50 makes a state transition from the node Node1 to the node Node2 at the initial stage. By outputting the character string data D1 of the utterance content to the speech synthesizing unit 53, it is confirmed whether or not the user can acquire the memory, and the internal state managing unit 34 (FIG. 5) is accessed. The parameter value of "affection" at that time is obtained.

  Next, the memory acquisition conversation generation unit 50 confirms that there is a positive response of “good” from the user at the node Node2 based on the voice recognition result of the auditory recognition function unit 32 (FIG. 5). The state transitions from the node Node2 to the node Node3. At this time, the acquired utterance associated with the arc Ac2, in this example, "What is Yukiko's favorite thing? Is output to the speech synthesizer 53.

  On the other hand, when the memory acquisition conversation generation unit 50 confirms that a negative response such as “Yadayo” has been received from the user at the node Node2 based on the speech recognition result of the auditory recognition function unit 32, the node The state transitions from Node2 to Node6. At this time, the character string data D1 of the utterance content of “I'm sorry” associated with the arc Ac6 is output to the speech synthesis unit 53.

  On the other hand, the memory acquisition conversation generation unit 50 confirms, based on the voice recognition result of the hearing recognition function unit 32, that the user has responded to the node Node3 that "I like tea is tea". The state transitions to the node Node4, and at this time, the character string data D1 of the utterance content of “tea” that is associated with the arc Ac3 is output to the speech synthesis unit 53.

  Subsequently, the memory acquisition conversation generation unit 50 confirms, based on the voice recognition result of the auditory recognition function unit 32, that the user has received a negative response such as “no” from the user at the node Node4. The state transits to Node3, and at this time, the acquired utterance associated with the arc Ac7, in this example, "What is Yukiko's favorite thing? Is output to the speech synthesizer 53.

  On the other hand, when the memory acquisition conversation generation unit 50 confirms that a positive response such as “yes” from the user has been received at the node Node4 based on the speech recognition result of the auditory recognition function unit 32, the node acquisition process proceeds to the node Node4. To the node Node5, and at this time, the character string data D1 of the utterance content of "Okay, I remembered" associated with the arc Ac4 is output to the speech synthesizer 53, and the internal state manager 34 (FIG. 5) To obtain the parameter value of “affection” at that time, and calculate the difference between the acquired parameter value of “affection” and the parameter value of “affection” acquired before the output of the acquired utterance.

  Further, the memory acquisition conversation generation unit 50 thereafter makes a state transition from the node Node5 to the node Node6, and at this time, “tea” as the value of the item “Favorite” of the user acquired as described above. The long-term memory 36 is stored in the long-term memory 36 (FIG. 5). At this time, the difference between the parameter value of “affection” calculated as described above and the degree of impression for the value is associated with “tea” in the long-term memory 36. To memorize.

  Then, when the state transition to the node Node6 is made, the memory acquisition conversation generation unit 50 notifies the situation determination unit 52 (FIG. 5) that a series of utterances has ended, and thereafter ends the current acquisition utterance generation process. .

  Incidentally, since only the first topic use method determination rule can be used for the acquired conversation as described above, all the acquired conversations are similarly performed in accordance with this one state transition model.

  On the other hand, FIGS. 10 and 11 show a first specific example of a conversation (use conversation) with a user using a value of an item performed under the control of the memory use conversation generation unit 51. In this example, it is assumed that the topic is "Birthday of the object ID1" and the first topic usage method determination rule is selected as the usage method.

  In this case, at the initial stage, the memory use conversation generation unit 51 makes a state transition from the node Node10 to the node Node11. By outputting the character string data D1 of the utterance content to the speech synthesizing unit 53, it is checked whether or not the user's talk is good, and the internal state management unit 34 (FIG. 5) is accessed to Get the parameter value of "love".

  Next, the memory use conversation generation unit 51 confirms that the node Node 11 has received a positive response from the user to “good” based on the voice recognition result of the auditory recognition function unit 32 (FIG. 5). The state transitions from the node Node11 to the node Node12. At this time, the usage utterance associated with the arc Ac11, in this example, “Yukiko's birthday is 73/5/2! Is output to the speech synthesizer 53. After that, the storage use conversation generation unit 51 further makes a state transition from the node Node12 to the node Node13.

  On the other hand, when the memory use conversation generation unit 51 confirms that a negative response such as “Yadayo” has been received from the user at the node Node11 based on the speech recognition result of the auditory recognition function unit 32, the node using the node The state transitions from Node 11 to Node 13, and at this time, the character string data D 1 of the utterance content “I'm sorry” associated with the arc Ac 13 is output to the speech synthesizer 53.

  Then, when the state transition to the node Node13 is made, the storage use conversation generation unit 51 notifies the situation determination unit 52 (FIG. 5) that a series of utterances has ended, and thereafter ends the current use conversation generation processing. .

  On the other hand, FIGS. 12 and 13 show a second specific example of a conversation (use conversation) with a user using a value of an item performed under the control of the memory use conversation generation unit 51. In this example, it is assumed that the topic is “Birthday of the object ID1” and the second topic usage method determination rule is selected as the usage method.

  In this case, the memory use conversation generation unit 51 makes a state transition from the node Node20 to the node Node21 at the initial stage, and at this time, the message "Yukikoko, do you want to talk?" By outputting the character string data D1 of the utterance content to the speech synthesizing unit 53, it is checked whether or not the user's talk is good, and the internal state management unit 34 (FIG. 5) is accessed to Get the parameter value of "love".

  Next, the memory use conversation generation unit 51 checks that the node Node 21 has received an affirmative response to the user from the user based on the voice recognition result of the auditory recognition function unit 32 (FIG. 5). The state transitions from the node Node21 to the node Node22. At this time, the value of "BIRTHDAY" of the "object ID1" is read from the long-term storage unit, and the obtained value "73/5/2" is used as a keyword. Search the anniversary database. Then, when the result of this search yields the item “traffic advertisement day”, the memory use conversation generation unit 51 uses the utterance, for example, “May 2 is a traffic advertisement day! Is output to the speech synthesizer 53. After that, the storage use conversation generation unit 51 further makes a state transition from the node Node22 to the node Node23.

  On the other hand, when the memory use conversation generation unit 51 confirms that a negative response such as “Yadayo” has been received from the user at the node Node 21 based on the voice recognition result of the auditory recognition function unit 32, The state transitions from Node 21 to Node Node 23, and at this time, the character string data D1 of the utterance content of “I'm sorry” associated with the arc Ac23 is output to the speech synthesis unit 53.

  Then, when the state transition to the node Node23 is made, the storage use conversation generation unit 51 notifies the situation determination unit 52 (FIG. 5) that a series of utterances has ended, and thereafter ends the use conversation generation processing of this time. .

  14 and 15 show a third specific example of a conversation (use conversation) with a user using a value of an item performed under the control of the memory use conversation generation unit 51. is there. In this example, it is assumed that the topic is “Birthday of the object ID1”, and the third topic usage method determination rule is selected as the usage method.

  In this case, the storage use conversation generation unit 51 transitions from the node Node 30 to the node Node 31 at the initial stage, and at this time, stores the value of “BIRTHDAY” of the “object ID 1” in the long-term storage unit 36 (FIG. 5). , Calculate the constellation using this as a keyword, and use the obtained "Taurus" item to say "Yukiko is Taurus, right?" Is generated and output to the speech synthesizer 53 (FIG. 5).

  Next, the memory use conversation generation unit 51 checks that the node Node 31 has received a positive response from the user, “Yes,” based on the voice recognition result of the auditory recognition function unit 32 (FIG. 5). The state transitions from the node Node31 to the node Node32. At this time, the horoscope database is searched using "Taurus" as a keyword. When the result of this search shows that "patience" is obtained, the memory use conversation generation unit 51 uses the utterance using this, for example, the character string data D1 of the utterance content "Taurus is patient". Is output to the speech synthesizer 53. After that, the storage use conversation generation unit 51 further makes a state transition from the node Node32 to the node Node33.

  On the other hand, when the memory use conversation generation unit 51 confirms that the node Node 31 has received a negative response such as “Yadayo” from the user based on the voice recognition result of the auditory recognition function unit 32, The state transitions from the node 31 to the node 33. At this time, the character string data D1 of the utterance content of “I'm sorry” associated with the arc Ac33 is output to the speech synthesizer 53.

  Then, when the state transition to the node Node33 is made, the storage use conversation generation unit 51 notifies the situation determination unit 52 (FIG. 5) that a series of utterances has ended, and thereafter ends the use conversation generation processing of this time. .

  FIGS. 16 and 17 show a fourth specific example of a conversation with a user (utilized conversation) using the value of a certain item under the control of the memory-based conversation generation unit 51. In this example, it is assumed that the topic is “Favorite of the object ID 1 (Favorite)” and the second topic use method determination rule is selected as the use method.

  In this case, the memory use conversation generation unit 51 makes a state transition from the node Node 40 to the node Node 41 at the initial stage, and at this time, “Can you talk about Rika, which is associated with the arc Ac40? By outputting the character string data D1 of the utterance content to the speech synthesizing unit 53, it is checked whether or not the user's talk is good, and the internal state management unit 34 (FIG. 5) is accessed to Get the parameter value of "love".

  Next, the memory use conversation generation unit 51 checks that the node Node 41 has received an affirmative response to the user from the user based on the voice recognition result of the auditory recognition function unit 32 (FIG. 5). The state transitions from the node Node41 to the node Node42. At this time, the value of "Favorite" of the "object ID1" is read from the long-term storage unit, and the obtained value "tulip" is used as a keyword, and the scenario database is used. Search for. Then, as a result of this search, the storage-based conversation generation unit 51 uses the scenario (collection of a plurality of sentences forming a script or script) stored in a location (not shown) represented by a scenario ID for “tulip”. Utterances using, for example, "Do you know the flower language of tulips? Each one is different depending on the color. There are many, for example, red tulips are confession of love, white tulips are lost love, yellow is hopeless love, purple is immortal love, green is beautiful eyes, and pink is old age. What did you say. When it gets warm, you can see it in many places. Tulips are so beautiful that they are good. I really like it. Is output to the speech synthesizer 53. After that, the storage use conversation generation unit 51 further makes a state transition from the node Node42 to the node Node43.

  On the other hand, when the memory use conversation generation unit 51 confirms that the node Node 41 has received a negative response such as “Yadayo” from the user based on the voice recognition result of the auditory recognition function unit 32, The state transitions from Node 41 to Node Node 43, and at this time, the character string data D 1 of the utterance content “I'm sorry” associated with the arc Ac 43 is output to the speech synthesis unit 53.

  When the state transition to the node Node 43 is made, the storage use conversation generation unit 51 notifies the situation determination unit 52 (FIG. 5) that a series of utterances has ended, and thereafter ends the use conversation generation processing of this time. .

  FIGS. 18 and 19 show a fifth specific example of a conversation (use conversation) with a user using a value of an item performed under the control of the memory use conversation generation unit 51. In this example, it is assumed that the topic is “Favorite of the object ID 1 (Favorite)” and the second topic use method determination rule is selected as the use method.

  In this case, the storage use conversation generation unit 51 makes a state transition from the node Node 50 to the node Node 51 at the initial stage, and at this time, "Can you talk about Jiro?" By outputting the character string data D1 of the utterance content to the speech synthesizing unit 53, it is checked whether or not the user's talk is good, and the internal state management unit 34 (FIG. 5) is accessed to Get the parameter value of "love".

  Next, the memory use conversation generation unit 51 checks that the node Node 51 has received an affirmative response to the user from the user based on the voice recognition result of the auditory recognition function unit 32 (FIG. 5). The state transitions from the node Node51 to the node Node52. At this time, the value of "Favorite" of the "object ID1" is read from the long-term storage unit, and the obtained value "soccer" is used as a keyword, and the sports database is used. Search for. Then, as a result of this search, the memory-based conversation generation unit 51 asks, “Do you want to play soccer? ”And“ Behavior to kick ”are obtained, and utterances using these are obtained, for example,“ Speaking of which, should we play soccer? Is output to the voice synthesizing unit 53 of the hardware-dependent layer control unit 40, and an action command representing "ball kicking action" is output to the action expression unit ( (Not shown). After that, the storage use conversation generation unit 51 further makes a state transition from the node Node52 to the node Node53.

  As a result, the robot 1 said, "Speaking of which, I wanted to play soccer! Not only is output as a voice through the speaker 12 but also the corresponding drive units 151 to 15n of the drive unit 4 are driven based on the action command, so that the ball in front of the eyes is kicked or simulated. An action as if kicking a ball can be expressed.

  On the other hand, when the memory use conversation generation unit 51 confirms that a negative response such as “Yadayo” has been received from the user at the node Node 51 based on the voice recognition result of the auditory recognition function unit 32, The state transitions from the node 51 to the node 53. At this time, the character string data D1 of the utterance content “I'm sorry” associated with the arc Ac53 is output to the speech synthesizer 53.

  Then, when the state transition to the node Node53 is made, the storage use conversation generation unit 51 notifies the situation determination unit 52 (FIG. 5) that a series of utterances has ended, and thereafter ends this use conversation generation processing. .

(3) Operation and effect of the present embodiment In the above configuration, the robot 1 acquires various information about the user through a conversation (acquired conversation) with the user, and stores the acquired information as a value for an item for a long term. In addition to the storage in the storage unit 36 (FIG. 5), the user communicates with the user using the values of the items stored in the long-term storage unit 36 (use conversation).

  Therefore, the robot 1 can perform a dialogue between the user and the user or contents related to the user, that is, a dialogue specific to the user, and the user can feel the personalization process in this way. Therefore, the user's feeling of closeness to the robot 1 can be improved.

  In this case, the robot 1 generates a topic in the next conversation within a range that can be associated with the topic of the immediately preceding conversation in accordance with the first to sixth topic generation rules described above. In the course of the process, the topic does not change extremely, and not only the information on the acquired user etc. is used as it is, but also the horoscope and the anniversary search etc. are performed according to the above-mentioned first to third topic use method determination rules. Therefore, the conversation with the robot 1 can be made more interesting and conversation variations can be increased.

  Further, in the robot 1, the memory acquisition conversation generation unit 50 and the memory use conversation generation unit 51 have a conversation history, and based on the conversation history, a combination of the same topic and its use method in a series of conversations with the same user. Is not generated, the utterance relating to the information acquired by the robot 1 and associated with the user is not output immediately after the acquisition, and it is possible to prevent a decrease in interest.

  In addition, under the control of the situation determination unit 52, the robot 1 switches between the acquired conversation and the used conversation according to the acquisition state of the information accompanying the user. There is no appearance of behavior using memory when there is no information attached to the information, and no appearance of unnatural behavior appears in the situation where all information has been acquired. It can be prevented before it happens.

  Further, in the robot 1, the value of each item for the user and the like is stored in the long-term storage unit 36 together with the impression degree, and the topic is selected based on the impression degree. It is possible to change the appearance frequency of topics related to the robot 1 and to cause the robot 1 to perform a highly entertaining interaction of speaking what the robot 1 wants to speak and not speaking what he does not want to speak.

  According to the above configuration, the robot 1 acquires various information about the user through a conversation with the user, and stores the acquired information as a value for the item in the long-term storage unit 36 (FIG. 5). By using the value of each item stored in the unit 36 to converse with the user, a conversation specific to the user as the target object can be performed with the user, and personalized in this way. Since the user can feel the process, the user's familiarity with the robot 1 can be improved, and thus a robot with higher entertainment can be realized.

(4) Other Embodiments In the above-described embodiment, the case where the present invention is applied to the entertainment robot configured as shown in FIGS. 1 to 3 has been described, but the present invention is not limited to this. The present invention is not limited to this, and can be widely applied to entertainment robots having other configurations, other robots, various devices other than robots, and various other interaction control devices that perform interaction control with users. Further, the present invention can be widely applied to various software such as software for video games.

  Further, in the above-described embodiment, a case has been described in which the impression degree of each item described above with reference to FIG. 4 is determined at the time of acquiring the value of the item and is not updated thereafter. Not limited to this, the impression degree may be updated. By doing so, the appearance frequency of various topics can be changed, and a conversation corresponding to the situation at that time can be made accordingly, so that the entertainment performance of the robot 1 is further improved. be able to.

  Further, in the above-described embodiment, the impression degree of each item described above with reference to FIG. 4 is calculated based on the difference between the parameter values of “affection” held in the internal state management unit 34 before and after acquiring the item. However, the present invention is not limited to this, and may be calculated based on the parameter values of other instinct elements or emotional elements held in the internal state management unit 34. It may be obtained by a technique.

  Further, in the above-described embodiment, a case has been described in which the RAM 22 or the non-volatile memory 24 (FIG. 2), which is an internal memory of the control unit 2, is used as storage means for storing the value of each item of the object. However, the present invention is not limited to this, and another storage medium may be used, or an external storage medium such as an external memory may be used.

  Furthermore, in the above-described embodiment, a topic related to the topic used in the immediately preceding conversation is selected, and the acquired conversation or the long-term storage unit 36 for acquiring the value of the item in the selected topic has already been stored. A conversation generating means for generating a use conversation using the value of an item in the topic as the next conversation; a storage acquisition conversation generation unit 50 as a storage acquisition conversation generation means for generating an acquisition conversation; and a use conversation. The user selects one of the storage use conversation generation unit 51 as the storage use conversation generation unit and one of the storage acquisition conversation generation unit 50 and the storage use conversation generation unit 51, and selects the selected storage use conversation generation unit 50 or the storage use conversation generation unit. Although the description has been made of the case where the unit 51 is constituted by three modules of the situation determination unit 52 as the situation determination unit for generating the next conversation, the present invention Not limited to Les, the construction of the conversation generation means can be widely applied this various other configurations.

  Furthermore, in the above-described embodiment, a case has been described in which six rules of the above-described first to seventh topic generation rules are prepared as topic generation rules that define a method of generating a topic of the next conversation. However, the present invention is not limited to this, and various other rules can be widely applied as such topic generation rules. Also, the number may be a number other than 7.

  Similarly, in the above-described embodiment, three rules of the above-described first to third topic use method determination rules are prepared as topic use method determination rules that define the use method of the generated topic. Although the case has been described, the present invention is not limited to this, and various other rules can be widely applied as such topic use method determination rules. Also, the number may be a number other than three.

  Further, in the above-described embodiment, as shown in FIGS. 17 and 18, the next topic related to the item is selected using the topic item used in the immediately preceding conversation as a keyword, and the previously stored relevant topic is selected. Although the case where the use conversation is generated using the scenario corresponding to the item has been described, the present invention is not limited to this, and the point is that the conversation is made consistent by connecting the keywords after the conversation. If it can be provided, it can be applied to various other specific examples. In addition, if an action command for expressing an action corresponding to a use conversation can be generated and an action corresponding to the action command can be expressed, the conversation can be further expanded in addition to consistency.

  The interaction control device and method and the robot device can be applied to an amusement robot, a care robot, and the like.

FIG. 2 is a block diagram schematically illustrating a functional configuration of the robot according to the present embodiment. FIG. 3 is a block diagram illustrating a configuration of a control unit. FIG. 3 is a block diagram illustrating a software configuration of the robot. It is a conceptual diagram provided for explanation of the recording format of acquisition information. FIG. 2 is a conceptual diagram showing a configuration of a main part relating to a dialogue control function in the action control system. It is a conceptual diagram provided for explanation of a conversation generation procedure. It is a flowchart which shows a dialog generation processing procedure. FIG. 14 is a conceptual diagram explaining a specific processing example of a memory acquisition conversation generation unit during a conversation generation process. FIG. 14 is a conceptual diagram explaining a specific processing example of a memory acquisition conversation generation unit during a conversation generation process. FIG. 10 is a conceptual diagram serving to explain a first specific processing example of a memory-based conversation generation unit during a conversation generation process. FIG. 10 is a conceptual diagram serving to explain a first specific processing example of a memory-based conversation generation unit during a conversation generation process. It is a conceptual diagram explaining the 2nd specific example of a process of the memory use conversation production | generation part at the time of a conversation production | generation process. It is a conceptual diagram explaining the 2nd specific example of a process of the memory use conversation production | generation part at the time of a conversation production | generation process. FIG. 18 is a conceptual diagram serving to explain a third specific processing example of the storage-based conversation generation unit during the conversation generation processing. FIG. 18 is a conceptual diagram serving to explain a third specific processing example of the storage-based conversation generation unit during the conversation generation processing. FIG. 18 is a conceptual diagram serving to explain a fourth specific processing example of the memory use conversation generation unit during the conversation generation processing. FIG. 18 is a conceptual diagram serving to explain a fourth specific processing example of the memory use conversation generation unit during the conversation generation processing. FIG. 21 is a conceptual diagram serving to explain a fifth specific processing example of the storage-based conversation generation unit during the conversation generation processing. FIG. 21 is a conceptual diagram serving to explain a fifth specific processing example of the storage-based conversation generation unit during the conversation generation processing.

Explanation of reference numerals

  DESCRIPTION OF SYMBOLS 1 ... Robot, 2 ... Control unit, 10 ... CCD camera, 11 ... Microphone, 12 ... Speaker, 21 ... CPU, 23 ... ROM, 31 ... Visual recognition function part, 32 ... Auditory recognition Functional unit 34 Internal state management unit 36 Long-term storage unit 38 Situation-dependent behavior hierarchy 50 Memory acquisition conversation generation unit 51 Memory use conversation generation unit 52 Situation determination unit , 53... Voice synthesis unit, D1... Character string data, RT1.

Claims (40)

Storage means for storing various information associated with the object as a value of a corresponding item of the object;
When the item of the object is defined as a topic, the topic that is related to the topic used in the immediately preceding conversation is selected, and the acquired conversation for acquiring the value of the item in the selected topic or Conversation generating means for generating, as the next conversation, a use conversation using the value of the item in the topic already stored in the storage means,
The above conversation generation means,
A dialogue control device, wherein the value acquired by the acquired conversation is stored in the storage means as the value of the corresponding item. The above conversation generating means,
Another optional item of the same object as the topic used in the immediately preceding conversation is selected as the next topic, and the storage unit uses the value of the item already stored and uses the item. The dialogue control device according to claim 1, wherein the conversation control device generates a conversation. The above conversation generation means,
The item related to the same object as the topic used in the immediately preceding conversation is selected as the next topic, and the value of the item already stored in the storage means is used for the use conversation. The dialogue control device according to claim 1, wherein: The above conversation generating means,
Selecting any of the items of the object that can be specified from the value of the item of the topic used in the immediately preceding conversation as the next topic, and any of the items already stored in the storage means; The dialogue control device according to claim 1, wherein the use conversation is generated by using the value of?. The above conversation generation means,
The same topic of the same object as the topic used in the immediately preceding conversation is selected as the next topic, and the use conversation is generated using the value of the item already stored in the storage means. The interactive control device according to claim 1, wherein: The above conversation generating means,
The same item of another object having the same value as the value of the item in the topic used in the previous conversation is selected as the next topic, and the same item already stored in the storage unit is selected. The interaction control device according to claim 1, wherein the use conversation is generated by using the value of the item. The above conversation generating means,
The item of another object having the value related to the value of the item in the topic used in the immediately preceding conversation is selected as the next topic, and the item stored in the storage unit is selected. The interaction control device according to claim 1, wherein the use conversation is generated by using a related value. The above conversation generation means,
Using the item of the topic used in the immediately preceding conversation as a keyword, selecting the next topic related to the item, and using the scenario corresponding to the item already stored in the storage means, The conversation control device according to claim 1, wherein the conversation control device generates a use conversation. The interaction control device according to claim 8, further comprising: an action generation unit that generates an action command for expressing an action corresponding to the use conversation generated by the conversation generation unit. The above conversation generating means,
Another optional item of the same object as the topic used in the immediately preceding conversation is selected as the next topic, and the acquired conversation for acquiring the value of the optional item is generated. The interactive control device according to claim 1, wherein: The above conversation generating means,
Selecting the item related to the same object as the topic used in the immediately preceding conversation as the next topic, and generating the acquired conversation for acquiring the value of the related item. The dialogue control device according to claim 1, wherein: The above conversation generation means,
Selecting one of the items of the object that can be identified from the value of the item in the topic used in the immediately preceding conversation as the next topic, and acquiring the value of the one of the items. The interaction control device according to claim 1, wherein the conversation control device generates an acquisition conversation. The above conversation generation means,
The interaction control device according to claim 1, wherein the use conversation is generated by using a matter obtained based on the value of the item in the selected topic. The above conversation generation means,
Memory acquisition conversation generation means for generating the acquisition conversation;
Storage use conversation generation means for generating the use conversation,
Selecting one of the storage acquisition conversation generation means and the storage utilization conversation generation means, and causing the selected storage acquisition conversation generation means or the storage utilization conversation generation means to generate the next conversation. The interactive control device according to claim 1, further comprising: The situation determining means is as follows:
A first degree, which is the degree of the number of the items that have not obtained the value with respect to the total value of the items in the dialogue partner, and the first degree of the item that has obtained the value with respect to the total value in the dialogue partner; Determining which of the storage acquisition conversation generation means and the storage use conversation generation means should generate the acquisition conversation or the use conversation based on the second degree of the number. Item 15. The dialogue control device according to Item 14. The situation determining means is as follows:
When the first degree is equal to or greater than the second degree, the storage-acquired conversation generation means is configured to generate the acquired conversation. When the first degree is smaller than the second degree, the storage-based conversation generation is performed. The dialogue control device according to claim 14, wherein said means generates said use conversation. The above conversation generation means,
The interaction control device according to claim 1, wherein a history of the used topics is held, and the acquired conversation or the use conversation is generated while referring to the history. The above conversation generation means,
The dialogue control according to claim 17, wherein the acquired conversation or the use conversation is generated so as not to use the same topic in one conversation with the same conversation partner while referring to the history. apparatus. The storage means,
The value of the item of the object is stored together with the impression degree serving as a measure of whether the value may be used for the conversation,
The above conversation generation means,
The interaction control device according to claim 1, wherein the topic to be used in the next conversation is selected based on the impression degree. An internal state management means for holding a parameter representing an internal state and changing the parameter value according to an external stimulus;
The impression degree is
20. The dialogue control device according to claim 19, wherein the dialogue control device comprises a difference between the parameter value of the predetermined parameter held by the internal state management unit before and after the corresponding value is obtained. A first step of storing various information associated with the object as a value of a corresponding item of the object;
When the item of the object is defined as a topic, the topic that is related to the topic used in the immediately preceding conversation is selected, and the acquired conversation for acquiring the value of the item in the selected topic or A second step of generating, as the next conversation, a use conversation using the above-mentioned value of the item in the topic already stored,
In the second step,
A dialogue control method characterized by storing the value acquired by the acquired conversation as the value of the corresponding item. In the second step,
Another optional item of the same object as the topic used in the immediately preceding conversation is selected as the next topic, and the storage unit uses the value of the item already stored and uses the item. The conversation control method according to claim 21, wherein a conversation is generated. In the second step,
The item related to the same object as the topic used in the immediately preceding conversation is selected as the next topic, and the value of the item already stored in the storage means is used for the use conversation. The interactive control method according to claim 21, wherein: In the second step,
Selecting any of the items of the object that can be specified from the value of the item of the topic used in the immediately preceding conversation as the next topic, and any of the items already stored in the storage means; 22. The interaction control method according to claim 21, wherein the use conversation is generated by using the value of?. In the second step,
The same topic of the same object as the topic used in the immediately preceding conversation is selected as the next topic, and the use conversation is generated using the value of the item already stored in the storage means. The interactive control method according to claim 21, wherein: In the second step,
The same item of another object having the same value as the value of the item in the topic used in the previous conversation is selected as the next topic, and the same item already stored in the storage unit is selected. 20. The interaction control method according to claim 19, wherein the use conversation is generated by using the value of the item. In the second step,
The item of another object having the value related to the value of the item in the topic used in the immediately preceding conversation is selected as the next topic, and the item stored in the storage unit is selected. 22. The interaction control method according to claim 21, wherein the use conversation is generated using a related value. In the second step,
Using the item of the topic used in the immediately preceding conversation as a keyword, selecting the next topic related to the item, and using the scenario corresponding to the item already stored in the storage means, 22. The interaction control method according to claim 21, wherein a use conversation is generated. In the second step,
The interaction control method according to claim 28, wherein an action command for expressing an action corresponding to the generated use conversation is generated. In the second step,
Another optional item of the same object as the topic used in the immediately preceding conversation is selected as the next topic, and the acquired conversation for acquiring the value of the optional item is generated. The interactive control method according to claim 21, wherein: In the second step,
Selecting the item related to the same object as the topic used in the immediately preceding conversation as the next topic, and generating the acquired conversation for acquiring the value of the related item. The interactive control method according to claim 21, characterized in that: In the second step,
Selecting one of the items of the object that can be identified from the value of the item in the topic used in the immediately preceding conversation as the next topic, and acquiring the value of the one of the items. The interaction control method according to claim 21, wherein an acquisition conversation is generated. In the second step,
22. The interaction control method according to claim 21, wherein the use conversation is generated using a matter obtained based on the value of the item in the selected topic. In the second step,
A first degree, which is the degree of the number of the items that have not obtained the value with respect to the total value of the items in the dialogue partner, and the first degree of the item that has obtained the value with respect to the total value in the dialogue partner; 22. The interaction control method according to claim 21, wherein it is determined whether the acquired conversation or the use conversation is to be generated as the next conversation based on a second degree of the number. In the second step,
35. The method according to claim 34, wherein the acquired conversation is generated when the first degree is equal to or higher than the second degree, and the use conversation is generated when the first degree is smaller than the second degree. 3. The interaction control method according to item 1. In the second step,
22. The interaction control method according to claim 21, wherein a history of the used topics is retained, and the acquired conversation or the use conversation is generated while referring to the history. In the second step,
37. The dialogue control according to claim 36, wherein the acquired conversation or the use conversation is generated so as not to use the same topic in one conversation with the same conversation partner while referring to the history. Method. In the first step,
The value of the item of the object is stored together with the impression degree serving as a measure of whether the value may be used for the conversation,
In the second step,
22. The interaction control method according to claim 21, wherein the topic to be used for the next conversation is selected based on the impression degree. In the first step,
Holding a parameter representing the internal state, changing the parameter value according to the external stimulus,
The impression degree is
39. The interaction control method according to claim 38, comprising a difference between the parameter value of the predetermined parameter held by the internal state management unit before and after the acquisition of the corresponding value. Storage means for storing various information associated with the object as a value of a corresponding item of the object;
When the item of the object is defined as a topic, the topic that is related to the topic used in the immediately preceding conversation is selected, and the acquired conversation for acquiring the value of the item in the selected topic or Conversation generating means for generating, as the next conversation, a use conversation using the value of the item in the topic already stored in the storage means,
The above conversation generating means,
The robot device according to claim 1, wherein the value acquired by the acquired conversation is stored in the storage unit as the value of the corresponding item.

Images