KR102598435B1 - Apparatus and method for determning emotional state of robot reflecting emotional change pattern of human - Google Patents

Abstract

An apparatus and method for determining the emotional state of a robot are disclosed. An apparatus for determining the emotional state of a robot according to an embodiment receives a vector corresponding to the personality of the robot as a first vector on an emotional space plane divided into a plurality of emotional regions, and receives an emotional space corresponding to the current emotional state of the robot. a receiving unit that receives a second vector on a plane; a conversion unit that converts the emotional state of the input sentence into a third vector on the emotional space plane using a previously learned model; A fourth vector on the emotional space plane corresponding to the maximum emotional state of the robot changed by the input sentence in the current emotional state is calculated based on the result of the operation between the first vector, the second vector, and the third vector. A calculation unit that calculates; a connection path determination unit that determines a connection path passing through the second vector and the fourth vector; and an emotion determination unit that determines a subsequent emotional state for the current emotional state based on the connection path.

Description

Apparatus and method for determining the emotional state of a robot that reflects human emotional change patterns {APPARATUS AND METHOD FOR DETERMNING EMOTIONAL STATE OF ROBOT REFLECTING EMOTIONAL CHANGE PATTERN OF HUMAN}

The disclosed embodiments relate to techniques for determining the emotional state of a robot.

[Explanation on state-supported research and development]

This research was supported by the Ministry of Science and ICT, the National Research Council of Science and Technology, research operating expenses (R&D), [Early prediction of dementia in the elderly generation, development of treatment and patient care technology (development of life care robots to support dementia patients), detailed project number: CRC-15- This was achieved with the support of [04-KIST].

Today, robot systems are expanding their application areas in various fields, such as assistant/personal assistant robots that provide chatbot services based on conversations with users, and reading service robots that read books/articles. In particular, robot services that require interaction with people need to provide responses (facial expressions and gestures) appropriate to the situation to provide more natural services.

In this regard, technology that generates robot emotions appropriate to the context by detecting image information showing the user's face and sound information containing the user's voice has been actively researched.

However, there is a limitation in existing technology that user information is needed to generate robot emotions. In other words, the generated robot's emotions are merely the output of the input of user information. Because of this, existing technology has a problem in that it is difficult to generate robot emotions corresponding to user information when user information does not exist.

Additionally, changes in a person's emotions due to external stimuli vary depending on the person's personality and tend to decrease over time. However, the emotions of robots generated by existing technology are fixed without change unless user information is input. Moreover, there is a limitation in existing technology that the robot's emotions change unnaturally as the number of outputs increases. In the case of people, even if emotions change due to external factors, they converge and change to emotions similar to the current emotion. On the other hand, in the case of a robot, for example, when the current emotion is 'happy' and the user's information including the 'sad' emotion is repeatedly input to the robot, the robot's emotions change from 'happy' to 'sad' and ' There is a radical change in the pattern of ‘sadness’. These differences cause a sense of heterogeneity in users.

As such, in the field of technology for generating robot emotions, only the technology for generating robot emotions in response to user information is being actively researched, and the technology for generating robot emotions in patterns similar to human emotional change patterns is that. Research is insufficient.

Therefore, for more natural interactions, the robot's emotions need to be changed gradually and recoverably like human emotions, taking into account the robot's uttered sentences.

Korean Patent No. 10-2278162 (registered on July 12, 2021)

The disclosed embodiments are intended to provide an apparatus and method for determining the emotional state of a robot.

A method for determining the emotional state of a robot according to an embodiment includes receiving a vector corresponding to the personality of the robot as a first vector on an emotional space plane divided into a plurality of emotional regions; Receiving a vector corresponding to the current emotional state of the robot as a second vector on the emotional space plane; Converting the emotional state of the input sentence into a third vector on the emotional space plane using a previously learned model; A fourth vector on the emotional space plane corresponding to the maximum emotional state of the robot changed by the input sentence from the current emotional state is calculated based on the result of the operation between the first vector, the second vector, and the third vector. calculating step; determining a connection path passing through the second vector and the fourth vector; and determining a subsequent emotional state for the current emotional state based on the connection path.

The emotional space plane may include a pleasure axis and an arousal axis.

In the calculating step, the fourth vector can be calculated using Equation 1 below.

[Equation 1]

(At this time, is the fourth vector, is the second vector, is the third vector, is the first vector, and is a weight variable with a random real value between 0 and 1)

Determining the connection path may include determining a first output emotional region for the emotional region mapped to each of the second vector and the third vector based on a predefined node graph rule; and determining the connection path as a preliminary connection path with the shortest path length among a plurality of preliminary connection paths passing through the second vector, the fourth vector, and the first output emotion area.

Determining the subsequent emotional state may include determining any one of a plurality of vectors on the connection path; and determining the subsequent emotional state based on an emotional region mapped to the arbitrary one of the plurality of emotional regions.

The step of determining the subsequent emotional state may determine, as the subsequent emotional state, an emotional region mapped to a vector spaced apart from the second vector by a preset interval along the connection path.

A method for determining an emotional state of a robot according to an embodiment includes: determining a regression path passing through the first vector and the second vector when the input sentence is not input within a preset threshold time interval; and determining the subsequent emotional state based on the regression path.

Determining the regression path may include determining a second output emotional region for the emotional region mapped to each of the first vector and the second vector based on a predefined node graph rule; and determining, as the regression path, a preliminary regression path with the shortest path length among a plurality of preliminary regression paths passing through the first vector, the second vector, and the second output emotion area.

The step of determining the subsequent emotional state may determine, as the subsequent emotional state, an emotional region mapped to a vector spaced apart from the second vector by a preset interval along the regression path.

An apparatus for determining the emotional state of a robot according to an embodiment receives a vector corresponding to the personality of the robot as a first vector on an emotional space plane divided into a plurality of emotional regions, and receives an emotional space corresponding to the current emotional state of the robot. a receiving unit that receives a second vector on a plane; a conversion unit that converts the emotional state of the input sentence into a third vector on the emotional space plane using a previously learned model; A fourth vector on the emotional space plane corresponding to the maximum emotional state of the robot changed by the input sentence from the current emotional state is calculated based on the result of the operation between the first vector, the second vector, and the third vector. A calculation unit that calculates; a connection path determination unit that determines a connection path passing through the second vector and the fourth vector; and an emotion determination unit that determines a subsequent emotional state for the current emotional state based on the connection path.

The emotional space plane may include a pleasure axis and an arousal axis.

The calculation unit may calculate the fourth vector using Equation 1 below.

[Equation 1]

(At this time, is the fourth vector, is the second vector, is the third vector, is the first vector, and is a weight variable with a random real value between 0 and 1)

The connection path determination unit determines a first output emotion region for the emotion region mapped to each of the second vector and the third vector based on a predefined node graph rule, and the second vector and the fourth vector. And among the plurality of preliminary connection paths passing through the first output emotion area, a preliminary connection path with the shortest path length may be determined as the connection path.

The emotion determination unit may determine an arbitrary one of a plurality of vectors on the connection path and determine the subsequent emotional state based on an emotional region mapped to the arbitrary one of the plurality of emotional regions.

The emotion determination unit may determine, as the subsequent emotional state, an emotional region mapped to a vector separated by a preset interval from the second vector along the connection path.

The apparatus for determining the emotional state of a robot according to an embodiment further includes a regression path determination unit that determines a regression path passing through the first vector and the second vector when the input sentence is not input within a preset threshold time interval. It can be included.

The regression path determination unit determines a second output emotional region for the emotional region mapped to each of the first vector and the second vector based on a predefined node graph rule, and determines a second output emotional region for the emotional region mapped to each of the first and second vectors. And among the plurality of preliminary regression paths passing through the second output emotion area, a preliminary regression path with the shortest path length may be determined as the regression path.

The emotion determination unit may determine, as the subsequent emotional state, an emotional area mapped to a vector separated by a preset interval from the second vector along the regression path.

In the disclosed embodiments, the emotional state of the robot can be determined without user information by using the robot's spoken sentence as an input sentence.

Disclosed embodiments may provide subsequent emotional states that are similar to a person's emotional change pattern by limiting categories of subsequent emotional states through connection paths.

In the disclosed embodiments, when there is no input sentence within a preset time, a subsequent emotional state in which the emotional state is recovered may be determined.

1 is a block diagram illustrating an apparatus for determining an emotional state of a robot according to an embodiment.
2 is a graph showing an emotional space plane according to one embodiment.
Figure 3 is a graph showing the node graph rules used by the connection path determination unit of Figure 1
Figure 4 is an example diagram showing a subsequent emotional state determined on the emotional space plane according to one embodiment.
Figure 5 is a block diagram illustrating an apparatus for determining an emotional state of a robot according to an additional embodiment.
Figure 6 is an example diagram showing the subsequent emotional state of a robot recovered by personality according to an embodiment.
7 is a flowchart illustrating a method for determining the emotional state of a robot according to an embodiment.
8 is a flowchart illustrating a method for determining the emotional state of a robot according to an additional embodiment.

The terminology used in this specification is a general term that is currently widely used as much as possible while considering function, but this may vary depending on the intention or practice of a technician working in the art or the emergence of new technology. In addition, in certain cases, there are terms arbitrarily selected by the applicant, and in this case, the meaning will be described in the explanation part of the relevant specification. Therefore, we would like to clarify that the terms used in this specification should be interpreted based on the actual meaning of the term and the overall content of this specification, not just the name of the term.

Additionally, the embodiments described herein may have aspects that are entirely hardware, partly hardware and partly software, or entirely software. In this specification, “unit,” “module,” “device,” “server,” or “system” refers to hardware, a combination of hardware and software, or software, etc. Refers to a computer-related entity. For example, a unit, module, device, server, or system may refer to hardware constituting part or all of a platform and/or software such as an application for running the hardware.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings and the contents described in the accompanying drawings, but the claimed scope is not limited or limited by the embodiments.

Figure 1 is a block diagram for explaining an apparatus 100 for determining the emotional state of a robot according to an embodiment.

According to one embodiment, the apparatus 100 for determining the emotional state of a robot may determine the subsequent emotional state of the robot that is changed by the input sentence to be similar to the emotional change pattern of a person.

According to one embodiment, the apparatus 100 for determining the emotional state of a robot includes the first vector 201, the second vector 202, the third vector 203, the fourth vector 204, and the first output of FIG. The vector 206 corresponding to the subsequent emotional state determined based on the emotional region 205 may be mapped to the emotional region of the emotional space plane 200 to determine the subsequent emotional state.

As shown in FIG. 2, the emotional space plane 200 may be a two-dimensional plane composed of two independent parameters, a pleasure axis and an arousal axis.

That is, the first vector 201, the second vector 202, the third vector 203, the fourth vector 204, the first output emotion region 205, and the vector corresponding to the subsequent emotional state on the emotion space plane. The vector on the emotional space plane including (206) may be a two-dimensional vector composed of positivity and arousal.

At this time, the emotional state determination device 100 of the robot uses a predefined classifier (e.g., Support Vector Machine (SVM), Stochastic Gradient Descent (SGD), K-nearest neighbor). The emotional space plane 200 can be divided into a plurality of emotional regions using (K-Nearest Neighbor; KNN) algorithm and Random forest (RF) algorithm).

As shown in FIG. 2, the emotional state determination device 100 of the robot divides each vector on the emotional space plane 200 into a happy region, neutral region, and surprise region through a predefined classifier. , the emotional space plane 200 can be divided into a plurality of emotional areas by classifying them into an angry area and a sad area.

Meanwhile, in the above-described example, the type and number of emotional areas included in the emotional space plane 200 are merely exemplary, and the emotional state determination device 100 of the robot is not limited to the type and number of emotional areas, and the emotional space plane ( 200) can be divided into various emotional areas.

Meanwhile, in the above-mentioned example, the predefined classifiers are exemplified as support vector machines, stochastic gradient descent, K-nearest neighbor algorithm, and random forest algorithms, but are not necessarily limited thereto, and various classification methods may be used depending on the embodiment. It may further include.

Referring to FIG. 1, the apparatus 100 for determining the emotional state of a robot includes a reception unit 110, a conversion unit 120, a calculation unit 130, a connection path determination unit 140, and an emotion determination unit 150. .

According to one embodiment, the reception unit 110, the conversion unit 120, the calculation unit 130, the connection path determination unit 140, and the emotion determination unit 150 are implemented using one or more physically separate devices. , It may be implemented by one or more processors or a combination of one or more processors and software, and, unlike the example shown, specific operations may not be clearly distinguished.

The receiving unit 110 receives a vector corresponding to the robot's personality as the first vector 201 on the emotional space plane 200 divided into a plurality of emotional regions, and receives the vector corresponding to the robot's current emotional state as the first vector 201 on the emotional space plane 200 corresponding to the robot's current emotional state. ) receives the second vector 202 on the

At this time, the robot's personality can be expressed in a vector format including openness, extraversion, agreeableness, and neuroticism among the Big Five personality traits as parameters.

In other words, the robot's personality can be expressed as a vector P(O, E, A, N). At this time, O represents the degree of openness, E represents the degree of extroversion, A represents the degree of likability, and N represents the degree of neuroticism. (However, O, E, A, N are random real numbers between -1 and 1)

The first vector 201 corresponding to the robot's personality can be converted using Equation 1 below.

[Equation 1]

At this time, each parameter value may be a value preset by user input. For example, if the vector P is preset to (1, 0.5, 1, 0.1) as shown in FIG. 2, the first vector 201 may be preset to (0.676, 0.507) on the emotional space plane 200. there is.

The conversion unit 120 converts the emotional state of the input sentence into the third vector 203 on the emotional space plane 200 using a previously learned model.

The input sentence may include a sentence generated based on the output information of the robot. For example, the input sentence may include a sentence generated based on output information such as text information, voice information, video information, etc. randomly output by the robot.

Meanwhile, in the above example, the input sentence is exemplified as being generated based on text information, voice information, and video information arbitrarily output by the robot, but it is not necessarily limited thereto, and is generated through various output information of the robot depending on the embodiment. It can be.

The calculation unit 130 divides the fourth vector 204 on the emotional space plane 200 corresponding to the maximum emotional state of the robot changed by the input sentence from the current emotional state into the first vector 201 and the second vector 202. ) and the third vector 203.

At this time, the maximum emotional state of the robot may mean the emotional state that can be maximally removed from the current emotional state by the input sentence on the emotional space plane 200.

According to one embodiment, the calculation unit 130 may calculate the fourth vector 204 using Equation 2 below.

[Equation 2]

At this time, is the fourth vector 204, is the second vector 202, is the third vector (203), is the first vector 201, and represents a weight variable having a random real number value between 0 and 1.

That is, according to one embodiment, the calculation unit 130 may calculate the fourth vector 204 by adding the weight vectors caused by the first vector 201 and the third vector 203.

According to one embodiment, the calculation unit 130 may calculate the fourth vector 204 without considering the first vector 201. In other words, by user input When is set to 0, the calculation unit 130 may calculate the fourth vector 204 corresponding to the maximum emotional state without considering the robot's personality.

According to one embodiment, the calculation unit 130 may calculate the fourth vector 204 without considering the third vector 203. In other words, by user input When set to 0 by this user input, the calculation unit 130 can calculate the fourth vector 204 corresponding to the maximum emotional state without considering the input sentence.

The connection path determination unit 140 may determine a connection path passing through the second vector 202 and the fourth vector 204.

At this time, the connection path may mean a limited area within the emotional space plane 200 used to determine the subsequent emotional state.

Specifically, the connection path determination unit 140 outputs a first output emotion area 205 for the emotion area mapped to each of the second vector 202 and the third vector 203 based on a predefined node graph rule. , and the preliminary connection path with the shortest path length among the plurality of preliminary connection paths passing through the second vector 202, the fourth vector 204, and the first output emotion area 205 may be determined as the connection path. .

At this time, the node graph rule is a graph representing the operation of transitioning from the current state to the subsequent state for an arbitrary event, and may include, for example, a finite state machine (FSM).

Accordingly, according to one embodiment, the connection path determination unit 140 determines that the state of the emotional region mapped to the second vector 202 is the current node, and the emotional region mapped to the third vector 203 may occur as a random event. In this case, the second output emotion area 205 may be determined as the next node to be transferred based on predefined node graph rules.

Meanwhile, the preliminary connection path with the shortest path length may be determined based on a predefined shortest path algorithm. The predefined shortest path algorithm may include, for example, an algorithm that can determine a path that avoids or passes through a predetermined point, such as the A* algorithm.

The emotion determination unit 150 determines a subsequent emotional state for the current emotional state based on the connection path.

According to one embodiment, the emotion determination unit 150 may determine any one of a plurality of vectors on the connection path and determine a subsequent emotional state based on the emotional region mapped to any one of the plurality of emotional regions. .

As a specific example, the emotion determination unit 150 may determine an emotion region mapped to a vector separated by a preset interval from the second vector 202 along the connection path as the subsequent emotional state. At this time, the preset interval can be determined from the results of a predefined linear equation.

FIG. 3 is a graph showing predefined node rules used by the connection path unit of FIG. 1.

Referring to Figure 3, the graph includes nodes where arrows come out, nodes where arrows come in, and edges. At this time, the graph may mean a rule model for determining the first output emotional region 205.

In other words, the graph may mean a rule model for determining an emotional region that determines a connection path from the second vector 202 to the fourth vector 204 that exists in an arbitrary emotional region.

Each node represents an emotional area to which each vector on the emotional space plane 200 can be mapped. The node where the arrow appears means the emotional area mapped to the second vector 202. On the other hand, the node where the arrow enters means the first output emotion area 205. Edge refers to the emotional area mapped to the third vector 203.

In other words, the graph means that the first output emotion area 205 can be moved between the nodes described above according to the direction of the edge arrow.

For a specific example, assume that the second vector 202 representing the robot's current emotion is mapped to the happiness area, and the fourth vector 204, which takes into account the maximum emotional change range that the robot can have, is mapped to the sadness area. do.

At this time, in order to determine the connection path between the second vector 202 and the fourth vector 204 as shown in FIG. 3, the connection path determination unit 140 sequentially changes the robot's emotions into happy, neutral, and sad regions. decide.

At this time, the connection path determination unit 140 determines the first output emotional area 205 that satisfies the sequential emotional change pattern of the robot, and determines the emotional area of the first output area 205.

As such, according to one embodiment, the apparatus 100 for determining the emotional state of a robot may generate the emotional change pattern of the robot using a human emotional change pattern or a pattern predefined by the user by applying a graph-based model.

Meanwhile, the results of the calculation between the second vector 202 and the third vector 203 are explained by illustrating a graph consisting of 5 nodes and 14 edges, but are not necessarily limited thereto, and may vary in emotional change patterns depending on the embodiment. may include.

FIG. 4 is an exemplary diagram showing a subsequent emotional state determined by the emotional state determining device 100 of a robot according to an embodiment of the present invention on the emotional space plane 200.

Meanwhile, the robot's personality is (0.5, 0.9, 0.5, 0.2), is 0.7, is assumed to be 0.8, and the preset interval (Q) is preset to 0.2.

As shown in FIG. 4, the emotional state determining device 100 of the robot converted 10 random sentences including sad emotions into the third vector 203, and then the emotional state determining device 100 of the robot converted the third vector 203 into the third vector 203. Starting with vector 201, 10 subsequent emotional states are determined. At this time, the 10 subsequent emotional states are determined sequentially from #1 to #10.

That is, even if the apparatus 100 for determining the emotional state of a robot according to one embodiment repeatedly receives the third vector 203 mapped to the same emotional region, the vector corresponding to the subsequent emotional state as shown in FIG. 4 It does not emit (206).

At this time, the result shown in FIG. 4 is a second vector 202 corresponding to the robot's current emotional state, a first vector 201 corresponding to the robot's personality, and a third vector corresponding to the emotional state of the input sentence ( 203) It can be said to be the result of the directional force acting.

FIG. 5 is a block diagram illustrating an apparatus 100 for determining the emotional state of a robot according to an additional embodiment.

Referring to FIG. 5 , the apparatus 100 for determining the emotional state of a robot according to an additional embodiment further includes a regression path determination unit 510.

According to one embodiment, the regression path determination unit 510 may be implemented using one or more physically separate devices, one or more processors, or a combination of one or more processors and software, and, unlike the example shown, Specific actions may not be clearly distinguished.

In the example shown in FIG. 5, the receiving unit 110, converting unit 120, calculating unit 130, connection path determining unit 140, and emotion determining unit 150 have the same configuration as shown in FIG. 1. , redundant explanation for this will be omitted.

When an input sentence is not input within a preset critical time interval, the regression path determination unit 510 determines a regression path based on the first vector 201 and the second vector 202.

Specifically, when the input sentence is not input within a preset critical time interval, the regression path determination unit 510 applies each of the first vector 201 and the second vector 202 based on a predefined node graph rule. A second output emotion area (not shown) is determined for the mapped emotion area, and a regression path is determined to pass through the second output emotion area (not shown).

At this time, the preset critical time interval may be a preset value based on the user's input, but the preset critical time interval may be changeable after the input sentence is entered.

According to one embodiment, the regression path determination unit 510 determines that the path length is the longest among a plurality of preliminary regression paths passing through the first vector 201, the second vector 202, and the second output emotion area (not shown). A short preliminary regression path can be determined as the regression path.

At this time, the return path determination unit 510 may determine a preliminary connection path with the shortest path length based on a predefined shortest path algorithm. For example, the return path determination unit 510 may determine a preliminary connection path with the shortest path length based on an algorithm that can determine a path that avoids or passes through a predetermined point, such as the A* algorithm. there is.

Figure 6 is an exemplary diagram showing the subsequent emotional state of a robot recovered by personality according to an embodiment.

At this time, robot A is a robot with a positive personality corresponding to the happiness domain, and robot B is a robot with a negative personality corresponding to the anger domain.

The #A-5 vector of robot A shown in FIG. 6 means the first vector 201 of robot A.

The #B-5 vector of robot B means the first vector (201) of robot B. #A-2 to #A-5 vectors refer to the sixth vector sequentially determined for robot A per critical time interval when an input sentence is not input within a preset threshold time interval for robot A.

Likewise, the illustrated vectors #B-2 to #B-5 refer to the sixth vector sequentially determined for robot B per threshold time interval when no input sentence is input within the preset threshold time interval for robot B. do.

As shown, when no input sentence is entered within the preset critical time interval, Robot A's emotion moves from #A-1 to #A-5, passing through the areas of anger, sadness, neutrality, and happiness. do.

Likewise, it shows that robot B's emotions have moved from #B-1 to #B-5, going through the neutral, sad, and angry areas.

That is, if no input sentence is input within the preset critical time interval, the emotions of robot A and robot B gradually revert to emotions corresponding to their personalities over time.

FIG. 7 is a flowchart illustrating a method for determining the emotional state of a robot according to an embodiment.

The method shown in FIG. 7 may be performed, for example, by the emotional state determination device 100 of the robot shown in FIG. 1 .

Referring to FIG. 7, the robot's emotional state determination device 100 receives a vector corresponding to the robot's personality as a first vector 201 on the emotional space plane 200 divided into a plurality of emotional regions (710). .

Afterwards, the robot's emotional state determination device 100 receives a vector corresponding to the robot's current emotional state as the second vector 202 on the emotional space plane 200 (720).

Afterwards, the emotional state determination device 100 of the robot converts the emotional state of the input sentence into the third vector 203 on the emotional space plane 200 using a previously learned model (730).

Thereafter, the robot's emotional state determination device 100 divides the fourth vector 204 on the emotional space plane 200 corresponding to the maximum emotional state of the robot by the input sentence into the first vector 201 and the second vector 202. ) and the third vector 203 (740).

Thereafter, the emotional state determination device 100 of the robot determines a connection path passing through the second vector 202 and the fourth vector 204 based on the result of the operation between the second vector 202 and the third vector 203. Do (750).

Thereafter, the robot emotional state determination device 100 determines the subsequent emotional state for the current emotional state based on the connection path (760).

Figure 8 is a flowchart for explaining a method for determining the emotional state of a robot according to an additional embodiment.

The method shown in FIG. 8 may be performed, for example, by the emotional state determination device 100 of the robot shown in FIG. 5 .

Referring to FIG. 8, the robot's emotional state determination device 100 receives a vector corresponding to the robot's personality as a first vector 201 on the emotional space plane 200 divided into a plurality of emotional regions (810). .

Thereafter, the robot's emotional state determination device 100 receives a vector corresponding to the robot's current emotional state as the second vector 202 on the emotional space plane 200 (820).

Afterwards, the emotional state determination device 100 of the robot determines whether the input sentence is received within a preset threshold time interval (830).

Afterwards, when the emotional state determining device 100 of the robot determines that the input sentence has been entered within a preset critical time interval, the emotional state determining device 100 of the robot evaluates the emotional state of the input sentence using the previously learned model. It is converted into a third vector 203 on the spatial plane 200 (840).

Thereafter, the robot's emotional state determination device 100 converts the fourth vector 204 on the emotional space plane 200 corresponding to the maximum emotional state of the robot changed by the input sentence from the current emotional state into the first vector 201. , is calculated based on the result of the operation between the second vector 202 and the third vector 203 (850).

Afterwards, the emotional state determination device 100 of the robot determines a connection path passing through the second vector 202 and the fourth vector 204 (860).

Thereafter, the robot emotional state determination device 100 determines the subsequent emotional state for the current emotional state based on the connection path (870).

Meanwhile, when the emotional state determining device 100 of the robot determines that the input sentence has not been input within the preset threshold time interval, the emotional state determining device 100 of the robot uses the first vector A return path passing through (201) and the second vector (202) is determined (880).

Thereafter, the robot emotional state determination device 100 determines the subsequent emotional state based on the regression path (890).

In FIGS. 7 and 8, the method is divided into a plurality of steps, but at least some of the steps are performed in a different order, combined with other steps, omitted, or divided into detailed steps. Alternatively, one or more steps not shown may be added and performed.

In FIGS. 7 and 8, the method is explained with reference to the flow chart shown in the drawings. For purposes of illustration, the method is shown and described as a series of blocks; however, the invention is not limited to the order of the blocks, and some blocks may occur simultaneously or in a different order than shown and described herein with other blocks. and various other branches, flow paths, and sequences of blocks may be implemented that achieve the same or similar results. Additionally, not all blocks shown may be required for implementation of the methods described herein.

Furthermore, the method according to an embodiment of the present invention may be implemented in the form of a computer program for performing a series of processes, and the computer program may be recorded on a computer-readable recording medium. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magneto-optical media such as floptical disks. optical media), and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, etc.

Although the above has been described with reference to the embodiments, those skilled in the art can make various modifications and changes to the present invention without departing from the spirit and scope of the present invention as set forth in the claims below. You will understand.

100: Robot emotional state determination device
110: receiving unit
120: conversion unit
130: Calculation unit
140: Connection path decision unit
150: Emotion decision unit
200: emotional space plane
201: first vector
202: Second vector
203: Third vector
204: Fourth vector
205: first output emotion area
206: Vectors corresponding to subsequent emotional states
510: Return path decision unit

Claims (18)

Receiving a vector corresponding to the personality of the robot as a first vector on an emotional space plane divided into a plurality of emotional regions;
Receiving a vector corresponding to the current emotional state of the robot as a second vector on the emotional space plane;
Converting the emotional state of the input sentence into a third vector on the emotional space plane using a previously learned model;
A fourth vector on the emotional space plane corresponding to the maximum emotional state of the robot changed by the input sentence in the current emotional state is calculated based on the result of the operation between the first vector, the second vector, and the third vector. calculating step;
determining a connection path passing through the second vector and the fourth vector;
Determining a subsequent emotional state to the current emotional state based on the connection path.
In claim 1,
The emotional space plane includes a pleasure axis and an arousal axis.
In claim 1,
In the calculating step, the fourth vector is calculated using Equation 1 below.
[Equation 1]

(At this time, is the fourth vector, is the second vector, is the third vector, is the first vector, and is a weight variable with random real values)
In claim 1,
Determining the connection path may include determining a first output emotional region for the emotional region mapped to each of the second vector and the third vector based on a predefined node graph rule; and
Determining the emotional state of the robot, comprising determining a preliminary connection path with the shortest path length among a plurality of preliminary connection paths passing through the second vector, the fourth vector, and the first output emotion area as the connection path. method.
In claim 1,
The step of determining the subsequent emotional state is:
determining any one of a plurality of vectors on the connection path; and
A method for determining an emotional state of a robot, comprising determining the subsequent emotional state based on an emotional region mapped to the arbitrary one of the plurality of emotional regions.
In claim 1,
The step of determining the subsequent emotional state is:
A method for determining an emotional state of a robot, wherein an emotional region mapped to a vector separated by a preset interval from the second vector along the connection path is determined as the subsequent emotional state.
In claim 1,
If the input sentence is not input within a preset threshold time interval, determining a regression path based on the first vector and the second vector; and
The method of determining the emotional state of a robot further comprising determining the subsequent emotional state based on the regression path.
In claim 7,
The step of determining the regression path is,
determining a second output emotional region for the emotional region mapped to each of the first vector and the second vector based on a predefined node graph rule; and
Determining the emotional state of the robot, comprising determining a preliminary regression path with the shortest path length among a plurality of preliminary regression paths passing through the first vector, the second vector, and the second output emotional area as the regression path. method.
In claim 7,
The step of determining the subsequent emotional state is:
A method for determining an emotional state of a robot, wherein an emotional region mapped to a vector separated by a preset interval from the second vector along the regression path is determined as the subsequent emotional state.
a receiving unit that receives a vector corresponding to the robot's personality as a first vector on an emotional space plane divided into a plurality of emotional regions, and receives a second vector on an emotional space plane corresponding to a current emotional state of the robot;
a conversion unit that converts the emotional state of the input sentence into a third vector on the emotional space plane using a previously learned model;
A fourth vector on the emotional space plane corresponding to the maximum emotional state of the robot changed by the input sentence from the current emotional state is calculated based on the result of the operation between the first vector, the second vector, and the third vector. A calculation unit that calculates;
a connection path determination unit that determines a connection path passing through the second vector and the fourth vector;
An emotional state determination device for a robot, comprising an emotion determination unit that determines a subsequent emotional state for the current emotional state based on the connection path.
In claim 10,
The emotional space plane includes a pleasure axis and an arousal axis.
In claim 10,
The emotional state determination device of a robot, wherein the calculation unit calculates the fourth vector using Equation 1 below.
[Equation 1]

(At this time, is the fourth vector, is the second vector, is the third vector, is the first vector, and is a weight variable with random real values)
In claim 10,
The connection path determination unit determines a first output emotion region for the emotion region mapped to each of the second vector and the third vector based on a predefined node graph rule, and the second vector and the fourth vector. and determining the connection path as the preliminary connection path with the shortest path length among the plurality of preliminary connection paths passing through the first output emotion area.
In claim 10,
The appraisal decision unit,
Determining an arbitrary one of the plurality of vectors on the connection path, and determining the subsequent emotional state based on an emotional region mapped to the arbitrary one of the plurality of emotional regions.
In claim 10,
The appraisal decision unit,
An emotional state determination device for a robot, which determines as the subsequent emotional state an emotional region mapped to a vector separated by a preset interval from the second vector along the connection path.
In claim 10,
If the input sentence is not input within a preset threshold time interval, the device further includes a regression path determination unit that determines a regression path based on the first vector and the second vector.
In claim 16,
The return path determination unit,
Based on a predefined node graph rule, a second output emotion area is determined for the emotion area mapped to each of the first vector and the second vector, and the first vector, the second vector, and the second output emotion are determined. An emotional state determination device for a robot, wherein a preliminary regression path with the shortest path length among a plurality of preliminary regression paths passing through an area is determined as the regression path.
In claim 16,
The appraisal decision unit,
An emotional state determination device for a robot that determines, as the subsequent emotional state, an emotional region mapped to a vector separated by a preset interval from the second vector along the regression path.