Robot bionic character establishing and cultivating method and expression robot

Abstract

The invention relates to a robot bionic character establishing and cultivating method and an expression robot. The method comprises the steps that a character model, an age model and an environment model are established, and the emotional state of the robot is initialized; external environment input is received and analyzed, and short-term emotional tendency and long-term emotional tendency of a target object are extracted; acquiring current age information of the robot from the age model; updating a short-term emotion vector of the robot according to the short-term emotion tendency and the current age information, and updating a character vector of the robot according to the long-term emotion tendency, the environment vector and the current age information; and circulating the above steps, and after a preset time, outputting a final character to form a report so as to simulate the growth process of the child. The user can establish a character model, an age model and an environment model, and then the expression robot can update the character model in real time through factors such as language communication and behavior actions of parents to the child / robot, the family environment where the child is located and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence, and in particular to a method for establishing and cultivating a robot bionic personality and an expression robot. Background Art

[0002] With the continuous development of artificial intelligence, the primary task of the new generation of artificial intelligence systems is to have the ability to connect with "emotions", so that they can meet people's general psychological and emotional needs in a more real way, thereby gradually building trust and dependence. In other words, artificial intelligence should not be just an ordinary intellectual tool, and the development path of artificial intelligence should not blindly follow the rational route, but should fall into the intersection of sensibility and rationality.

[0003] Emotional computing will play a vital role in the development of artificial intelligence. We have also seen more and more products with "emotions", but this is just the beginning. Faced with the extremely complex issue of human emotions, artificial intelligence still has a long way to go.

[0004] In the prior art, a method, medium and system for a chat robot with simulated emotions, with publication number CN117808011B, discloses initializing the initial emotional state of the chat robot based on a large language model; receiving user input and performing semantic analysis to extract short-term topic text and short-term emotional tendency text; performing semantic analysis based on all user input to extract long-term emotional tendency text; updating the robot's short-term emotional vector; fusing the short-term emotional vector with the long-term emotional vector as the answer emotional vector; obtaining the chat robot's output text based on the user's question, and using the text emotion adjustment model to adjust according to the answer emotional vector to obtain the output text with emotion, and output it to the user; repeating the above steps; thereby facilitating the robot to establish an emotional connection with the user and improve the user experience. However, the robot is a process of suggesting emotions during the communication process, not a process in which the robot imitates a certain personality proportion of the person himself and simulates the change of personality proportion under the influence of the environment.

[0005] In particular, in today's era, parents pay more and more attention to the character development of their children. However, many parents do not know how to get along with their children and what to do to be more conducive to their children's growth. However, existing robots cannot meet the needs of users. Therefore, the solution of this application designs a robot that can simulate the character formation process of children during their growth, in the hope that it can help parents solve the problems they are worried about and allow more children to grow up physically and mentally healthy. Summary of the invention

[0006] The first purpose of the present invention is to provide a method for establishing and cultivating a robot bionic personality, which aims to solve the technical problem that existing expression robots cannot simulate the growth process of children.

[0007] In order to solve the above technical problems, a method for establishing and cultivating a robot bionic personality is provided, comprising:

[0008] Establishing a model, establishing a personality model, an age model and an environment model, wherein the personality model includes a personality vector and a short-term emotion vector, and the environment model includes an environment vector, wherein the personality vector is set manually, and the personality vector and the short-term emotion vector are both vectors including a plurality of emotion indicators, each element of the vector represents a weight of a corresponding emotion indicator, and a value range of each element in the personality vector is greater than a value range of each element in the short-term emotion vector;

[0009] Initialize emotional state, initialize the robot's emotional state;

[0010] Acquire emotional tendencies, receive and analyze external environmental input, and extract the short-term and long-term emotional tendencies of the target object;

[0011] Get the current age information of the robot from the age model;

[0012] Updating the personality model, updating the short-term emotion vector of the robot according to the short-term emotion tendency and the current age information, and updating the personality vector of the robot according to the long-term emotion tendency, the environment vector and the current age information;

[0013] Outputting emotion expression features, fusing the short-term emotion vector with the personality vector, and controlling the robot to output the emotion expression features;

[0014] The steps of initializing the emotional state to outputting the emotional expression characteristics are repeated, and after a preset time, a final personality formation report is output.

[0015] Furthermore, the receiving and analyzing of external environment input includes: preprocessing the user's input text to extract valid information, using natural language processing technology NLP to perform in-depth semantic analysis on the user input, extracting short-term topic content and short-term emotional tendencies, and combining the user's historical input to obtain the long-term emotional tendencies.

[0016] Furthermore, the receiving and analyzing of external environment input also includes: obtaining environmental change factors through user input or actively, and forming corresponding long-term emotional tendencies through different environmental change factors.

[0017] Furthermore, the environmental change factors include a major change in the family environment and / or the sudden death of at least one of the users. When the family environment undergoes a major change, a first influencing factor is formed, and when at least one of the users suddenly dies, a second influencing factor is formed. The personality vector is adjusted according to the first influencing factor and the second influencing factor, and the influence of the second influencing factor on the personality vector is greater than the influence of the first influencing factor on the personality vector.

[0018] Furthermore, the receiving of external environment input and analysis also includes: when the user's behavior conforms to a special behavior event, extracting the target object's short-term emotional tendency and long-term emotional tendency, and the number of times a special behavior event is formed is positively correlated with the values ​​of the short-term emotional tendency and long-term emotional tendency extracted at that time.

[0019] Furthermore, the robot includes a free mode and a guided mode. The free mode includes setting an initial personality model and an initial age model. In the free mode, the user can view / export the robot's current age information, current personality model and education advice report at any time. The guided mode includes setting an initial personality model, an age model and a target personality model. In the guided mode, when the user's behavior is not conducive to the transformation from the initial personality model to the target personality model, the robot is controlled to send a warning message to the user.

[0020] Furthermore, an emotion vector T is constructed to describe the personality model of the expression robot. The specific formula is as follows:

[0021] T=[t1,t2,t3,...,t n ] (1);

[0022]

[0023] Among them, t n represents the value of the ith emotion feature. There are n emotion features in total. The emotion vectors include short-term emotion vectors T s , character vector T L , environmental change factors T D .

[0024] Furthermore, the short-term sentiment vector Short-term sentiment vector Environmental change factors

[0025]

[0026] Furthermore, the change ΔT of the short-term emotion vector is obtained by using the following convolutional neural network or recurrent neural network model: S , the change of environmental change factors ΔT D, the formula is as follows:

[0027] S pos =f pos (C), S neg =f neg (C) (3);

[0028] ΔT S =g(S pos , S neg ) (4);

[0029] D pos =f pos (C),D neg =f neg (C) (5);

[0030] ΔT D =g(S pos , S neg ) (6);

[0031] Among them, g is a function that maps the sentiment score to the change of the emotion vector;

[0032] Updated short-term sentiment vector T s′ =T s +ΔT S +ΔT D (7);

[0033] Updated personality vector T L′ =ωT L +(1-ω)(T s +αΔT S +βΔT D );

[0034] Among them, ω is a weight parameter, the specific value of ω is determined according to the specific situation, the value of β depends on the first influencing factor, the second influencing factor and the current age information of the robot, and the value of α depends on the special behavior event and the current age information of the robot.

[0035] The second object of the present invention is to provide an expression robot, including the establishment and cultivation method of the above-mentioned robot bionic personality.

[0036] Implementing the embodiments of the present invention will have the following beneficial effects:

[0037] In the method for establishing and cultivating the robot's bionic personality in this embodiment, the user (parent) can establish a personality model, an age model and an environment model, and then the expression robot can update the personality model in real time through factors such as the parents' language communication and behavioral actions to the children, the parents' language communication and behavioral actions to the expression robot, and the child's family environment, surrounding environment, social environment, etc. In addition, the expression robot can also obtain the user's short-term emotional tendencies, and can make emotional language expressions or body language expressions during communication. In other words, the expression robot of this application is no longer a stereotyped response. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 The present invention is a flowchart of a method for establishing and developing a robot bionic personality according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. The preferred embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present invention more thorough and comprehensive.

[0041] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0043] The purpose of designing the expression robot of the present application is to simulate the growth process of infants and children, and to help parents understand how to accompany their children's growth. Especially for some new parents, they don't know what to do to make their children grow up as their parents hope, and the development of character is also very important for children. Therefore, the scheme of the present application designs a method that can simulate the character formation process of children during their growth, in the hope that it can help the parents of children solve the problems they are worried about, so that more children can grow up physically and mentally healthy.

[0044] Please refer to Figure 1 The embodiment of the present invention provides a method for establishing and developing a robot bionic personality, which is applied to the robot and includes:

[0045] Establish a model, establish a personality model, an age model and an environment model, the personality model includes a personality vector and a short-term emotion vector, and the environment model includes an environment vector, wherein the personality vector is set manually, and the personality vector and the short-term emotion vector are both vectors including multiple emotion indicators, each element of the vector represents the weight of the corresponding emotion indicator, and the value range of each element in the personality vector is greater than the value range of each element in the short-term emotion vector; illustratively, establishment refers to the internal confirmation of a fixed personality, and cultivation refers to the process of strengthening the formation of the personality development direction defined by the model rules under the influence of various parameter indicators of external information, which is external and is affected by external information, environmental factors, social interaction and other factors. These external factors may have a shaping effect on the personality of an individual, so that the individual gradually forms or strengthens certain personality traits in the interaction with the outside world. Developing personality is a dynamic process involving the interaction and adaptation between the individual and the external environment.

[0046] In the model building stage, users (referring to parents) can build a personality model based on their own children as reference objects. Of course, users can also set up a new personality model (not based on any object as reference), such as couples who do not have children yet and want to adapt to or cultivate the experience of educating children in advance. Users can build an environmental model based on their family's living environment and surrounding environment. The impact of the environment on the growth of children is also very important. The indicators of the age model include initial age, end age, and age change rate. Children of different ages have different factors such as acceptance, perception, performance, and learning ability. Therefore, setting the age model is to make the robot's simulation closer to reality. Setting the initial age refers to when the robot simulates human age at the beginning of the simulation. At this time, the various ability indicators possessed by the robot should be consistent with the current state of the simulated age, so that the simulation results of the expression robot are closer to reality. The end age refers to when the robot needs to simulate human age. In addition, it should be noted that the various ability indicators of the current robot can be retrieved from the pre-trained big data model. The speed of age change determines the speed of the simulation process. It can be understood that the speed of age change is equal to the simulated age duration divided by the simulation time, where the simulated age duration is equal to the end age minus the initial age, and the simulation time refers to the time required from the initial age simulation to the end age. In addition, you can also set the gender model, boy or girl, different genders will also have different ability indicators.

[0047] The personality vector and short-term emotion vector include multiple emotion indicators, such as, but not limited to, happiness, sadness, surprise, anger, fear, relaxation, tension, excitement, boredom, etc. In the constructed personality model, each emotion indicator has a weight value, which indicates the intensity of the emotion. When the user first starts to build a personality model, the weight of each emotion indicator is manually set. The long-term emotion vector is the same as the short-term emotion vector at the time of initialization, but the value range of each element of the long-term emotion vector is larger than that of the short-term emotion vector to reflect the stability of long-term emotions.

[0048] Initialize emotional state, initialize the robot's emotional state;

[0049] The initialization process can perform statistical analysis on the conversation history corpus and use sentiment analysis tools such as TextBlob or VADER to assign an initial weight to each sentiment dimension.

[0050] Acquire emotional tendencies, receive and analyze external environmental input, and extract the short-term and long-term emotional tendencies of the target object;

[0051] Exemplarily, the external environment input includes the user's language, the user's behavior, and changes in the external environment. The short-term and long-term emotional tendencies of the target object (i.e., the parents) are comprehensively judged through the user's spoken text, tone, body movements, and changes in the external environment.

[0052] Get the current age information of the robot from the age model;

[0053] Update the personality model, update the robot's short-term emotion vector according to the short-term emotion tendency and current age information, and update the robot's personality vector according to the long-term emotion tendency, environment vector and current age information;

[0054] Output emotional expression features, fuse short-term emotional vectors with personality vectors, and control the robot to output emotional expression features; illustratively, outputting emotional expression features means that the expression robot fuses short-term emotional vectors with personality vectors and controls the robot to output expressions, language and body movements with emotional characteristics. This process simulates the way humans express emotions, where the short-term emotional vector reflects the emotional state of the robot at a specific moment, and the personality vector represents the robot's long-term personality tendency. Specifically, after the external environment input is received and analyzed, the robot extracts the short-term emotional tendency and long-term emotional tendency of the target object, and then updates the short-term emotional vector and personality vector based on this information and the current age information. The updated vectors will be used to control the robot's output, including facial expressions, language expressions and body movements, so that the robot can express emotions and personality in a more human-like way. For example, if a user expresses anger to the robot, the robot will analyze this emotion and decide how to respond based on its personality model. If the robot's personality model tends to avoid conflict, it may choose to respond with soothing language and body language to avoid further conflict. Conversely, if the robot’s personality model tends to be directly confrontational, it may use stronger language and body language to express its stance.

[0055] The steps of initializing the emotional state to outputting the emotional expression characteristics are repeated, and after a preset time, the final personality formation report is output. Exemplarily, the personality formation report can be exported not only by the user but also by the robot manufacturer. For example, the personality formation report can be exported according to regional divisions, which is conducive to research experts to study the personality trends, local customs and other aspects of each region, so as to hope that the robot of this application can contribute to the research of social humanities.

[0056] In the method for establishing and cultivating the robot's bionic personality in this embodiment, the user (parent) can establish a personality model, an age model and an environment model, and then the expression robot can update the personality model in real time through factors such as the parents' language communication and behavioral actions to the children, the parents' language communication and behavioral actions to the expression robot, and the child's family environment, surrounding environment, social environment, etc. In addition, the expression robot can also obtain the user's short-term emotional tendencies, and can make emotional language expressions or body language expressions during communication. In other words, the expression robot of this application is no longer a stereotyped response.

[0057] The expression robot of the present application can be applied to different scenarios. The first aspect is the simulation of the growth process of children mentioned above. The second aspect is that the expression robot can also realize the function of companionship and is suitable for accompanying different objects. The expression robot has the characteristics of user-defined personality model, which makes it show different application scenarios and functions when accompanying different users. For example, when facing the elderly, because the elderly often face loneliness, the expression robot can provide emotional support by simulating emotional expressions such as care and comfort by establishing an emotional model. Health monitoring can also be carried out. By setting a health model, the robot can remind the elderly to take medicine on time, do physical exercise, and notify family members or doctors in time in abnormal situations. When facing children, parents hope that their children can form good personality traits for their growth process. The expression robot can guide children to develop good habits and morals through language and behavior according to the personality model set by parents. Parents can set a safety model to ensure that the robot will sound an alarm and take protective measures when the child is in a dangerous environment. For parents, parents can get parenting advice through the robot and understand how to better educate and cultivate their children's personality characteristics. Parents can share parenting experience, relieve parenting pressure, and get emotional support through communication with the robot. For partners, the expression robot of the present application can also play an important role. For example, partners can set each other's personality models to allow the robot to recognize and understand each other's emotional state, thereby providing corresponding emotional support and comfort. For example, when a person feels frustrated, the robot can play soothing music or provide encouraging words. Partners can set important dates (such as birthdays, anniversaries, etc.) and let the robot remind them in advance and plan special celebrations. In this way, the robot can help partners maintain romance and surprises, and enhance the depth and persistence of feelings. Partners can talk about their inner troubles and emotional problems through communication with the robot. The robot can provide professional psychological counseling and support based on the set personality model to help partners relieve psychological pressure and promote mental health. Partners can use the robot to analyze the relationship status of both parties (such as communication frequency, emotional satisfaction, etc.), and put forward suggestions and methods to improve the relationship. In this way, the robot can help partners discover and solve problems in the relationship in a timely manner and maintain a harmonious and stable emotional life.

[0058] In addition, it should be noted that the personality model set by the user at the beginning is not unchangeable. The user may not know what kind of personality model he needs at the beginning. Therefore, the robot of this application allows the personality model to be adjusted during use. The first way is that the user manually inputs the personality model. The second way is that the robot can automatically adjust it. The basis for automatic adjustment is that the robot can obtain the personality model information that the user actually wants from the user's dialogue, body movements, expressions, etc.

[0059] In one possible implementation, receiving external environment input and analyzing it includes: preprocessing the user's input text to extract valid information, using natural language processing technology NLP to perform in-depth semantic analysis on the user input to extract short-term topic content and short-term emotional tendencies, and combining the user's historical input to obtain long-term emotional tendencies.

[0060] When users input in language, they need to preprocess the input text, including Chinese word segmentation, stop word filtering, part-of-speech tagging, etc., in order to extract effective information. Natural language processing (NLP) technology, such as BERT or GPT model, is used to conduct in-depth semantic analysis of user input to extract the topic and short-term emotional tendency of the user's speech. This step also requires analyzing the user's historical input to extract long-term emotional tendencies. Time series analysis technology, such as long short-term memory network (LSTM), can be introduced here to track the changing trend of user emotions. In addition, unsupervised learning methods, such as topic model LDA, can be used to identify long-term topic tendencies in user discourse.

[0061] In a possible implementation, receiving external environment input and analyzing it further includes: obtaining environmental change factors through user input or actively, and forming corresponding long-term emotional tendencies through different environmental change factors. For example, the user sets an initial environmental model before the robot starts simulation. If the home environment or living environment changes during the simulation, the user can actively input the changes in the current environment. Of course, the robot can also update and confirm the surrounding environment in real time from the user's chat content or the robot actively observes the surrounding environment.

[0062] In one possible implementation, environmental change factors include a major change in the family environment and / or the sudden death of at least one of the users. When the family environment undergoes a major change, a first influencing factor is formed, and when at least one of the users dies suddenly, a second influencing factor is formed. The personality vector is adjusted based on the first influencing factor and the second influencing factor, and the impact of the second influencing factor on the personality vector is greater than the impact of the first influencing factor on the personality vector.

[0063] In a possible implementation, receiving external environment input and analyzing it also includes: when the user's behavior conforms to a special behavior event, extracting the short-term emotional tendency and long-term emotional tendency of the target object, and the number of special behavior events formed is positively correlated with the values ​​of the short-term emotional tendency and long-term emotional tendency extracted at that time. Exemplarily, the special behavior events here refer to some behaviors that will have a significant impact on the physical and mental health of children, such as beating, scolding, corporal punishment, and even abuse of children, which will have a significant impact on the development of children's later personality. In addition, it should be noted that since the expression robot of the present application simulates the growth process of children in the family, then the communication and interaction between parents and children in the family, including the beating, scolding, and abuse of children mentioned above, the expression robot will synchronize these behavior events to its own personality model, that is, not only the user's behavior on the expression robot will affect the personality model of the expression robot, but the user's behavior on the child will also affect the personality model of the expression robot, which is conducive to better simulating the real growth environment and further improving the authenticity and reliability of the simulation. In addition, it should be noted that the more the number of special behavior events, the greater the impact on the personality model, that is, the number of special behavior events will have a superimposed effect. For example, when a special behavior event occurs for the first time, the impact on the robot's personality model is 0.1, when a special behavior event occurs for the second time, the impact on the robot's personality model may become 0.15, and when a special behavior event occurs for the third time, the impact on the robot's personality model may become 0.18. Of course, this is just an example, and the specific impact parameter value needs to be set according to the actual situation.

[0064] In one possible implementation, the robot includes a free mode and a guided mode. The free mode includes setting an initial personality model and an initial age model. In the free mode, the user can view / export the robot's current age information, current personality model, and education advice report at any time. The guided mode includes setting an initial personality model, an age model, and a target personality model. In the guided mode, when the user's behavior is not conducive to the transformation from the initial personality model to the target personality model, the robot is controlled to send a warning message to the user.

[0065] For example, in free mode, the user only needs to set the initial personality model and the initial age model before the simulation begins, and then let it grow freely. Of course, during the simulation, the user can view or export the information of the robot's current age stage and the current personality model at any time, and the user can obtain the robot's growth status information in time to make corresponding adjustments. The educational suggestion report can include the current robot's age stage, current personality model information, and how it develops from the initial personality model to the current personality model, and record the events or factors that have a greater impact on it for the user to view. It can also give educational suggestions based on the simulation situation to tell the user how to better cultivate the child. In guided mode, before the robot simulation, the user sets the initial personality model and the target personality model at the beginning. For example, if the current user's child is 6 years old, the user can set the initial personality model with reference to his own child, and at the same time, set the target personality model to what kind of person the parents hope their child will grow up to be when he is 18 years old. At this time, during the simulation process, if the parents' language expression, body behavior, etc. are not conducive to the robot forming the target personality model, the robot will issue a warning message, including a red light, a sound alarm prompt, or directly give educational advice to the parents, informing them that a certain behavior or speech is not conducive to the child's growth, or suggesting what the parents should do at this time. In other words, in the guidance mode, the user can use the expression robot of this application to know which behaviors can be done and which behaviors cannot be done if they want their children to grow into people with the target personality model. In addition, it should be noted that in the guidance mode, the user can also view / export the robot's current age information, current personality model, and educational advice report at any time.

[0066] In addition, it should be noted that the longer a dominant emotional characteristic in a personality model is formed, the more difficult it is to change. For example, if one of the emotions in a personality model, such as responsibility, has a low value, it is difficult to improve it all at once.

[0067] The method of the present application constructs an emotion vector T to describe the personality model of the expression robot. The specific formula is as follows:

[0068] T=[t1,t2,t3,...,t n ] (1);

[0069] In formula (1), t n represents the value of the i-th emotional feature, such as "happy", "sad" and other emotional features, and there are n emotional features in total. Among them, the emotional vector includes the short-term emotional vector T s , the long-term sentiment vector T L , environmental change factors T DThe function T in formula (1) indicates that the current personality model of the expression robot includes n emotional features, including happiness, sadness, relaxation and other emotional features.

[0070]

[0071] The function T in formula (2) represents the current personality model of the expression robot, which is established and considered from three dimensions, including the short-term emotion vector T s , the long-term sentiment vector T L , environmental change factors T D Compared with the solution recorded in the patent document with publication number CN117808011B, the technical solution of the present application takes into account environmental change factors, which can make the result orientation of personality simulation more accurate. At the same time, in terms of robot receiving information, the expression robot of the present application can not only receive parents' dialogue and behavior information to the robot itself, but also receive parents' dialogue and behavior information to their children.

[0072] Among them, the emotion vector T = t1, t2, t3, ..., t n The specific emotions for each emotion feature in are as follows:

[0073] t1: represents the weight of the emotional feature "happy";

[0074] t2: represents the weight of the emotional feature "sadness";

[0075] t3: represents the weight of the emotional feature of "relaxation";

[0076] t4: represents the weight of the emotional feature of "tension";

[0077] t5: represents the weight of the emotional feature "excitement";

[0078] t6: represents the weight of the emotional feature of "boredom";......

[0079] t n : represents the weight of the emotional feature "anger";

[0080] The above emotional features can be customized, or the most frequent emotional words can be obtained by analyzing large-scale corpus, and the corresponding emotional features are selected to construct vector T. The value range of each element is [-1, +1], which represents the emotional feature t i The degree and intensity of positive and negative.

[0081] Initialize the short-term sentiment vector T s , character vector T L , environmental change factors T D The value of , where the value range of each dimension is recommended to be set to:

[0082] T s =T L =T D =T0

[0083] Wherein, T0 represents the preset initial short-term emotion vector value, initial personality vector value and initial environmental change factor.

[0084] Using the following convolutional neural network or recurrent neural network model, the change of short-term emotion vector ΔT is obtained S , the change of environmental change factors ΔT D , the formula is as follows:

[0085] S pos =f pos (C), S neg =f neg (C) (3);

[0086] ΔT S =g(S pos , S neg ) (4);

[0087] D pos =f pos (C),D neg =f neg (C) (5);

[0088] ΔT D =g(S pos , S neg ) (6);

[0089] Among them, g is a function that maps the emotion score to the change of the emotion vector. Function f is mentioned as a function that maps the emotion score to the change of the emotion vector. Specifically, this function is used in the document to describe how to convert the numerical value (score) of the emotion into the change of the emotion vector, and then update the robot's short-term emotion vector and personality vector. Function f involves complex algorithms and models, such as convolutional neural networks (CNN) or recurrent neural networks (RNN), which can process and learn complex relationships between emotion data and predict emotion changes.

[0090] Updated short-term sentiment vector T s′ =T s +ΔT S +ΔT D (7);

[0091] Updated personality vector T L′ =ωT L +(1-ω)(T s+αΔT S +βΔT D ); (8)

[0092] Among them, ω is a weight parameter. Among them, the specific value of ω is determined according to the specific situation, among which, the value of β depends on the first influencing factor, the second influencing factor and the current age state simulated by the expression robot, and the value of α depends on the special behavior event and the current age state simulated by the expression robot, and the more special behavior events there are, the larger the value of α. In addition, it should be noted that the value of β depends on the degree of change in the family environment and living environment. In particular, the impact of changes in the family environment on the healthy growth of children is related. It is understandable that the greater the impact, the greater the value of β. It is understandable that at different ages, facing the same behavior, the α value and β value are also different. For example, facing the special behavior event of parents beating and scolding their children, the older the age, the stronger their acceptance ability, and the impact on the personality model is relatively small. The specific α value and β value are collected from the big data model library to determine their specific values.

[0093] After the above update, the emotion vectors are fused and used as the final emotion vector of the reply sentence. is a fusion function, such as concatenation or weighting;

[0094]

[0095] in, is a fusion function, such as concatenation or weighting.

[0096] The second object of the present invention is to provide an expression robot, including the above-mentioned robot bionic personality establishment and cultivation method. Exemplarily, the expression robot includes a user interface (UI), an artificial intelligence engine, an environment simulation module, a database and an output module.

[0097] The user interface (UI) allows the user to input personality models, age models, and environmental models;

[0098] The artificial intelligence engine is responsible for processing the data input by the user and simulating the character's personality development process according to the preset algorithm.

[0099] The environmental simulation module influences the robot's character development based on the initial age and initial character model set by the user, combined with actual family atmosphere, social environment and other factors.

[0100] The database stores personality trait data at different age stages, as well as possible personality change patterns under various circumstances.

[0101] The output module displays the robot's current personality status, including facial expressions, language behaviors, etc.

[0102] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for establishing and developing a robot bionic personality, applied to a robot, characterized in that: include: Establishing a model, establishing a personality model, an age model and an environment model, wherein the personality model includes a personality vector and a short-term emotion vector, and the environment model includes an environment vector, wherein the personality vector is set manually, and the personality vector and the short-term emotion vector are both vectors including a plurality of emotion indicators, each element of the vector represents a weight of a corresponding emotion indicator, and a value range of each element in the personality vector is greater than a value range of each element in the short-term emotion vector; Initialize emotional state, initialize the robot's emotional state; Acquire emotional tendencies, receive and analyze external environmental input, and extract the short-term and long-term emotional tendencies of the target object; Get the current age information of the robot from the age model; Updating the personality model, updating the short-term emotion vector of the robot according to the short-term emotion tendency and the current age information, and updating the personality vector of the robot according to the long-term emotion tendency, the environment vector and the current age information; Outputting emotion expression features, fusing the short-term emotion vector with the personality vector, and controlling the robot to output the emotion expression features; The steps of initializing the emotional state to outputting the emotional expression characteristics are repeated, and after a preset time, a final personality formation report is output.

2. The method for establishing and developing a robot bionic personality according to claim 1, characterized in that: The receiving and analyzing of external environment input includes: preprocessing the user's input text to extract effective information, using natural language processing technology NLP to perform in-depth semantic analysis on the user input, extracting short-term topic content and short-term emotional tendencies, and combining the user's historical input to obtain the long-term emotional tendencies.

3. The method for establishing and developing a robot bionic personality according to claim 2, characterized in that: The receiving and analyzing the external environment input also includes: obtaining environmental change factors through user input or actively, and forming corresponding long-term emotional tendencies through different environmental change factors.

4. The method for establishing and developing a robot bionic personality according to claim 3, characterized in that: The environmental change factors include a major change in the family environment and / or the sudden death of at least one of the users. When the family environment undergoes a major change, a first influencing factor is formed. When at least one of the users dies suddenly, a second influencing factor is formed. The personality vector is adjusted according to the first influencing factor and the second influencing factor, and the influence of the second influencing factor on the personality vector is greater than the influence of the first influencing factor on the personality vector.

5. The method for establishing and developing a robot bionic personality according to claim 3, characterized in that: The receiving of external environment input and analysis also includes: when the user's behavior conforms to a special behavior event, extracting the short-term emotional tendency and long-term emotional tendency of the target object, and the number of times the special behavior event is formed is positively correlated with the values ​​of the short-term emotional tendency and long-term emotional tendency extracted at that time.

6. The method for establishing and developing a robot bionic personality according to claim 1, characterized in that: The robot includes a free mode and a guided mode. The free mode includes setting an initial personality model and an initial age model. In the free mode, the user can view / export the robot's current age information, current personality model, and education suggestion report at any time. The guidance mode includes setting an initial personality model, an age model and a target personality model. In the guidance mode, when the user's behavior is not conducive to the transformation from the initial personality model to the target personality model, the robot is controlled to send a warning message to the user.

7. The method for establishing and developing a robot bionic personality according to claim 1, characterized in that: Construct an emotion vector T to describe the personality model of the expression robot. The specific formula is as follows: T=[t1,t2,t3,…,t n ] (1); Among them, t n represents the value of the ith emotion feature. There are n emotion features in total. The emotion vectors include short-term emotion vectors T s , character vector T L , environmental change factors T D .

8. The method for establishing and developing a robot bionic personality according to claim 7, characterized in that: Short-term sentiment vector Short-term sentiment vector Environmental change factors 9. The method for establishing and developing a robot bionic personality according to claim 8, characterized in that: Using the following convolutional neural network or recurrent neural network model, the change of short-term emotion vector ΔT is obtained S , the change of environmental change factors ΔT D , the formula is as follows: S pos =f pos (C),S neg =f neg (C) (3); ΔT S =g(S pos ,S neg ) (4); D pos =f pos (C),D neg =f neg (C) (5); ΔT D =g(S pos ,S neg ) (6); Among them, g is a function that maps the sentiment score to the change of the emotion vector; Updated short-term sentiment vector T s′ =T s +ΔT S +ΔT D (7); Updated personality vector T L′ =ωT L +(1-ω)(T s +αΔT S +βΔT D ); Among them, ω is a weight parameter, the specific value of ω is determined according to the specific situation, the value of β depends on the first influencing factor, the second influencing factor and the current age information of the robot, and the value of α depends on the special behavior event and the current age information of the robot.

10. An expression robot, characterized in that: The invention comprises a method for establishing and developing a robot bionic personality as described in any one of claims 1 to 9.

Images