Emotion output method and device and robot

Abstract

The invention relates to an emotion output method and device and a robot. The method comprises the following steps: acquiring scene information in real time; the scene information comprises ontology information, environment information and interaction information; performing emotion adjustment in response to the change of the scene information, and obtaining an emotion change vector; wherein the reference coordinate system of the emotion change vector is an emotion coordinate system constructed based on a two-dimensional emotion model; and obtaining a current emotion coordinate according to the historical emotion coordinate and the emotion change vector before the scene information changes, and outputting an emotion expression corresponding to the current emotion coordinate. The emotion change is quantified based on the two-dimensional emotion model, fine adjustment of emotion expression is achieved by outputting the emotion change vector, direct emotion jumping is not needed any more, the robot is guided to convert emotions more naturally, and therefore the man-machine interaction experience is improved.

Description

Technical Field

[0001] The present application relates to the field of robot control technology, and in particular to an emotion output method, device and robot. Background Art

[0002] Large models refer to machine learning models with large-scale parameters and complex computing structures. These models are usually built from deep neural networks and have billions or even hundreds of billions of parameters. Large models are designed to improve the expressiveness and predictive performance of the model and to handle more complex tasks and data. Therefore, relying on the development of large model technology, the human-computer interaction model is also undergoing revolutionary changes.

[0003] In traditional technology, the large model senses the current scene and makes the robot output corresponding emotions for underlying control. For example, when the user praises the robot, the robot shows happy emotions; when the user criticizes the robot, the robot shows frustrated emotions.

[0004] However, due to scene changes and large model misjudgments, the robot may jump between different emotions, such as quickly switching between happiness and worry, resulting in a poor user experience. Summary of the invention

[0005] Based on this, it is necessary to provide an emotion output method, device and robot with stable and natural emotion changes to address the above technical problems.

[0006] In a first aspect, the present application provides an emotion output method. The method comprises:

[0007] Obtain scene information in real time; scene information includes entity information, environmental information and interaction information;

[0008] In response to the change of scene information, the emotion is adjusted to obtain an emotion change vector; wherein the reference coordinate system of the emotion change vector is an emotion coordinate system constructed based on a two-dimensional emotion model;

[0009] The current emotion coordinates are obtained according to the historical emotion coordinates and emotion change vector before the scene information changes, and the emotion expression corresponding to the current emotion coordinates is output.

[0010] In one embodiment, making emotion adjustment in response to changes in scene information includes:

[0011] According to preset emotion adjustment rules, make emotion adjustments in response to changes in scene information;

[0012] The preset emotion adjustment rules include emotion parameter quantitative adjustment rules corresponding to the changes of entity information, environment information and interaction information respectively;

[0013] Obtaining the emotion change vector includes:

[0014] The emotion change vector is obtained by comprehensively adjusting the emotion parameters corresponding to the change of the ontological information, the emotion parameters corresponding to the change of the environmental information, and the emotion parameters corresponding to the interaction information.

[0015] In one embodiment, the emotion parameter quantization adjustment corresponding to the change of the ontology information, the emotion parameter quantization adjustment corresponding to the change of the environment information and the emotion parameter quantization adjustment corresponding to the interaction information are integrated to obtain the emotion change vector, including:

[0016] The quantitative adjustments of emotion parameters corresponding to changes in ontological information, the quantitative adjustments of emotion parameters corresponding to changes in environmental information, and the quantitative adjustments of emotion parameters corresponding to changes in interactive information are weighted according to their priorities, and an emotion change vector is obtained based on the weighted quantitative adjustments of emotion parameters corresponding to changes in ontological information, the quantitative adjustments of emotion parameters corresponding to changes in environmental information, and the quantitative adjustments of emotion parameters corresponding to changes in interactive information.

[0017] In one embodiment, the emotion parameters in the preset emotion adjustment rules include excitement and emotion, and the emotion parameter quantification adjustment rules include the slopes of the changes of excitement and emotion over time, and the extreme values ​​of the changes of excitement and emotion.

[0018] In one embodiment, the slope and the extremum are dynamically adjusted according to changes in the entity information.

[0019] In one embodiment, after acquiring scene information, the environment information and interaction information are input into the visual model and the language model to extract environment features and interaction features; the environment features, interaction features and ontology information are input into the large model to obtain the emotion change vector; or,

[0020] After acquiring the scene information, the ontological information, environmental information, and interaction information are input into the multimodal large model to obtain the emotion change vector.

[0021] In a second aspect, the present application also provides an emotion output device. The device comprises:

[0022] The acquisition module is used to obtain scene information in real time; the scene information includes entity information, environmental information and interaction information;

[0023] A response module, used to make emotion adjustments in response to changes in scene information and obtain an emotion change vector; wherein the reference coordinate system of the emotion change vector is an emotion coordinate system constructed based on a two-dimensional emotion model;

[0024] The output module is used to obtain the current emotion coordinates according to the historical emotion coordinates and emotion change vector before the scene information changes, and output the emotion expression corresponding to the current emotion coordinates.

[0025] In a third aspect, the present application further provides a robot, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps in the above-mentioned emotion output method when executing the computer program.

[0026] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps in the above-mentioned emotion output method when executed by a processor.

[0027] In a fifth aspect, the present application further provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned emotion output method are implemented.

[0028] The above-mentioned emotion output method, device and robot obtain scene information in real time; the scene information includes entity information, environmental information and interaction information; the emotion is adjusted in response to the change of scene information, and the emotion change vector is obtained; wherein, the reference coordinate system of the emotion change vector is the emotion coordinate system constructed based on the two-dimensional emotion model; the current emotion coordinate is obtained according to the historical emotion coordinate and the emotion change vector before the scene information changes, and the emotion expression corresponding to the current emotion coordinate is output. The present invention quantifies the emotion change based on the two-dimensional emotion model, and fine-tunes the emotion expression by outputting the emotion change vector, instead of directly jumping the emotion, and guides the robot to convert the emotion more naturally, thereby improving the human-computer interaction experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a flow chart of an emotion output method in one embodiment;

[0030] Figure 2 is a schematic diagram of a two-dimensional emotion model in one embodiment;

[0031] Figure 3 is a schematic diagram of the change of excitement over time in one embodiment;

[0032] Figure 4 is a schematic diagram of emotion changing over time in one embodiment;

[0033] Figure 5 is a structural block diagram of an emotion output network in one embodiment;

[0034] Figure 6 is a structural block diagram of an emotion output network in another embodiment;

[0035] Figure 7 FIG. 4 is a structural block diagram of an emotion output device in an embodiment. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0037] The emotion output method provided in the embodiment of the present application comprises the following steps:

[0038] Step 102, acquiring scene information in real time; the scene information includes entity information, environment information and interaction information.

[0039] The scene information is mainly acquired through the hardware configured on the robot, including cameras, microphones, infrared sensors, temperature sensors, etc. The scene information includes body information, such as power, posture, position, etc.; environmental information, such as temperature, humidity, light intensity, noise, concentration of air pollutants, etc.; and interaction information, including but not limited to interaction with the owner, surrounding people, and home IoT devices and toys.

[0040] Step 104 , making emotion adjustments in response to changes in scene information, and acquiring an emotion change vector; wherein a reference coordinate system of the emotion change vector is an emotion coordinate system constructed based on a two-dimensional emotion model.

[0041] This embodiment can use the existing Thayer emotion model as the two-dimensional emotion model. Figure 3 As shown. The model was originally a two-dimensional music emotion psychology model, which reflects the emotional characteristics of music works through the two dimensions of energy and pressure. The model is expressed in the form of a coordinate system. The ordinate of the coordinate system represents the energy dimension, which changes from calm to vitality, reflecting the degree of emotional activity; the abscissa represents the pressure dimension, which changes from negative to positive, reflecting the subjective feelings of the subject. Finally, the music emotions are divided into four representative categories, which are located in the four quadrants of the coordinate system.

[0042] When the scene information changes, the large model simulates and quantifies the emotional changes that the robot should experience based on deep learning, such as increasing the emotional activity dimension by 0.5 points and decreasing the pressure dimension by 0.3 points, correspondingly generating an emotional change vector (0.5, -0.3).

[0043] In this embodiment, the large model is a neural network model with a large number of parameters trained using a large amount of behavior and emotion data. The parameters of the large model are continuously adjusted through training so that it can achieve good performance in the task of emotion feedback. The emotions in the training data are also labeled using the Thayer emotion model.

[0044] Step 106 , obtaining the current emotion coordinates according to the historical emotion coordinates and the emotion change vector before the scene information changes, and outputting the emotion expression corresponding to the current emotion coordinates.

[0045] Emotions will change in real time as the scene changes, so the emotion coordinates will also change in real time. The historical emotion coordinates before the scene information changes are combined with the emotion change vector to obtain the emotion coordinates after the scene changes, that is, the current emotion coordinates. The current emotion coordinates reflect the emotions that the robot should have for the current scene. Finally, the robot makes corresponding emotional expressions based on the current emotion coordinates, which can be expressed through expression display, action, voice, etc.

[0046] When the scene information changes, we hope that the robot can give corresponding feedback to achieve good interaction. The traditional technical solution is to correspond each scene to an expression, such as a happy expression in the scene of being praised, and a neutral expression in the scene without interaction. This design solution may cause the robot to make a happy expression in the previous second and become indifferent in the next second. Such expression jumps lead to a strong sense of emotional change and separation, which greatly affects the user experience.

[0047] Therefore, this embodiment realizes the scientific classification of emotions by introducing a two-dimensional emotion model, and uses a large model to obtain the emotion change vector in response to the change of scene information, completes the quantitative analysis of the emotion change, and finally obtains the current emotion coordinates based on the historical emotion coordinates combined with the emotion change vector. That is, the current emotion is not only associated with the emotion change in response to the scene change, but also with the historical emotion before the scene information changes, thereby realizing the continuity of emotion change, effectively avoiding the jump of emotion, and improving the user experience.

[0048] In this embodiment, hardware such as cameras and microphones collect scene information in real time, and then hand over the collected scene information to the large model for processing to obtain emotion coordinates. The large model generates execution instructions corresponding to the emotion coordinates and hands them over to the robot body for execution. The robot externalizes emotions such as happiness, fear, worry, sadness, etc. by displaying expressions on the display screen and playing voice through the speaker.

[0049] In one embodiment, in step 102, emotion adjustment is made in response to changes in scene information, including: according to preset emotion adjustment rules, emotion adjustment is made in response to changes in scene information; wherein the preset emotion adjustment rules include emotion parameter quantification adjustment rules corresponding to changes in entity information, environmental information, and interactive information, respectively.

[0050] In this embodiment, the preset emotion adjustment rules are used to specify the emotion parameter adjustment schemes corresponding to various scene information changes. The preset emotion adjustment rules are also followed in the large model training stage, so that the large model can output the emotion change vector more accurately. Under the Thayer emotion model, the horizontal coordinate of the emotion coordinate represents excitement, and the vertical coordinate represents emotion. Both excitement and emotion can be regarded as emotion parameters. The preset emotion adjustment rules include the rules for the corresponding changes of the horizontal and vertical coordinates of the emotion coordinates when the scene information changes. In order to further demonstrate the ideas for formulating the preset emotion adjustment rules, Table 1 lists several scenes and corresponding emotion coordinate change schemes:

[0051] Table 1

[0052] Scene Change Excitement emotion Interact with the host Increase Increase Praise from the owner Slightly increase Increase Criticized by the owner Slightly lower reduce Find interesting items Increase Increase Comfortable temperature and light conditions Slightly increase Slightly increase

[0053] In Table 1, "slightly increased" and "slightly decreased" are relative expressions corresponding to "increased" and "decreased", respectively, and the specific values ​​are determined according to the parameter adjustment of the large model. This is only an example for understanding the preset emotion adjustment rules, and does not limit the method itself.

[0054] In step 102, obtaining the emotion change vector includes: comprehensively adjusting the emotion parameters corresponding to the change of the entity information, adjusting the emotion parameters corresponding to the change of the environment information, and adjusting the emotion parameters corresponding to the interaction information to obtain the emotion change vector.

[0055] The collection of scene information is real-time and comprehensive. Changes in the entity information, environmental information and interactive information may occur independently or simultaneously. Therefore, when the large model generates an emotion change vector, it is necessary to consider the quantitative adjustment of emotion parameters corresponding to the changes in the entity information, the quantitative adjustment of emotion parameters corresponding to the changes in the environmental information, and the quantitative adjustment of emotion parameters corresponding to the interactive information. Each adjustment can be determined according to its corresponding adjustment rules.

[0056] In one embodiment, the quantitative adjustments of emotion parameters corresponding to changes in ontological information, the quantitative adjustments of emotion parameters corresponding to changes in environmental information, and the quantitative adjustments of emotion parameters corresponding to changes in interactive information are integrated to obtain an emotion change vector, including: weighting the quantitative adjustments of emotion parameters corresponding to changes in ontological information, the quantitative adjustments of emotion parameters corresponding to changes in environmental information, and the quantitative adjustments of emotion parameters corresponding to changes in interactive information according to priority, and obtaining the emotion change vector according to the weighted quantitative adjustments of emotion parameters corresponding to changes in ontological information, the quantitative adjustments of emotion parameters corresponding to changes in environmental information, and the quantitative adjustments of emotion parameters corresponding to changes in interactive information.

[0057] When adjusting emotions in response to changes in scene information, there may be conflicts in the direction of emotion adjustment. For example, changes in environmental information reduce the robot's excitement, while changes in interactive information increase the robot's excitement. The large model will conflict when processing the emotional parameter adjustments corresponding to these two situations, resulting in the robot clearly being supposed to give feedback to the owner's interactive information, but due to the influence of changes in environmental information, it gives feedback that does not match the interactive information. In order to reduce the impact of conflicts, this embodiment assigns different priorities to different scene information. For example, interactive information with the owner has the highest priority, and environmental information has the lowest priority. In other embodiments, the scene information can be further refined and assigned different priorities. Taking the scene information in Table 1 as an example, the order of priority is: entertaining interaction with the owner> being praised by the owner = being criticized by the owner> discovering fun items> comfortable temperature and lighting conditions.

[0058] While determining the priority corresponding to different scene information, different scene information is weighted according to the priority. The big model obtains the emotion change vector based on the quantitative adjustment of emotion parameters corresponding to the weighted ontological information changes, the environmental information changes, and the interactive information changes. The emotion change vector is obtained comprehensively to reduce the impact of conflicts, output more accurate and appropriate emotion expressions, and improve user experience.

[0059] In one embodiment, the emotion parameters in the preset emotion adjustment rule include excitement and emotion, and the emotion parameter quantification adjustment rule includes the slopes of the changes of excitement and emotion over time, and the extreme values ​​of the changes of excitement and emotion.

[0060] In order to further optimize the robot's emotional expression and make the emotional changes more coherent and natural, this embodiment sets the slope of the emotional parameters changing over time. For example, when the excitement increases by 0.5, the robot will not directly jump to the expression corresponding to the excitement increasing by 0.5, but will change at a fixed slope within a certain period of time until the excitement increases to 0.5. At the same time, in order to avoid the same scene changes causing the same emotional parameters to change infinitely, extreme values ​​are set to constrain this situation. For example, when the robot is praised all the time, the excitement will continue to increase, but when the excitement reaches the extreme value, it will no longer continue to change, but maintain the current excitement, waiting for the next scene information change that causes the excitement to change.

[0061] like Figure 3 and Figure 4As shown in the figure, the slope graph of excitement changing over time and the slope graph of emotion changing over time are respectively shown. In these two figures, the five scenes of playing with the owner, playing by oneself, being praised, comfortable environment and being criticized are used as examples for explanation. The four scenes of playing with the owner, playing by oneself, being praised, and comfortable environment are positively correlated with excitement and emotion at the same time, and the scene of being criticized is negatively correlated with excitement and emotion at the same time. In terms of excitement, the slopes from large to small are playing with the owner, playing by oneself, being praised, and comfortable environment, and the extreme values ​​also follow this order; and in terms of emotion, the slopes from large to small are being praised, playing with the owner, playing by oneself, and comfortable environment, and the extreme values ​​also follow this order.

[0062] In this embodiment, the slope can be used as the input of the large model, and the large model can choose how much to adjust each time, or the large model can only output increase or decrease, and the amount of adjustment is determined by advance rules.

[0063] This embodiment ensures stable changes in the robot's emotions through the slope, and avoids extreme emotions through extreme value constraints, thereby improving the user experience.

[0064] In one embodiment, the slope and the extremum are dynamically adjusted according to changes in the entity information.

[0065] The ontological information reflects the robot's own situation, such as low battery, stuck, etc. Some ontological information will affect the robot's overall mood. This embodiment makes adjustments for special situations. For example, when the robot's battery is low, the overall excitement will be reduced, and the slope and extreme value will be adjusted accordingly to make the robot behave calmer.

[0066] In one embodiment, after acquiring scene information, environmental information and interaction information are input into a visual model and a language model to extract environmental features and interaction features; environmental features, interaction features and ontological information are input into a large model to obtain an emotion change vector; or, after acquiring scene information, ontological information, environmental information and interaction information are input into a multimodal large model to obtain an emotion change vector.

[0067] like Figure 5 As shown in the figure, a network structure is used to realize the output of emotions. The collected environmental information and the state of the owner are input into the visual model, and the owner's voice (including language and tone) is input into the language model. The visual model can extract environmental features and some interactive features, and the voice model can extract some interactive features. Then, the output of the visual model and the output of the language model are input into the large model together with the ontology information. The large model outputs the current emotional coordinates, and the robot outputs the corresponding emotional expression according to the current emotional coordinates.

[0068] like Figure 6As shown in FIG. 1 , another network structure for realizing emotion output is shown. After acquiring the scene information, the entity information, environment information and interaction information are directly input into the multimodal large model. The multimodal large model outputs the current emotion coordinates, and the robot outputs the corresponding emotion expression according to the current emotion coordinates.

[0069] Figure 5 and Figure 6 In the figure, the interactive information is simplified into the owner's status and the owner's voice, and the environmental information is simplified into visual information. In actual situations, the environmental information and interactive information rely on more complex sensor systems, and the information form is also more complex, which will not be elaborated here.

[0070] Large models usually refer to deep learning models with a large number of parameters. They have billions or even tens of billions of parameters. Large models can handle complex tasks because they have stronger representation and learning capabilities, but they also require a lot of computing resources and data to train. Usually large models are used to handle single-modality tasks such as natural language processing, image recognition, and speech recognition.

[0071] Multimodal large models are models that can process multiple different types of data at the same time, such as text, images, audio, etc. They can fuse and understand information from different modalities. Multimodal large models can combine data from multiple modalities to complete more complex tasks. For example, they can understand the content of text and images at the same time, thereby providing richer answers or generating more precise descriptions.

[0072] Simply put, large models focus on single-modal information processing, while multimodal large models are able to process and fuse multiple types of data to provide more comprehensive intelligent services.

[0073] According to the characteristics of large models and multimodal large models, the following Figure 5 The network structure based on the large model shown and Figure 6 The network structure based on the multimodal large model is shown.

[0074] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0075] Based on the same inventive concept, the embodiment of the present application also provides an emotion output device for implementing the emotion output method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more emotion output device embodiments provided below can refer to the limitations of the emotion output method above, and will not be repeated here.

[0076] In one embodiment, Figure 7 As shown, an emotion output device is provided, including: a collection module 702, a response module 704 and an output module 706, wherein:

[0077] The acquisition module 702 is used to acquire scene information in real time; the scene information includes entity information, environmental information and interaction information.

[0078] The response module 704 is used to make emotion adjustments in response to changes in scene information and obtain an emotion change vector; wherein the reference coordinate system of the emotion change vector is an emotion coordinate system constructed based on a two-dimensional emotion model.

[0079] The output module 706 is used to obtain the current emotion coordinates according to the historical emotion coordinates and the emotion change vector before the scene information changes, and output the emotion expression corresponding to the current emotion coordinates.

[0080] The response module 704 is also used to make emotion adjustments in response to changes in scene information according to preset emotion adjustment rules; wherein the preset emotion adjustment rules include emotion parameter quantization adjustment rules corresponding to changes in entity information, changes in environmental information, and changes in interactive information. The response module 704 is also used to synthesize the emotion parameter quantization adjustment corresponding to changes in entity information, the emotion parameter quantization adjustment corresponding to changes in environmental information, and the emotion parameter quantization adjustment corresponding to interactive information to obtain an emotion change vector.

[0081] The response module 704 is also used to weight the quantitative adjustments of emotion parameters corresponding to the changes in the ontological information, the quantitative adjustments of emotion parameters corresponding to the changes in the environmental information, and the quantitative adjustments of emotion parameters corresponding to the interactive information according to the priority, and obtain the emotion change vector according to the weighted quantitative adjustments of emotion parameters corresponding to the changes in the ontological information, the quantitative adjustments of emotion parameters corresponding to the changes in the environmental information, and the quantitative adjustments of emotion parameters corresponding to the interactive information.

[0082] The emotion parameters in the emotion preset adjustment rules include excitement and emotion, and the emotion parameter quantitative adjustment rules include the slopes of the changes of excitement and emotion over time, and the extreme values ​​of the changes of excitement and emotion.

[0083] The slope and extreme value are dynamically adjusted according to the changes of entity information.

[0084] After acquiring the scene information, the environmental information and interaction information are input into the visual model and the language model to extract the environmental features and interaction features; the environmental features, interaction features and ontological information are input into the large model to obtain the emotion change vector; or, after acquiring the scene information, the ontological information, environmental information and interaction information are input into the multimodal large model to obtain the emotion change vector.

[0085] Each module in the above-mentioned emotion output device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0086] In one embodiment, a robot is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in all the method embodiments of the above method when executing the computer program.

[0087] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in all the method embodiments of the above method are implemented.

[0088] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in all the method embodiments of the above method when executed by a processor.

[0089] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0090] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited thereto. The processors involved in each embodiment provided in this application may be general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, data processing logic devices based on quantum computing, etc., but are not limited thereto.

[0091] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0092] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present 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 application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for outputting emotions, characterized in that: The method comprises: Acquire scene information in real time; the scene information includes entity information, environment information and interaction information; In response to the change of the scene information, an emotion adjustment is made to obtain an emotion change vector; wherein a reference coordinate system of the emotion change vector is an emotion coordinate system constructed based on a two-dimensional emotion model; The current emotion coordinates are acquired according to the historical emotion coordinates before the scene information changes and the emotion change vector, and the emotion expression corresponding to the current emotion coordinates is output.

2. The method according to claim 1, characterized in that The making emotion adjustment in response to the change of the scene information comprises: According to preset emotion adjustment rules, making emotion adjustment in response to the change of the scene information; The preset emotion adjustment rules include emotion parameter quantification adjustment rules corresponding to the entity information change, the environment information change and the interaction information change respectively; The obtaining of the emotion change vector comprises: The emotion change vector is obtained by comprehensively analyzing the quantitative adjustment of the emotion parameters corresponding to the change of the entity information, the quantitative adjustment of the emotion parameters corresponding to the change of the environmental information, and the quantitative adjustment of the emotion parameters corresponding to the interaction information.

3. The method according to claim 2, characterized in that The step of comprehensively adjusting the emotion parameter corresponding to the change of the entity information, adjusting the emotion parameter corresponding to the change of the environment information, and adjusting the emotion parameter corresponding to the interaction information to obtain the emotion change vector includes: The quantitative adjustment of the emotion parameters corresponding to the change of the ontological information, the quantitative adjustment of the emotion parameters corresponding to the change of the environmental information, and the quantitative adjustment of the emotion parameters corresponding to the interactive information are weighted according to the priority, and the emotion change vector is obtained according to the weighted quantitative adjustment of the emotion parameters corresponding to the change of the ontological information, the quantitative adjustment of the emotion parameters corresponding to the change of the environmental information, and the quantitative adjustment of the emotion parameters corresponding to the interactive information.

4. The method according to claim 2, characterized in that: The emotion parameters in the emotion preset adjustment rule include excitement and emotion, and the emotion parameter quantification adjustment rule includes the slopes of the excitement and emotion changing over time, and the extreme values ​​of the excitement and emotion changing respectively.

5. The method according to claim 4, characterized in that The slope and the extreme value are dynamically adjusted according to the change of the entity information.

6. The method according to any one of claims 1 to 5, characterized in that: After acquiring the scene information, inputting the environment information and the interaction information into a visual model and a language model to extract environment features and interaction features; Input the environmental features, the interaction features and the ontology information into a large model to obtain the emotion change vector; or, After acquiring the scene information, the entity information, the environment information and the interaction information are input into a multimodal large model to acquire the emotion change vector.

7. An emotion output device, characterized in that: The device comprises: The acquisition module is used to obtain scene information in real time; the scene information includes entity information, environment information and interaction information; A response module, used to make emotion adjustments in response to the change of the scene information and obtain an emotion change vector; wherein the reference coordinate system of the emotion change vector is an emotion coordinate system constructed based on a two-dimensional emotion model; The output module is used to obtain the current emotion coordinates according to the historical emotion coordinates before the scene change information and the emotion change vector, and output the emotion expression corresponding to the current emotion coordinates.

8. A robot comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images