Emotional companion robot interaction method for single users, robot and storage medium

Abstract

This invention discloses an interactive method, robot, and storage medium for emotional companionship robots targeting single users, relating to the field of artificial intelligence. It includes: during a conversation between the robot and a single user, using conversation segments as the smallest update unit, forming a textual personality evidence package for each segment before updating the route; separately updating the stable personality layer, temporary state layer, and dependent state record layer; updating the personality profile only based on the textual personality evidence package of the stable personality layer, avoiding distortion of the personality profile caused by directly writing the user's short-term state at a specific time or event into the stable personality layer; and then determining a companionship dialogue strategy for the single user based on the long-term personality profile and the current interaction state to generate and play robot voice, completing personalized interaction between the robot and the single user, satisfying the single user's personalized needs for in-depth expression and emotional companionship, thereby enhancing the emotional companionship experience for single users.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence, and in particular to an interactive method, robot, and storage medium for emotional companionship robots for single users. Background Technology

[0002] With the rapid development of artificial intelligence, intelligent robots have already touched upon all aspects of people's lives. For example, intelligent robots have gradually taken on diverse functions of human-computer interaction and information services. For instance, users can ask robots questions via voice or touch to obtain various information such as weather information, current events, encyclopedic knowledge answers, and general knowledge.

[0003] However, in the process of human-computer interaction between robots and users, robots usually respond to users' questions directly based on preset rules. This makes it impossible for robots to meet the personalized needs of single users for in-depth expression and emotional companionship in scenarios where single users communicate with robots for a long time, resulting in a poor emotional companionship experience for single users. Summary of the Invention

[0004] In view of the above-mentioned technical problems, the present invention provides an interactive method, robot and storage medium for emotional companionship robot for single users, aiming to overcome the above problems or at least partially solve the above problems.

[0005] The first aspect of this invention provides an interaction method for an emotional companion robot for single users, applied to an emotional companion robot, the emotional companion robot comprising: a voice interaction layer, an evidence processing layer, and a profiling and control layer; the profiling and control layer comprising: a stable personality layer, a temporary state layer, a dependency state recording layer, and a temporary storage layer; the method comprising: The voice interaction layer collects the voice of single users, converts the user's voice into user text, and adds it to the conversation record; The evidence processing layer divides the dialogue record into conversation segments; it performs text personality evidence analysis on each conversation segment, and generates a text personality evidence package for each conversation segment based on the evidence analysis results; it performs update routing judgment on the text personality evidence package of each conversation segment, routing text personality evidence packages belonging to long-term states and with sensitivity below the sensitivity threshold to the stable personality layer, routing text personality evidence packages belonging to short-term states or involving stressful events to the temporary state layer, and routing text personality evidence packages belonging to high-frequency emotional expression states to the dependency state recording layer; The profiling and control layer updates the long-term personality profile based on the textual personality evidence package in the stable personality layer; generates the current interaction state based on the textual personality evidence package in the temporary state layer and the dependent state record layer; and determines the companionship dialogue strategy for single users based on the long-term personality profile and the current interaction state.

[0006] A second aspect of the present invention provides a robot, the robot including a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein when the program or instructions are executed by the processor, the steps of the emotional companion robot interaction method for single users as described in the first aspect of the present invention are implemented.

[0007] A third aspect of the present invention provides a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the emotional companion robot interaction method for single users as described in the first aspect of the present invention.

[0008] In the emotional companionship robot interaction method for single users proposed in this invention, for long-term emotional companionship scenarios for single users, the dialogue between the robot and the single user is divided into multiple conversation segments. Then, a text personality evidence package is generated for each conversation segment and an update routing judgment is performed to determine whether the text personality evidence package is routed to the stable personality layer, the temporary state layer, or the dependent state record layer. Then, the long-term personality profile is updated based on the text personality evidence package in the stable personality layer, and the current interaction state is generated based on the text personality evidence packages in the temporary state layer and the dependent state record layer. Finally, the companionship dialogue strategy for single users is determined based on the long-term personality profile and the current interaction state. Thus, in scenarios where single users engage in long-term interactions with a robot, this invention uses conversation fragments as the smallest update unit. It forms textual personality evidence packages of these fragments before routing updates, separating and updating the stable personality layer, temporary state layer, and dependent state record layer. The personality profile of the single user is updated only based on the textual personality evidence packages entering the stable personality layer. This ensures that personality profile updates have a clear source, evidentiary justification, and processing path, avoiding distortion of the personality profile caused by directly writing short-term states (such as late-night confessions, expressions of stress, or sensitive topics) into the stable personality layer, and preventing direct contamination of the long-term personality profile. Furthermore, this invention determines a companionship dialogue strategy for single users based on the long-term personality profile and the current interaction state. The robot's voice can be generated and played according to this strategy to complete a personalized current round of interaction between the robot and the single user. This companionship dialogue strategy is not limited to the robot's answers but also includes dialogue control parameters such as the robot's response methods, thereby meeting the single user's personalized needs for in-depth expression and emotional companionship, and ultimately enhancing the single user's emotional companionship experience. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart illustrating the steps of an interactive method for an emotional companion robot for single users, as shown in an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the structure of a personality evidence package according to an embodiment of the present invention; Figure 3 This is a detailed schematic diagram illustrating an interactive method for an emotional companion robot for single users, as shown in an embodiment of the present invention; Figure 4 This is a schematic diagram of a robot according to an embodiment of the present invention. Detailed Implementation

[0011] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0012] The emotional companion robot interaction method for single users provided in this embodiment is applied to emotional companion robots. Please refer to... Figure 1 , Figure 1 This is a flowchart illustrating the steps of an interactive method for an emotional companion robot for single users, as shown in an embodiment of the present invention. The interactive method for an emotional companion robot for single users provided in this embodiment is applied to an emotional companion robot, which includes: a voice interaction layer, an evidence processing layer, and a profiling and control layer; wherein, the profiling and control layer includes: a stable personality layer, a temporary state layer, a dependency state recording layer, and a temporary storage layer. Figure 1 As shown, the emotional companion robot interaction method for single users provided in this embodiment includes at least the following steps: Step S11: The voice interaction layer collects the voice of single users, converts the user's voice into user text, and adds it to the conversation record.

[0013] In this embodiment, the long-term emotional companionship scenario for single users refers to a scenario where single, solitary, or rarely accompanied by real people engage in long-term verbal communication with an emotional companionship robot in a private space such as home. This scenario is more likely to involve deep confiding, requests for emotional companionship, frequent communication, and a tendency to rely on the robot's responses. In this embodiment, single users are the target users of the emotional companionship robot, referring to long-term interaction partners identified through account binding, device binding, wake-up confirmation, user self-reporting, etc. The personality profile in this embodiment is only created and updated for single users. Deep confiding refers to an interactive state where the user repeatedly expresses loneliness, stress, depression, relationship problems, or a need for companionship to the robot within a certain period. Deep confiding also includes dialogue segments where the user continuously expresses their views on topics such as intimate relationships, self-evaluation, and life choices.

[0014] During a conversation between an emotional companionship robot and a single user, the robot can collect the user's voice through a voice interaction layer, convert it into text, and add it to the dialogue log. The voice interaction layer then continuously sends updated dialogue logs to the evidence processing layer. The dialogue log contains the user's text, the robot's questions, the dialogue time, the topic, and historical context (user text and robot text from previous rounds). In one embodiment, the robot uses ASR (Automatic Speech Recognition) to convert the user's voice into text and outputs the corresponding ASR confidence score. The ASR confidence score is used to control the reliability of personality evidence and update routing.

[0015] Step S12: The evidence processing layer divides the dialogue record into conversation segments; performs text personality evidence analysis on each conversation segment, and generates a text personality evidence package for each conversation segment based on the evidence analysis results; updates the routing judgment for the text personality evidence package of each conversation segment, routing text personality evidence packages belonging to long-term states and with sensitivity below the sensitivity threshold to the stable personality layer, routing text personality evidence packages belonging to short-term states or involving stressful events to the temporary state layer, and routing text personality evidence packages belonging to high-frequency emotional expression states to the dependency state recording layer.

[0016] In this embodiment, the evidence processing layer can divide the dialogue record into session fragments to obtain fragment division results, which include one or more session fragments obtained from the division.

[0017] In this embodiment, the dialogue record can be divided into conversation segments based on one or more conditions, such as whether the topic has changed, whether the question and answer have been completed, whether a deep conversation has started or ended, whether there is a continuous request for companionship, whether the conversation is long, or whether the user has corrected their stance. For example, a topic belongs to one conversation segment, a complete question and answer belongs to one conversation segment, the beginning and end of a deep conversation belongs to one conversation segment, and a long conversation can be divided into multiple conversation segments based on a preset conversation duration, etc. There are no restrictions on this.

[0018] In this embodiment, the evidence processing layer performs text-based personality evidence analysis on each conversation segment, obtaining the evidence analysis results for each segment, and generating a text-based personality evidence package for each segment based on these results. The text-based personality evidence analysis includes: extracting personality evidence from user text, bot questions, and historical context; and identifying whether the conversation segment contains deep venting, sensitive topics related to intimate relationships, high-frequency emotional expression status markers, and interaction dependency markers. In a specific example, the text-based personality evidence analysis can employ a large language model, or it can use a text classification model, rule templates, keyword / semantic tagging systems, or a combination of these methods.

[0019] The text-based personality evidence package is a structured evidence package corresponding to a conversation segment, recording basic information, textual evidence, personality evidence, companionship status, and processing records. Personality evidence refers to personality-related clues extracted from textual content such as user expressions in the conversation segment, the relationship between the robot's questions and user responses, and historical dialogue summaries. It is used to explain the conversation segment on which the profile update is based, the personality dimensions involved, the strength of the evidence, and subsequent processing steps.

[0020] In one example, the textual personality evidence package includes one or more of the following: ASR confidence, strength of evidence, sensitivity markers, state / trait markers, depth of expression markers, interaction dependency state markers, etc.

[0021] In one embodiment, the deep emotional expression markers and interaction dependency state markers can be obtained based on time windows, wake-up frequency, conversation duration, repeated requests for comfort, topic type, semantic tags, or a combination of methods. In a specific optional example, if a single user wakes up the robot for extended periods of time on consecutive nights, repeatedly saying things like "Talk to me a little longer" or "Don't stop," the robot can mark these conversation fragments as high-frequency emotional expression or interaction dependency states.

[0022] In this embodiment, the evidence processing layer updates the routing judgment for the text personality evidence package of each session fragment to determine whether the text personality evidence package is routed to the stable personality layer, the temporary state layer, or the dependent state record layer. The stable personality layer represents a long-term, relatively stable personality tendency. In this embodiment, stable, clear, and low-sensitivity text personality evidence packages enter the stable personality layer; that is, text personality evidence packages belonging to a long-term state and with sensitivity below a sensitivity threshold are routed to the stable personality layer. The sensitivity of the text personality evidence package is determined by the sensitivity markers within it. A stable and clear text personality evidence package indicates that it belongs to a long-term state. Clarity is defined by whether the ASR confidence level in the text personality evidence package is greater than a preset confidence threshold. If the ASR confidence level is greater than the preset confidence threshold, the text personality evidence package is determined to belong to a clear interaction state; if the ASR confidence level is not greater than the preset confidence threshold, the text personality evidence package is determined not to belong to a clear interaction state.

[0023] In this embodiment, the temporary state layer represents short-term behaviors caused by recent stress, emotions, or special events. The stable personality layer updates more slowly, while the short-term companionship layer updates more quickly and is more prone to decay. This embodiment separates the stable personality layer from the temporary state layer to avoid including a single outpouring, stress expression, or short-term state in the long-term personality profile. In this embodiment, the evidence processing layer routes textual personality evidence packages representing late-night outpourings and / or stressful events to the temporary state layer; that is, textual personality evidence packages belonging to short-term states or involving stressful events are routed to the temporary state layer. Whether a textual personality evidence package belongs to a short-term state is determined by a deep outpouring marker within the textual personality evidence package. If the deep outpouring marker indicates the presence of deep outpouring behavior, the textual personality evidence package is determined to be a short-term state; if the deep outpouring marker indicates the absence of deep outpouring behavior, the textual personality evidence package is determined not to be a short-term state.

[0024] In this embodiment, the dependency state recording layer represents the user's interactive dependency on the robot; the evidence processing layer routes text personality evidence packages belonging to the high-frequency emotional expression state to the dependency state recording layer. Whether a text personality evidence package belongs to the high-frequency emotional expression state is determined by the high-frequency emotional expression state marker within the text personality evidence package. If the high-frequency emotional expression state marker in the text personality evidence package indicates the existence of a high-frequency emotional expression state, then the text personality evidence package is determined to belong to the high-frequency emotional expression state; if the high-frequency emotional expression state marker in the text personality evidence package indicates the absence of a high-frequency emotional expression state, then the text personality evidence package is determined not to belong to the high-frequency emotional expression state. Furthermore, a high-frequency emotional expression state refers to the interactive state in which the target user continuously or repeatedly expresses emotions such as loneliness, stress, depression, relationship difficulties, or a need for companionship to the robot within a certain time window.

[0025] Step S13: The profiling and control layer updates the long-term personality profile based on the text personality evidence package in the stable personality layer; generates the current interaction state based on the text personality evidence package in the temporary state layer and the dependent state record layer; and determines the companionship dialogue strategy for single users based on the long-term personality profile and the current interaction state.

[0026] In this embodiment, the profiling and control layer can update the long-term personality profile of a single user based on the text personality evidence package in the stable personality layer. Therefore, the long-term personality profile update in this embodiment does not directly save chat summaries, but rather first forms fragment-level text personality evidence packages. It is understood that in this embodiment, text personality evidence packages not routed to the stable personality layer (such as text personality evidence packages routed to the temporary state layer or dependent state record layer) cannot be used to update the long-term personality profile.

[0027] In one specific example, the long-term personality profile can be updated using weighted averaging, time decay, manual verification followed by updates, or other time-series update methods.

[0028] In one optional example, the long-term personality profile is used to control the robot's subsequent response methods, proactive follow-up questions, and TTS output.

[0029] In this embodiment, the profiling and control layer can generate the current interaction state based on the textual personality evidence package in the temporary state layer and the textual personality evidence package in the dependent state record layer. The current interaction state represents the recorded interactions between a single user and the emotional companion robot regarding emotional companionship at present and in the past.

[0030] After determining the latest long-term personality profile and the latest current interaction state, the profiling and control layer can determine a personalized companionship dialogue strategy for single users based on these parameters. This strategy controls the content and manner of the robot's responses and includes at least one of the following control parameters: response length, follow-up question depth, continuous companionship duration, frequency of proactive greetings, and frequency of follow-up questions. The frequency of proactive greetings and the frequency of follow-up questions allow the robot's interaction with single users to be more adapted to the current interaction state.

[0031] Subsequently, the emotional companionship robot can generate robot text tailored to the user's voice through the voice interaction layer, based on dialogue records and companionship dialogue strategies for single users. This robot text is then converted into robot speech and played via TTS (Text-to-Speech) to complete personalized interaction between the emotional companionship robot and the single user. The robot text includes text replies, text questions, or reminders generated by the robot.

[0032] This embodiment considers that late-night venting, expressions of loneliness, and work stress are only short-term states of single users under specific times or events. Directly writing these into the stable personality layer could easily distort the personality profile. Therefore, it is necessary to distinguish between the stable personality layer, the temporary state layer, and the dependency state recording layer, and update these layers separately to prevent late-night venting, stress expressions, or sensitive topics from directly contaminating the long-term personality profile. In one embodiment, the companionship dialogue strategy is determined based on the single user's long-term personality profile and the single user's current interaction state. Thus, the determined companionship dialogue strategy also includes deep venting state processing and interaction dependency control, which is not only suitable for the single user's long-term personality profile but also more suitable for the current interaction dependency state and short-term companionship state between the single user and the robot.

[0033] In this embodiment, in scenarios where single users engage in long-term communication with a robot, the present invention uses conversation fragments as the smallest update unit. Textual personality evidence packages of these fragments are formed before update routing, separating and updating the stable personality layer, temporary state layer, and dependent state record layer. The personality profile of the single user is updated only based on the textual personality evidence packages entering the stable personality layer. This ensures that personality profile updates have a clear source, evidentiary reason, and processing path, avoiding distortion of the personality profile caused by directly writing short-term states (such as late-night confessions, expressions of stress, or sensitive topics) into the stable personality layer, thus preventing direct contamination of the long-term personality profile. Furthermore, the present invention determines a companionship dialogue strategy for single users based on the long-term personality profile and the current interaction state. Robotic voice can be generated and played according to this strategy to complete a personalized current round of interaction between the robot and the single user. This companionship dialogue strategy is not limited to the robot's answers but also includes dialogue control parameters such as the robot's response methods, thereby meeting the single user's personalized needs for in-depth expression and emotional companionship, and ultimately enhancing the single user's emotional companionship experience.

[0034] In conjunction with the above embodiments, in one implementation, the present invention also provides an interactive method for an emotional companion robot for single users. In addition to the steps described above, this method may further include steps S21 to S23: Step S21: The profiling and control layer determines the TTS parameters for single users based on the long-term personality profile and the current interaction status.

[0035] In this embodiment, the profiling and control layer can also determine TTS parameters for single users based on long-term personality profiles and current interaction status. These TTS parameters include parameters such as TTS speech rate, pauses, and volume.

[0036] Step S22: The profiling and control layer feeds back the companionship dialogue strategy and TTS parameters for single users to the voice interaction layer.

[0037] In this embodiment, after the profiling and control layer generates a companion dialogue strategy and TTS parameters for single users, it feeds these strategies and TTS parameters back to the voice interaction layer.

[0038] Step S23: Based on the dialogue record, the voice interaction layer generates a text reply according to the companion dialogue strategy for single users and the TTS parameters, and outputs it through TTS.

[0039] In this embodiment, the voice interaction layer can generate a text response corresponding to the user's semantics based on the dialogue record and the companion dialogue strategy for single users, and then convert the text response into robot voice according to the TTS parameters and play it.

[0040] In this embodiment, a companion dialogue strategy and TTS parameters are determined based on the long-term personality profile and current interaction state of a single user to generate and play robot voice, thereby further satisfying the single user's personalized needs for in-depth expression and emotional companionship, and enhancing the single user's emotional companionship experience. Thus, this embodiment, targeting the long-term emotional companionship scenario for single users, not only enables variations in response content but also achieves stable control over the TTS conversion parameter values ​​and the robot's long-term speaking style by stably mapping the long-term personality profile, short-term companionship state, and interaction dependency state to dialogue control parameters, thereby fulfilling the single user's personalized needs for in-depth expression and emotional companionship.

[0041] In conjunction with any of the above embodiments, the present invention also provides an interactive method for an emotional companion robot for single users, wherein, in addition to the steps described above, the method may further include steps S31 and S32: Step S31: When the profiling and control layer recognizes that the current interaction state is a high-frequency emotional expression state, it triggers the voice interaction layer to provide a companionship boundary prompt.

[0042] In this embodiment, after generating the current interaction state based on the textual personality evidence package in the temporary state layer and the dependent state recording layer, the profiling and control layer identifies the current interaction state. When the current interaction state is identified as a high-frequency emotional expression state, the profiling and control layer triggers the voice interaction layer to provide a companionship boundary prompt. This companionship boundary prompt refers to adding gentle control content to the robot's response when the robot identifies a high-frequency emotional expression or a robot interaction dependency state. For example, this might involve reducing continuous questioning, prompting for rest, guiding the user to focus on real-world tasks, or reducing the frequency of active wake-up. In this embodiment, the companionship dialogue strategy also includes the intensity of the boundary prompt.

[0043] Step S32: Based on the dialogue record, the voice interaction layer generates companionship boundary prompt text according to the companionship dialogue strategy for single users and TTS parameters, and outputs it through TTS.

[0044] In this embodiment, the voice interaction layer can generate companionship boundary prompt text based on the dialogue record and a companionship dialogue strategy for single users. Then, it converts the generated companionship boundary prompt text into robot speech according to TTS parameters and plays it. This companionship boundary prompt text is used to guide single users to perform target tasks, thereby reducing the frequency of interaction with the emotional companionship robot.

[0045] In one example, the voice interaction layer provides companionship boundary prompts to prompt users to pause continuous conversations, switch to preset real-world tasks, take short breaks, or reduce the frequency of subsequent proactive greetings. For example, the companionship boundary prompt text generated by the voice interaction layer includes: "How about taking a short break and making some mung bean and pork rib soup?"; "How about closing your eyes and resting for ten minutes before doing some exercise?"; "How about going to the park for a stroll and enjoying nature?"; etc., without imposing any restrictions.

[0046] In a specific example, when a single user expresses work stress, fatigue, and loneliness about "having no one to talk to at home" to the robot late at night, the robot will mark the target conversation segment as a deep outpouring and short-term state. At this point, the target conversation segment enters a temporary companionship layer, without directly increasing the long-term neuroticism judgment, and determines that the single user's current interaction state is a high-frequency emotional outpouring state. The robot then reduces follow-up questions, uses more summarizing and confirming responses, and appropriately reduces the TTS speaking speed and increases pauses.

[0047] In conjunction with any of the above embodiments, in one implementation, the present invention also provides an interactive method for an emotional companion robot for single users. In addition to the steps described above, this method may further include step S41: Step S41: When the profiling and control layer recognizes that the current interaction state is that the user is repeatedly seeking comfort or continuously engaging in deep conversation, it sends an instruction to the voice interaction layer to reduce the depth of follow-up questions and the frequency of proactive greetings.

[0048] In this embodiment, after generating the current interaction state based on the textual personality evidence package in the temporary state layer and the dependency state recording layer, the profiling and control layer identifies the current interaction state. When it is identified that the current interaction state is that the user repeatedly seeks comfort or continuously engages in deep venting, it is determined that the single user is currently in a state of heavy interaction dependence with the robot. The profiling and control layer then sends instructions to the voice interaction layer to reduce the depth of follow-up questions and the frequency of proactive greetings, in order to prevent the interaction method from further reinforcing the high-frequency emotional venting or interaction dependence state, and to prevent the single user from becoming overly reliant on the emotional companionship robot. Thus, the voice interaction layer can adjust the companionship dialogue strategy based on the received instructions to reduce the depth of follow-up questions and the frequency of proactive greetings, and generate reply text based on the adjusted companionship dialogue strategy, convert it into robot voice, and play it to achieve personalized interaction with the single user.

[0049] In one example, when a single user repeatedly wakes up the robot for extended periods over several consecutive nights, saying things like "Talk to me a little longer" or "Don't stop," the robot marks these snippets as high-frequency emotional expressions or heavy robot interaction dependence, updating the dependence status record instead of directly altering the stable personality layer. Subsequently, the robot reduces personality-probing follow-up questions and continuous probing, and increases gentle summaries, pause suggestions, or guidance on realistic tasks.

[0050] In conjunction with any of the above embodiments, in one implementation, the present invention also provides an interactive method for an emotional companion robot for single users. In this method, step S12, "updating the routing judgment of the text personality evidence package for each conversation segment," may further include step S51: Step S51: Route textual personality evidence packages with semantic clarity below the semantic clarity threshold, or speech-to-text confidence below the confidence threshold, or sensitivity above the sensitivity threshold, or denied by the user, to the temporary storage layer.

[0051] In this embodiment, the evidence processing layer can determine the semantic clarity, speech-to-text confidence (i.e., ASR confidence), and sensitivity of the text personality evidence package. Then, it compares the semantic clarity with a preset semantic clarity threshold, the speech-to-text confidence with a preset confidence threshold, and the sensitivity with a preset sensitivity threshold. Text personality evidence packages with semantic clarity lower than the semantic clarity threshold, speech-to-text confidence lower than the confidence threshold, sensitivity higher than the sensitivity threshold, or denied by the user, are routed to the temporary storage layer. The temporary storage layer indicates that it does not update the long-term personality profile or the current interaction state. In other words, this embodiment places text personality evidence packages with unclear semantics, low ASR confidence, sensitivity, or denial by the user into the temporary storage layer.

[0052] In conjunction with any of the above embodiments, in one implementation, the present invention also provides an interactive method for an emotional companion robot for single users. In addition to the steps described above, this method may further include steps S61 to S63: Step S61: The voice interaction layer collects the corrected speech of single users and converts the corrected speech into corrected text.

[0053] In this embodiment, when the voice interaction layer recognizes a correction instruction from a single user regarding a previous user's voice, it collects the single user's corrective voice and converts it into corrective text. The correction instruction represents a viewpoint expressed by the user that contradicts the previous user's voice. For example, when the voice interaction layer detects keywords such as "Don't remember this," "I misspoke," "You misunderstood," or "I think I misremembered," it can be determined that the single user has issued a correction instruction regarding a previous user's voice. At this point, the voice interaction layer collects the relevant corrective voice from the single user based on the correction instruction and converts it into corrective text.

[0054] Step S62: The evidence processing layer locates the textual personality evidence package that has been denied by the user based on the correction text and notifies the profiling and control layer.

[0055] In this embodiment, the evidence processing layer can locate the textual personality evidence package that has been denied by the user, based on the correction text, and determine which layer the denied textual personality evidence package is routed to, thus obtaining the location result. The location result is then notified to the profiling and control layer. The location result includes: the textual personality evidence package denied by the user corresponds to the layer in the profiling and control layer.

[0056] Step S63: Based on the text personality evidence package that has been denied by the user and located by the evidence processing layer, the profiling and control layer cancels the long-term personality profile update triggered by the text personality evidence package that has been denied by the user and rolls back the long-term personality profile, rolls back the current interaction state, and / or rolls back the companionship dialogue strategy and TTS parameters for single users.

[0057] In this embodiment, the profiling and control layer determines the text personality evidence package denied by the user and its corresponding layer based on the positioning results. Based on the text personality evidence package denied by the user located by the evidence processing layer, it cancels the long-term personality profile update triggered by the text personality evidence package denied by the user and rolls back the long-term personality profile, rolls back the current interaction state, and / or rolls back the companion dialogue strategy and TTS parameters for single users.

[0058] Specifically, when the profiling and control layer determines that the layer corresponding to the textual personality evidence package denied by the user is a stable personality layer, it cancels the long-term personality profile update triggered by the denied textual personality evidence package and rolls back the long-term personality profile, as well as rolls back the companion dialogue strategy and / or TTS parameters generated by the long-term personality profile for single users. When the profiling and control layer determines that the layer corresponding to the textual personality evidence package denied by the user is a temporary state layer or a dependent state record layer, it cancels the current interaction state generated triggered by the denied textual personality evidence package and rolls back the current interaction state, as well as rolls back the companion dialogue strategy and / or TTS parameters generated by the current interaction state for single users. Therefore, the personality evidence in this embodiment is fragment-level, traceable, and rollbackable personality evidence formed in natural dialogue.

[0059] In addition, in one embodiment, the voice interaction layer can also notify the evidence processing layer of the user's deletion command or memory ban command. The evidence processing layer locates the corresponding text personality evidence package based on the deletion command or memory ban command and notifies the profiling and control layer. The profiling and control layer deletes the text personality evidence package located by the evidence processing layer, and cancels the long-term personality profile update triggered by the text personality evidence package located by the evidence processing layer and rolls back the long-term personality profile, rolls back the current interaction state, and / or rolls back the companion dialogue strategy and TTS parameters for single users.

[0060] In conjunction with any of the above embodiments, in one implementation, the present invention also provides an interactive method for an emotional companion robot for single users. In addition to the steps described above, this method may further include steps S71 to S73: Step S71: In the case of an evidence gap in a personality dimension, after updating the route judgment, the evidence processing layer also judges the timing of supplementary questioning based on the current topic, user burden, sensitivity and cooldown time.

[0061] In this embodiment, the personality profile is a record of personality tendencies maintained by the robot around the target user over a long period of time, preferably using the Big Five personality traits: Openness (O), Conscientiousness (C), Extraversion (E), Agreeableness (A), and Neuroticism (N). In addition, the personality dimensions of the profile can be replaced with other personality dimensions, communication style dimensions, or companionship compatibility dimensions, etc.

[0062] In this embodiment, after generating the text personality evidence package, the evidence processing layer will also determine whether the personality evidence of each personality dimension in the text personality evidence package is sufficient. When the evidence of certain personality dimensions is insufficient, the robot needs to determine "when to ask, what to ask, and when not to ask". Otherwise, actively asking questions may interrupt the venting process or cause additional burden in sensitive topics such as intimate relationships, loneliness, and stressful events.

[0063] Based on this, in this embodiment, for any personality dimension, if the evidence processing layer determines that there is an evidence gap in that personality dimension (such as when the corresponding amount of personality evidence is less than a preset threshold), after performing the update route judgment, it can also determine the timing of supplementary questioning based on the current topic, user burden, sensitivity and cooldown time in the text personality evidence package.

[0064] In one embodiment, such as Figure 2 As shown, Figure 2 This is a schematic diagram illustrating the structure of a personality evidence package according to an embodiment of the present invention. Figure 2 The textual personality evidence package includes: basic information, textual evidence, personality evidence, and processing records. Basic information includes: fragment number, session / user ID, time, topic, and ASR confidence level; textual evidence includes: fragment text, original source, robot question, and historical context summary; personality evidence includes: Big Five dimensions, evidence direction, evidence strength, confidence level, and status / trait markers; processing records include: update path, sensitive markers, valid / invalid status, profile version, and rollback records.

[0065] Step S72: If the user is deeply confiding, discussing sensitive relationship topics, or repeatedly seeking comfort, then issue a prohibition instruction to the voice interaction layer for personality supplementary questioning; if the current topic is related to a personality dimension with insufficient evidence, the user is not deeply confiding, and is not repeatedly seeking comfort, then issue a personality supplementary questioning instruction with the timing for supplementary questioning to the voice interaction layer.

[0066] In this embodiment, when the evidence processing layer determines, based on the text personality evidence package, that the user is deeply confiding, discussing sensitive relationship topics, or repeatedly seeking comfort, the evidence processing layer issues a prohibition instruction to the voice interaction layer to prevent the voice interaction layer from actively asking follow-up questions and to reduce continuous questioning.

[0067] Based on the textual personality evidence package, if the evidence processing layer determines that the current topic is related to the personality dimension with insufficient evidence, the user has not expressed themselves in depth, and the user has not repeatedly sought comfort, the evidence processing layer sends a personality supplementary question instruction with the timing of supplementary questions to the voice interaction layer, instructing the voice interaction layer to proactively ask supplementary questions based on the timing of supplementary questions.

[0068] Step S73: Based on the dialogue record, the voice interaction layer generates supplementary questions according to the personality supplementary question instruction, the companion dialogue strategy for single users and the TTS parameters, and outputs them through TTS.

[0069] In this embodiment, after the voice interaction layer receives the personality supplementary question instruction from the evidence processing layer, it can generate supplementary question text according to the dialogue record and the personality supplementary question instruction, and according to the companion dialogue strategy for single users. The supplementary question text is then converted into robot semantics according to the TTS parameters and played, so as to form an active supplementary question for single users.

[0070] In other words, in this embodiment, when there is insufficient personality evidence in the personality dimension and the time for supplementary questioning has arrived, personality question text (i.e., supplementary question text) can be generated for the personality dimension where there is insufficient personality evidence. For example, a low-intrusion question, i.e., supplementary question text, can be naturally extended from the current topic in the conversation segment to supplement personality evidence.

[0071] In one optional example, the supplementary question text used for proactive supplementary questions can be implemented from template questions, retrieval questions, generative questions, or a pre-set question library.

[0072] In conjunction with any of the above embodiments, in one implementation, the present invention also provides an interactive method for an emotional companion robot for single users. In addition to the steps described above, this method may further include steps S81 to S83: Step S81: The voice interaction layer collects the voice responses of single users to follow-up questions, converts the voice responses into text responses, and adds them to the dialogue record.

[0073] In this embodiment, single users can respond to the robot's proactive follow-up questions. The semantic interaction layer collects the single user's voice response to the follow-up questions, converts the voice response into text response via ASR and adds it to the dialogue record, and sends the dialogue record with the added text response to the evidence processing layer.

[0074] Step S82: The evidence processing layer divides the dialogue record after adding the response text into conversation fragments; performs text personality evidence analysis on the conversation fragments containing the response text, and generates a text personality evidence package for the conversation fragments containing the response text based on the evidence analysis results; updates the routing judgment for the text personality evidence package of the conversation fragments containing the response text, routing the text personality evidence package of the conversation fragments containing the response text that belongs to the long-term state and whose sensitivity is lower than the sensitivity threshold to the stable personality layer, routing the text personality evidence package of the conversation fragments containing the response text that belongs to the short-term state or involves stressful events to the temporary state layer, and routing the text personality evidence package of the conversation fragments containing the response text that belongs to the high-frequency emotional expression state to the dependent state record layer.

[0075] In this embodiment, step S82 is similar to the aforementioned step S12. The evidence processing layer can divide the received dialogue record after adding the reply text into conversation fragments (the conversation fragment division method at this time can refer to step S12) to obtain the fragment division result, which includes: conversation fragments containing reply text.

[0076] In this embodiment, the evidence processing layer performs text personality evidence analysis on the conversation fragments containing the response text, obtains the evidence analysis results of the conversation fragments containing the response text, and generates a text personality evidence package containing the response text based on the evidence analysis results of the conversation fragments containing the response text, so as to accumulate personality evidence for personality dimensions where personality evidence is insufficient. For the specific method of generating the text personality evidence package, please refer to step S12.

[0077] Next, the evidence processing layer updates the routing of text personality evidence packages containing conversation fragments with response text to determine whether the package is routed to the stable personality layer, the temporary state layer, or the dependent state recording layer. Specifically, stable, clear, and low-sensitivity text personality evidence packages containing response text are routed to the stable personality layer; that is, the evidence processing layer routes text personality evidence packages containing response text that belong to a long-term state and have a sensitivity below the sensitivity threshold to the stable personality layer. Text personality evidence packages containing response text that represent late-night venting and / or stressful events are routed to the temporary state layer; that is, the evidence processing layer routes text personality evidence packages containing response text that belong to a short-term state or involve stressful events to the temporary state layer. The evidence processing layer routes text personality evidence packages containing response text that belong to a high-frequency emotional venting state to the dependent state recording layer.

[0078] Step S83: The profiling and control layer updates the long-term personality profile based on the textual personality evidence package containing conversational fragments of response text in the stable personality layer.

[0079] In this embodiment, the profiling and control layer updates the long-term personality profile of single users only based on the text personality evidence package containing conversation fragments with response text in the stable personality layer.

[0080] In this embodiment, the supplementary question answer is not directly written into the stable personality layer. Instead, it is added to the dialogue record and still generates a text personality evidence package and executes the relevant update route. This is because the supplementary question answer may also have low confidence in ASR, sensitive content, or short-term status, and cannot be directly used to update the long-term personality profile.

[0081] In one embodiment, after obtaining a conversation fragment, semantic analysis can be performed on the conversation fragment to obtain a textual personality evidence package. Then, based on the textual personality evidence package, it can be determined whether to route to a stable personality layer or a temporary state layer, etc. Figure 3 As shown, Figure 3 This is a detailed schematic diagram illustrating an interactive method for an emotional companion robot for single users, according to an embodiment of the present invention. Figure 3 The robot comprises a voice interaction layer, an evidence processing layer, and a profiling and control layer. In the voice interaction layer, the user speaks, the robot collects the user's voice, performs ASR transcription to obtain the user's text, outputs the ASR confidence score, and then writes the user's text into the dialogue record. Based on the ASR confidence score and the dialogue record (including historical context), textual personality evidence analysis is performed. For example, the content is first segmented, and then the segmented conversation fragments are semantically analyzed to obtain the textual personality evidence package corresponding to the conversation fragment. This fragment-level textual personality evidence package ensures that the personality profile update has a clear source, evidentiary reason, processing path, and rollback basis.

[0082] After obtaining the textual personality evidence package, an update judgment route is performed: if the textual personality evidence package represents stable, clear, and low-sensitivity evidence, it enters the stable personality layer; if the textual personality evidence package represents short-term clarity or stress (such as late-night confiding, stressful events), it enters the temporary state layer; if the textual personality evidence package represents high-frequency continuous confiding, it enters the marked dependency state record to determine the interaction dependency state of single users and enters the dependency state record layer; if the textual personality evidence package represents insufficient evidence or low confidence (such as unclear semantics, low confidence in ASR, sensitive content, or content denied by the user), it enters the temporary storage area or the non-updating area. This establishes a clear processing mechanism for short-term or low-reliability fragments such as low confidence in ASR, user denial, late-night stress, and sensitive social topics, enabling evidence-level processing of sensitive topics, refusal to answer, deleted memories, and prohibited memories in long-term one-on-one companionship scenarios. Thus, this embodiment proposes an update routing method based on textual personality evidence packages such as ASR confidence, evidence strength, state / trait markers, deep confession markers, interaction dependency state markers, and sensitivity markers to determine the stable personality layer, temporary state layer, dependency state record layer, temporary confirmation, no update, or rollback.

[0083] Specifically, only personality evidence from the stable personality layer is used to update the long-term personality profile, and content that the user explicitly requests not to remember is excluded from the long-term personality profile. The temporary state layer and dependent state records are used to update the current interaction state. Based on the updated long-term personality profile and the current interaction state, a companion dialogue strategy and control parameters (TTS parameters) are generated. Then, the robot's response is generated based on the companion dialogue strategy, and TTS output is performed based on the control parameters to play the robot's voice.

[0084] Furthermore, when insufficient evidence is determined for a certain personality dimension based on the textual personality evidence package, the timing for proactive follow-up questioning is judged based on the evidence gap, user status, sensitivity, and cooldown time. When the time for proactive follow-up questioning arrives, a robot-generated follow-up question is generated based on a companion dialogue strategy, and TTS output is performed based on TTS parameters, thus realizing the robot's proactive follow-up questioning. For example, this embodiment can control whether the robot asks follow-up questions, how it asks follow-up questions, and when it does not ask follow-up questions based on personality evidence gaps, current topic, user burden, deep confiding status, interaction dependence status, sensitivity, historical refusal to answer, and cooldown time.

[0085] Furthermore, when a single user triggers a deletion operation, a correction operation, or issues a command to prohibit memory recognition, the robot performs a rollback process: it can roll back evidence packages and profile increments. For example, based on stable personality layers and temporary state layers, it can revoke personality profile updates triggered by user voice, and / or revoke companionship dialogue strategies and TTS parameters determined based on user voice.

[0086] In a specific example, when a single user discusses relationship difficulties or social pressure, the robot identifies it as a sensitive topic and marks the conversation segment as a sensitive state. At this point, the conversation segment enters the short-term companionship layer, is temporarily stored for confirmation, or is not updated, and is not directly written into the stable personality layer. The robot then determines that the single user's short-term companionship state is a sensitive state related to intimate relationships. The robot avoids drawing conclusions and uses neutral confirmation and boundary expressions.

[0087] In another specific example, when a single user casually mentions wanting to try a new interest during the day, but there is insufficient evidence regarding openness, the robot determines that the current topic is relevant, the user's burden is low, and there is no deep venting, interaction dependence, historical refusal to answer, or cooling-off restrictions. Therefore, it generates a low-intrusion follow-up question. After the single user answers, the robot then treats that answer as relevant evidence for openness.

[0088] In this embodiment, the robot first converts the user's speech into text using ASR (Automatic Speech Recognition). Then, combining the robot's questions, historical context, and ASR confidence levels, it extracts personality evidence and companionship status information to update the target user's long-term personality profile and adjusts subsequent responses and TTS (Text-to-Speech) output accordingly. Thus, this embodiment solves the problems of personality evidence management, deep venting state processing, interaction dependency control, and adaptive interaction loop in language dialogue by using a structured personality evidence package associated with robot questions, user responses, historical context, and ASR confidence levels, based on an intermediate link of "fragment-level text personality evidence package - update route - proactive follow-up question - dialogue / TTS control - user rollback." Furthermore, this embodiment proposes a method for generating companionship dialogue strategies, boundary prompts, dialogue control parameters, and TTS output parameters based on long-term personality profiles, temporary companionship states, and interaction dependency states, ensuring that personalization is not limited to changes in response content.

[0089] It should be noted that the robot of the present invention (i.e., the emotional companion robot) can be a desktop companion robot, a smart speaker-shaped robot, a robot with a screen, a mobile robot, or a robot system that is linked with a mobile application.

[0090] It should also be noted that all privacy-related data in this invention, such as "personality profile" and "dialogue," were obtained with the user's consent and comply with relevant laws, regulations, and privacy requirements.

[0091] It should be noted that, for the sake of simplicity, the method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments of the present invention are not limited to the described order of actions, because according to the embodiments of the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0092] The terms "first," "second," etc., used in the specification and claims of this invention are used to distinguish similar objects and are not used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0093] Based on the same inventive concept, another embodiment of the present invention provides a robot, such as... Figure 4 As shown, Figure 4 This is a schematic diagram of a robot according to an embodiment of the present invention. The robot includes a memory, a processor, and a program or instructions stored in the memory and executable on the processor. When the program or instructions are executed by the processor, they implement the steps of the emotional companion robot interaction method for single users described in any of the above embodiments of the present invention.

[0094] Based on the same inventive concept, another embodiment of the present invention provides a readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the emotional companion robot interaction method for single users as described in any of the above embodiments of the present invention. The processor is a processor in an electronic device (or the aforementioned robot). The readable storage medium includes a computer-readable storage medium, such as a computer read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk.

[0095] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of the present invention is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0096] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0097] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of the present invention.

Claims

1. A method for interacting with an emotional companion robot for single users, characterized in that, An application is made to an emotional companion robot, which includes: a voice interaction layer, an evidence processing layer, and a profiling and control layer; the profiling and control layer includes: a stable personality layer, a temporary state layer, a dependency state recording layer, and a temporary storage layer; the method includes: The voice interaction layer collects the voice of single users, converts the user's voice into user text, and adds it to the conversation record; The evidence processing layer divides the dialogue record into conversation segments; it performs text personality evidence analysis on each conversation segment, and generates a text personality evidence package for each conversation segment based on the evidence analysis results; it performs update routing judgment on the text personality evidence package of each conversation segment, routing text personality evidence packages belonging to long-term states and with sensitivity below the sensitivity threshold to the stable personality layer, routing text personality evidence packages belonging to short-term states or involving stressful events to the temporary state layer, and routing text personality evidence packages belonging to high-frequency emotional expression states to the dependency state recording layer; The profiling and control layer updates the long-term personality profile based on the textual personality evidence package in the stable personality layer; generates the current interaction state based on the textual personality evidence package in the temporary state layer and the dependent state record layer; and determines the companionship dialogue strategy for single users based on the long-term personality profile and the current interaction state.

2. The interactive method for emotional companionship robots for single users according to claim 1, characterized in that, The method further includes: The profiling and control layer determines the TTS parameters for single users based on long-term personality profiles and current interaction status; The profiling and control layer will feed back the companionship dialogue strategy and TTS parameters for single users to the voice interaction layer; Based on the dialogue record, the voice interaction layer generates a text response according to the companion dialogue strategy for single users and TTS parameters, and outputs it through TTS.

3. The interactive method for emotional companionship robots for single users according to claim 2, characterized in that, The method further includes: When the profiling and control layer recognizes that the current interaction state is a high-frequency emotional expression state, it triggers the voice interaction layer to provide a companionship boundary prompt. Based on the dialogue record, the voice interaction layer generates companionship boundary prompt text according to the companionship dialogue strategy for single users and TTS parameters, and outputs it through TTS.

4. The interactive method for emotional companionship robots for single users according to claim 1, characterized in that, The method further includes: When the profiling and control layer recognizes that the current interaction state is that the user is repeatedly seeking comfort or continuously engaging in deep conversation, it sends instructions to the voice interaction layer to reduce the depth of follow-up questions and the frequency of proactive greetings.

5. The interactive method for emotional companionship robots for single users according to claim 1, characterized in that, The evidence processing layer updates the routing decision for the text personality evidence package of each conversation segment, and also includes: Packages of textual personality evidence with semantic clarity below the semantic clarity threshold, or speech-to-text confidence below the confidence threshold, or sensitivity above the sensitivity threshold, or denied by the user, are routed to the temporary storage layer.

6. The interaction method for an emotional companion robot for single users according to claim 1 or 2, characterized in that, The method further includes: The voice interaction layer collects the corrected speech from single users and converts the corrected speech into corrected text. Based on the corrective text, the evidence processing layer locates the textual personality evidence package that has been denied by the user and notifies the profiling and control layer. Based on the text personality evidence package that has been denied by the user and located by the evidence processing layer, the profiling and control layer cancels the long-term personality profile update triggered by the text personality evidence package that has been denied by the user and rolls back the long-term personality profile, rolls back the current interaction state, and / or rolls back the companionship dialogue strategy and TTS parameters for single users.

7. The interactive method for emotional companionship robots for single users according to claim 2, characterized in that, The method further includes: In the case of an evidence gap in a personality dimension, after updating the routing judgment, the evidence processing layer also judges the timing of supplementary questioning based on the current topic, user burden, sensitivity and cooldown time. If the user is deeply confiding, discussing sensitive relationship topics, or repeatedly seeking comfort, a prompt to prohibit personality follow-up questions will be sent to the voice interaction layer; if the current topic is related to a personality dimension with insufficient evidence, and the user is not deeply confiding or repeatedly seeking comfort, a prompt to ask personality follow-up questions with the timing of follow-up questions will be sent to the voice interaction layer. Based on the dialogue record and personality-based supplementary question instructions, the voice interaction layer generates supplementary questions according to the companion dialogue strategy for single users and TTS parameters, and outputs them through TTS.

8. The interactive method for an emotional companion robot for single users according to claim 7, characterized in that, The method further includes: The voice interaction layer collects the voice responses of single users to follow-up questions, converts the voice responses into text responses, and adds them to the conversation history; The evidence processing layer divides the dialogue records after adding the response text into conversation fragments; it performs text personality evidence analysis on the conversation fragments containing the response text, and generates a text personality evidence package for the conversation fragments containing the response text based on the evidence analysis results; it then performs an update routing judgment on the text personality evidence package of the conversation fragments containing the response text, routing the text personality evidence package of the conversation fragments containing the response text that belongs to the long-term state and whose sensitivity is below the sensitivity threshold to the stable personality layer, routing the text personality evidence package of the conversation fragments containing the response text that belongs to the short-term state or involves stressful events to the temporary state layer, and routing the text personality evidence package of the conversation fragments containing the response text that belongs to the high-frequency emotional expression state to the dependent state record layer; The profiling and control layer updates the long-term personality profile based on the textual personality evidence package containing conversational fragments with responses from the stable personality layer.

9. A robot, characterized in that, It includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the emotional companion robot interaction method for single users as described in any one of claims 1 to 8.

10. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the emotional companion robot interaction method for single users as described in any one of claims 1 to 8.

Images