Session control method and device of pet robot, and pet robot

Abstract

This invention provides a conversation control method, device, and pet robot for a pet robot, belonging to the field of robot control technology. The method includes: acquiring current user interaction habit information and interaction state information; performing speech recognition on user input speech in the current conversation to obtain the user's speech semantic information; determining the confidence level for ending the conversation based on the current user's interaction habit information, interaction state information, and speech semantic information; and determining the control result of the current conversation based on the confidence level for ending the conversation. This invention comprehensively assesses the likelihood of the user wanting to end the conversation by considering the user's interaction habits, current interaction state, and the specific content of the speech, thereby avoiding misjudgments caused by relying solely on fixed time intervals, reducing the occurrence of erroneous conversation endings, and improving the user experience.

Description

Technical Field

[0001] This invention relates to the field of robot control technology, and in particular to a conversation control method, device, and pet robot for a pet robot. Background Technology

[0002] Pet companion robots are intelligent interactive devices designed specifically for children, providing emotional companionship and educational support through voice dialogue.

[0003] Currently, existing smart interactive devices such as pet companion robots typically use a time threshold to determine whether to end the current session. Specifically, existing pet companion robots start a timer after completing a broadcast or asking a question. If the user does not continue speaking within a preset time, it is assumed that the user has ended the interaction, and the current session is exited directly.

[0004] However, this method relies on a single fixed time for judgment. Since children often pause naturally when interacting with the device, this method is prone to misinterpreting the user's normal pause as an intention to end the session, thus leading to the wrong termination of the session, reducing the accuracy of session control, and affecting the user experience. Summary of the Invention

[0005] This invention provides a conversation control method, device, and pet robot for pet robots, which solves the defects of existing technologies that are prone to accidental termination of conversations, realizes a comprehensive judgment of the user's actual interaction state, avoids accidental termination of conversations, and improves the user's interactive experience.

[0006] This invention provides a conversation control method for a pet robot, comprising the following steps: Obtain information about the current user's interaction habits and interaction status; Speech recognition is performed on the user's input speech in the current session to obtain the speech semantic information of the current user; Based on the current user's interaction habit information, interaction status information, and voice semantic information, determine the confidence level for ending the current session; The control result of the current session is determined based on the session termination confidence level corresponding to the current session.

[0007] According to a conversation control method for a pet robot provided by the present invention, determining the confidence level for ending the current conversation based on the current user's interaction habit information, interaction state information, and voice semantic information includes: Extract the implicit end signal from the speech semantic information. The implicit end signal is semantic information that represents a decrease in the user's willingness to interact and does not explicitly indicate the end of the current session. Determine the first session termination score corresponding to the implicit termination signal and the second session termination score corresponding to the interaction state information; Based on the interaction habit information, determine the scoring weight corresponding to the implicit end signal; perform weighted fusion of the first end session score and the scoring weight to obtain a first judgment score; The first judgment score and the second session termination score are fused together to obtain the session termination confidence level corresponding to the current session.

[0008] According to a conversation control method for a pet robot provided by the present invention, the step of determining the control result of the current conversation based on the confidence level of the current conversation ending includes: If the confidence level for ending the session is less than the first judgment threshold, then the current session is maintained; If the confidence level for ending the session is greater than or equal to the first judgment threshold and less than the second judgment threshold, then a confirmation prompt message is output. Obtain the current user's feedback on the confirmation prompt, and determine the control result of the current session based on the feedback; If the confidence level for ending the session is greater than or equal to the second judgment threshold, then the current session is ended; Wherein, the first judgment threshold is less than the second judgment threshold.

[0009] According to a conversation control method for a pet robot provided by the present invention, the step of determining the control result of the current conversation based on the feedback includes: If the feedback indicates that the current user has an intention to engage in an interactive session, then the current session is maintained. If the feedback situation is determined to be that no feedback information is received within a preset time period, or if the feedback situation indicates that the current user does not have the intention to have an interactive session, then the current session is terminated.

[0010] According to a session control method for a pet robot provided by the present invention, after the current session ends, it further includes: Determine the reason for ending the current session; If the reason for ending the session indicates that no explicit intention to end the session has been received from the current user, then the pet robot is controlled to enter a shallow sleep state. The shallow sleep state refers to the state in which the pet robot retains the context information of the current session and opens a short-term session recovery window.

[0011] According to a conversation control method for a pet robot provided by the present invention, after controlling the pet robot to enter a shallow sleep state, it further includes: If a wake-up command is received within the short session recovery window, the current session is restored based on the context information.

[0012] According to a session control method for a pet robot provided by the present invention, after the current session ends, it further includes: Determine the reason for ending the current session; If the reason for ending the session indicates that the current user has clearly indicated their intention to end the session, then the pet robot is controlled to enter a deep sleep state. The deep sleep state refers to the state in which the pet robot releases the temporary resources corresponding to the current session and stops voice semantic processing.

[0013] According to a conversation control method for a pet robot provided by the present invention, the interaction habit information includes at least one of the following: The current user's age information; The current user's historical average session duration; The current user's daily short response frequency; The current user's common methods for ending a session.

[0014] According to a conversation control method for a pet robot provided by the present invention, the interaction state information includes at least one of the following: The current user's voice pause information; The information on the change in the distance between the current user and the sound source of the pet robot; The current user's attention orientation relative to the pet robot; The interaction frequency information of the current user within a preset time period.

[0015] The present invention also provides a conversation control device for a pet robot, comprising the following modules: The information acquisition module is used to acquire information about the current user's interaction habits and interaction status. The speech analysis module is used to perform speech recognition on the user's input speech in the current session to obtain the speech semantic information of the current user; The confidence determination module is used to determine the confidence level of ending the current session based on the current user's interaction habit information, interaction state information, and voice semantic information. The session control module is used to determine the control result of the current session based on the session termination confidence level corresponding to the current session.

[0016] The present invention also provides a pet robot, comprising: The interactive component includes a voice module and a display module. The voice module is used to collect the current user's input voice and output voice interaction information during the conversation. The display module is used to display the pet robot's current sleep state or interactive feedback screen. Sensing components are used to collect the physical interaction state data of the current user; A controller is connected to both the interaction component and the sensing component, and the controller is used to execute the conversation control method of any of the pet robots described above.

[0017] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the conversation control method of any of the pet robots described above.

[0018] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the conversation control method for a pet robot as described above.

[0019] The pet robot conversation control method, device, and pet robot provided by this invention comprehensively assess the likelihood of the user wanting to end the conversation by considering the user's interaction habits, current interaction state, and specific content of speech. This avoids misjudgments caused by relying solely on fixed time intervals, reduces the occurrence of accidental conversation terminations, and improves the user experience. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating the conversation control method for the pet robot provided by the present invention.

[0022] Figure 2 This is a flowchart illustrating the process for determining the confidence level of ending a session, as provided by the present invention.

[0023] Figure 3 This is a schematic diagram of the conversation control device for the pet robot provided by the present invention.

[0024] Figure 4This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this 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 this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0026] It should be noted that, in the description of this invention, the terms "comprising," "including," or any other variations thereof are intended to cover a 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 limitation, 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 said element.

[0027] The terms "first," "second," etc., used in this invention are used to distinguish similar objects and not 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, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more.

[0028] Figure 1 This is a flowchart illustrating the conversation control method for the pet robot provided by the present invention, as shown below. Figure 1 As shown, the execution subject of the conversation control method for pet robots provided by the present invention can be a controller of the pet robot, a cloud computing platform, or a computer capable of executing the method of the present invention, etc. Unless otherwise specified, the controller of the pet robot will be used as an example in the following embodiments.

[0029] As an optional embodiment, the conversation control method of this pet robot mainly includes, but is not limited to, the following steps: Step 110: Obtain the current user's interaction habits and interaction status information.

[0030] Interaction habit information refers to the historical behavioral characteristics and personal expression preferences exhibited by the current user in past interactions with the pet robot. For example, interaction habit information could include the user's age group, historical average conversation duration, frequency of short daily responses, and commonly used conversation ending methods.

[0031] Interaction habit information can be obtained by reading historical data bound to the current user's identity from the local storage after the pet robot device is started or the user wakes up the device. For example, if the current user is a child aged 3 to 5, their interaction habit information may record that the user usually uses short responses.

[0032] Interaction status information refers to real-time data that reflects a user's current level of engagement and physical behavior during the current session. For example, interaction status information can be the user's voice status, including whether the user is continuing to speak and the duration of pauses; it can also be the location or distance of the sound source; the user's attention level and interaction frequency over a recent period; and whether the user proactively asks new questions and the speed at which the device responds.

[0033] Interaction status information can be collected and calculated in real time during the conversation using sensing components such as microphone arrays and cameras mounted on the pet robot. For example, the microphone array can determine whether the user's voice is gradually moving away from the pet robot device, and the camera can determine whether the user has not been facing the pet robot device for a long time or has obviously shifted their attention.

[0034] Step 120: Perform speech recognition on the user's input speech in the current session to obtain the speech semantic information of the current user.

[0035] Voice semantic information refers to the text content and corresponding true intent contained in the user's current voice input. For example, voice semantic information can be explicit ending expressions, such as "bye-bye," "see you later," "I'm done chatting," "I'm going to sleep," or "I'm going to play." It can also be explicit expressions of refusal to continue interaction, such as "Don't talk to me anymore," "Don't bother me," or "You can play by yourself." It can also be implicit ending signals, such as repeatedly replying with only short content such as "um," "oh," or "okay," or indicating that the current task has been completed and the user has not asked any new questions.

[0036] Specifically, after receiving the user's voice audio signal, the speech recognition algorithm is used to convert the audio signal into corresponding text data. Then, a semantic analysis model is used to extract and classify the intent of the text data, thereby determining the specific content category of the user's current expression.

[0037] For example, when a user is identified saying "Let's chat next time," semantic analysis determines that this voice message belongs to the category of explicitly wanting to end the current conversation. When a user's repeated responses consist of only one or two words, semantic analysis determines that this voice message belongs to the category of implicit termination signals indicating a decreasing willingness to interact. In this way, the specific meaning of the user's speech can be understood, providing a content-level basis for subsequent decisions on whether to end the conversation.

[0038] Step 130: Based on the current user's interaction habits, interaction status, and voice semantic information, determine the confidence level for ending the current session.

[0039] Conversation end confidence refers to a numerical value or evaluation metric used to measure the likelihood that a user actually wants to end the chat. For example, conversation end confidence can be a specific numerical score or a probability percentage. It can be calculated by comprehensively analyzing the user's usual speaking habits, current physical state, and the content of what was just said.

[0040] Specifically, the intent contained in the speech semantic information can be extracted, combined with the user behavior reflected in the interaction state information, and adjusted by referring to the degree of influence of interaction habit information on these intents and behaviors, thereby calculating a comprehensive score. For example, when the speech semantic information indicates that the user continuously replies with "um," but the interaction habit information shows that the young child is used to such short answers, the influence of such short answers in the calculation can be reduced. In this case, if the interaction state information shows that the user is still facing the pet robot device, then the confidence level of the final determined end of the conversation will be relatively low.

[0041] Step 140: Determine the control result of the current session based on the session termination confidence level corresponding to the current session.

[0042] The control outcome refers to the specific action or working state that the pet robot device will perform in response to the current chat state. For example, the control outcome could be to continue the normal conversation, exit the current session, or output a confirmation message to the user.

[0043] Specifically, the calculated confidence level for ending the session is compared with a preset threshold, and the next step is determined based on the comparison result. For example, if the confidence level for ending the session is lower than the preset threshold, it means the user is likely to continue chatting, and the control result is to maintain the current session, with the pet robot device continuing to wait for the user to speak. If the confidence level for ending the session reaches the preset threshold, it means the user is more likely to end the chat, and the control result could be to end the current session, with the device stopping waiting and entering a sleep state. Through this multi-dimensional comprehensive judgment and comparison with thresholds, erroneous exits caused by short pauses in the user's chat can be avoided.

[0044] The conversation control method for pet robots provided by this invention comprehensively assesses the likelihood of a user wanting to end the conversation by considering the user's interaction habits, current interaction state, and the specific content of their speech. This avoids misjudgments caused by relying solely on fixed time intervals, reduces the chance of accidentally ending the conversation, and improves the user experience.

[0045] Figure 2 This is a flowchart illustrating the process for determining the confidence level for ending a session, as provided by the present invention. Figure 2 As shown, as another optional embodiment provided by the present invention, the confidence level for ending the current session is determined based on the current user's interaction habits, interaction status information, and voice semantic information, including but not limited to the following steps: Step 210: Extract the implicit end signal from the speech semantic information. The implicit end signal is semantic information that represents a decrease in the user's willingness to interact and does not explicitly indicate the end of the current session.

[0046] Implicit end signals refer to indirect cues indicating a decreasing willingness to interact without explicitly stating an intention to end the interaction. Examples include repeated responses like "um," "oh," or "okay," the completion of a question-and-answer session, story, or explanation without further questions, or prolonged periods without any new proactive interaction.

[0047] Implicit termination signals can be extracted by rule matching of text length, dialogue turn number, and intent classification results in speech semantic information. For example, after obtaining speech semantic information, it can be determined whether the user's most recent responses are all less than a preset number of characters. If so, the implicit termination signal of consecutive short responses can be extracted.

[0048] Step 220: Determine the first session termination score corresponding to the implicit termination signal and the second session termination score corresponding to the interaction state information.

[0049] The first conversation end score refers to a quantitative score assigned to implicit end signals extracted from a user's speech, used to measure the degree to which these signals contribute to the likelihood of ending the conversation. For example, the first conversation end score could be 10 to 25 points for consecutive short responses, or 20 points for the completion of the current task.

[0050] The first session termination score can be determined by looking up a pre-set weak signal scoring map. For example, after extracting the implicit termination signal that the current task has been completed and the user has not made any further requests, the first session termination score corresponding to the implicit termination signal is determined to be 20 points according to the preset scoring rules.

[0051] The second end-of-conversation score refers to a quantitative score assigned to the user's current physical behavior and interaction state, used to measure the degree of the user's intention to end the conversation as reflected in their external state. For example, the second end-of-conversation score could be 15 points for when the user's voice continues to move away from the pet robot device, 15 points for when the user no longer faces the pet robot device, 10 points for when the interaction frequency decreases due to not asking questions for a long time, or 5 points for when no user voice is detected for a period of time.

[0052] The second end-of-session score can be determined by comparing the real-time interaction status information with a preset status scoring standard. For example, if the camera detects that the user is no longer facing the pet robot device and their attention has shifted, then the second end-of-session score corresponding to this interaction status information is determined to be 15 points according to the scoring standard.

[0053] Step 230: Determine the scoring weight corresponding to the implicit end signal based on the interaction habit information; perform weighted fusion of the first end session score and the scoring weight to obtain the first judgment score.

[0054] The first judgment score refers to the revised score obtained by re-evaluating implicit termination signals after taking into account the user's personal daily expression habits. For example, the first judgment score can be a reduced score after weight discounting, or a increased score after weight amplification.

[0055] Specifically, the user's age group and historical daily short response frequency are read from the interaction habit information. Based on these habit characteristics, a weight coefficient is assigned to the currently extracted implicit end signal. Then, the weight coefficient is multiplied by the first end-of-session score to obtain the first judgment score.

[0056] For example, for children aged 3 to 5, if interaction habit information shows that the user frequently uses short phrases such as "um," "oh," and "okay" in normal communication, a smaller scoring weight, such as 0.5, will be assigned to this implicit end signal of consecutive short responses. Assuming the original first end-of-conversation score was 20 points, the weighted score after fusion would become 10 points, thus reducing the impact of consecutive short responses on the end-of-conversation judgment. Conversely, for users with strong expressive abilities and fewer short responses, a larger scoring weight, such as 1.2, will be assigned to consecutive short answers, thus appropriately increasing the impact of this signal on the end-of-conversation judgment and reducing misjudgments caused by differences in individual expression habits.

[0057] Step 240: The first judgment score and the second session termination score are fused to obtain the session termination confidence level corresponding to the current session.

[0058] Specifically, the first judgment score, which has been adjusted for user habits, is directly added to the second session termination score, which reflects the user's current physical behavior, to obtain a comprehensive total score. This total score is used as the session termination confidence level for the current session.

[0059] For example, a first judgment score of 10 points is calculated for consecutive short responses. Simultaneously, a second session termination score of 15 points is determined by sensors detecting that the user's voice is continuously moving away from the pet robot device. These 10 and 15 points are added together to obtain a total score of 25 points. This 25 points represents the session termination confidence level for the current session. By accumulating the scores for spoken content and external behavior, the system can comprehensively reflect the user's true interaction state, providing reliable data support for subsequent session control.

[0060] The conversation control method for pet robots provided by this invention extracts implicit end signals from speech semantic information and determines the corresponding first conversation end score. It then determines a second conversation end score by combining the interaction state information. Next, it assigns score weights to the implicit end signals based on the user's interaction habits, and then performs weighted fusion to obtain a first judgment score. This score is then fused with the second conversation end score to obtain the conversation end confidence level. This method can fully consider the individual expression differences of different users and perform personalized quantitative correction for unclear end intentions, thereby improving the accuracy of conversation end confidence level calculation and further reducing the probability of misjudgment caused by different user expression habits.

[0061] In another embodiment of the present invention, determining the control result of the current session based on the session termination confidence level corresponding to the current session includes: maintaining the current session if the session termination confidence level is less than a first judgment threshold; outputting a confirmation prompt if the session termination confidence level is greater than or equal to the first judgment threshold and less than a second judgment threshold; obtaining the current user's feedback on the confirmation prompt to determine the control result of the current session based on the feedback; and terminating the current session if the session termination confidence level is greater than or equal to the second judgment threshold; wherein the first judgment threshold is less than the second judgment threshold.

[0062] The first and second judgment thresholds refer to two reference scores used to classify the strength of a user's intention to end the chat. For example, the first judgment threshold can be a lower score boundary, representing the initial judgment line that the user may want to end the chat, while the second judgment threshold can be a higher score boundary, representing the confirmation line that the user is highly likely to want to end the chat.

[0063] A confirmation message is a voice prompt played proactively to the user to inquire about their true intentions when it's unclear whether they genuinely want to end the chat. For example, a confirmation message might be a simple question like, "Is there anything else you want to say?", "And then?", or "Shall we continue chatting?"

[0064] Feedback refers to the specific actions or responses a user takes after hearing a question. For example, feedback could be a user asking a new question, starting a new topic, or explicitly indicating they want to continue chatting; it could also be a user not answering, giving only very brief responses, or explicitly stating they don't want to continue chatting.

[0065] Specifically, the calculated confidence score for ending the session can be compared with a first judgment threshold and a second judgment threshold. If the confidence score is less than the first judgment threshold, it is considered that there is not enough information to indicate that the user wants to end the session, and the normal conversation continues. If the confidence score is greater than or equal to the first judgment threshold and less than the second judgment threshold, it indicates that the user may want to end the session but is not yet completely certain. In this case, the session is not immediately ended, but a confirmation prompt is played for simple confirmation. The user's response is then listened to, and a decision is made based on the user's subsequent feedback to resume the normal conversation or confirm the end of the current session.

[0066] If the confidence level for ending the session is greater than or equal to the second judgment threshold, it is considered that there is a high probability of ending the session, and the process of ending the chat is initiated directly. This secondary confirmation mechanism under uncertain conditions avoids abruptly exiting the pet robot device, making the process of ending the chat more natural.

[0067] The conversation control method for pet robots provided by this invention, by presetting a first judgment threshold and a second judgment threshold of different sizes, actively outputs a confirmation prompt when the confidence level for ending the conversation is in an uncertain state between the two thresholds, and decides whether to end the conversation based on the user's feedback. This introduces a secondary confirmation mechanism, avoids abruptly exiting when the user's intention cannot be fully determined, reduces the probability of erroneous termination caused by children thinking, playing, or pausing temporarily, and makes the process of ending the chat for pet robot devices more natural.

[0068] In another embodiment of the present invention, determining the control result of the current session based on the feedback includes: if the feedback indicates that the current user has an intention to have an interactive session, then maintain the current session; if the feedback indicates that no feedback information is received within a preset time period, or that the feedback indicates that the current user does not have an intention to have an interactive session, then terminate the current session.

[0069] Specifically, after outputting the confirmation message, the system listens for the user's response. If the user asks a new question, starts a new topic, or explicitly indicates they want to continue, it means the user still has the intention to interact. In this case, the current conversation is canceled, and the normal session resumes. For example, after playing "Shall we continue chatting?", if the user replies "I want to hear more stories," the current conversation is maintained. Conversely, if the user does not respond within the set 60 seconds, continues to give only brief responses of one or two words, or explicitly indicates they do not want to continue, it means the user genuinely has no intention of continuing the conversation. In this case, the current conversation is confirmed to end.

[0070] It should be noted that, considering that children often experience brief shifts in attention, searching for toys, or pauses in thought during actual use, exiting the conversation based on only one initial judgment could easily disrupt their normal experience. Therefore, this invention adds a simple inquiry in uncertain situations and determines subsequent actions based on the user's actual feedback, thereby effectively reducing the probability of misjudgment and ensuring the continuity of multiple rounds of conversation.

[0071] The conversation control method for pet robots provided by this invention, after obtaining user feedback, performs operations to maintain or end the current conversation based on whether the user intends to engage in conversation or whether feedback information is received within a preset time period. This method can accurately identify the user's true intention after secondary confirmation, ensuring that the normal conversation is resumed in a timely manner when the user still wants to chat, and the conversation is decisively ended when the user has no intention or does not respond, thus further improving the accuracy and rationality of conversation state control.

[0072] In another embodiment of the present invention, after ending the current session, the method further includes: determining the session termination reason for ending the current session; if the session termination reason indicates that no explicit termination intention from the current user has been received, then controlling the pet robot to enter a shallow sleep state; wherein, the shallow sleep state is a state in which the pet robot retains the context information of the current session and opens a short-term session recovery window.

[0073] Specifically, after determining that the chat has ended, the specific triggering conditions that led to its termination are analyzed. If the termination was due to the user temporarily ceasing to speak, moving away, shifting attention, decreasing interaction frequency, or not receiving a response after secondary confirmation, it indicates that the user did not explicitly express an intention to end the chat. In this case, the pet robot device is controlled to enter a shallow sleep state. In this state, the context of the current conversation, the content of the most recent rounds of dialogue, and the current topic information are temporarily saved, and a short waiting window is set. For example, if the user did not reply within the timeout period because they went to get a toy, the pet robot device ends the conversation and enters a shallow sleep state, while simultaneously opening a short conversation recovery window of 3-10 minutes. During these 3-10 minutes, the pet robot device will retain the context information of the previous conversation.

[0074] It should be noted that, considering that children often leave briefly when interacting with pet robot devices due to external interference or searching for items, if all chat memories are cleared directly after each session ends, the child will need to re-establish the topic upon returning, resulting in an uncoordinated communication process. Therefore, this invention employs a shallow sleep mode and retains the context for a short period when the conversation does not explicitly end, thereby reducing the need for users to re-describe previous content and improving the continuity of multi-round chats.

[0075] As an optional embodiment, after controlling the pet robot to enter a shallow sleep state, the method further includes: if a wake-up command is received within a short session recovery window, then the current session is resumed based on context information.

[0076] Specifically, during the period of shallow sleep, if a wake-up command is received from the user before the end of the set short session recovery window, the previously saved context information, recent conversation content and topic information are extracted, and the conversation scene is reconstructed based on this information, and then the interaction continues with the previous chat content.

[0077] For example, if a user goes to get a toy and returns within 5 minutes, sending a wake-up command to the pet robot device, the robot will directly load the context of the topic being discussed and continue chatting with the user from there, without requiring the user to explain where they left off.

[0078] The pet robot conversation control method provided by this invention determines the reason for ending the current conversation. If no explicit intention to end the conversation is received from the current user, the pet robot is controlled to enter a shallow sleep state that retains context information and opens a short conversation recovery window. When a wake-up command is received within the short conversation recovery window, the current conversation is recovered based on the context information. This method enables hierarchical state management based on different reasons for ending the conversation. It retains necessary chat memories when the user may only be temporarily interrupting the chat, so that when the user wakes up the pet robot again in a short time, the previous topic can be resumed directly. This reduces the need for the user to re-describe the context and improves the continuity of multi-round conversations.

[0079] In another embodiment of the present invention, after ending the current session, the method further includes: determining the session termination reason for ending the current session; if the session termination reason indicates that the current user has been given a clear intention to end the session, then controlling the pet robot to enter a deep sleep state; wherein, the deep sleep state is the state in which the pet robot releases the temporary resources corresponding to the current session and stops voice semantic processing.

[0080] Specifically, after determining that the chat has ended, the specific triggering conditions that led to its termination are analyzed. If the user explicitly said "goodbye" or "I don't want to chat anymore," it means the user genuinely doesn't want to continue the conversation. In this case, the pet robot device is put into a deep sleep state. In this state, temporary resources such as cache used during the current chat are cleared, and background voice recognition and semantic analysis functions are disabled, retaining only basic low-power wake-up functions. For example, when the system recognizes that the user has explicitly said "goodbye" or "I don't want to chat anymore," confirming the user's clear intention to end the chat, the pet robot device is then put into a deep sleep state, releasing the temporary resources used during the chat and ceasing continuous voice processing.

[0081] It should be noted that, considering that if the pet robot device continues to maintain context memory and run the voice processing program in the background after the user has clearly expressed the intention to end the chat, it will cause unnecessary power consumption and memory occupation. Therefore, this invention adopts deep sleep and releases relevant resources when it receives a clear intention to end the chat, thereby reducing the power consumption and computing resource occupation of the pet robot device while ensuring user experience.

[0082] The pet robot conversation control method provided by this invention determines the reason for ending the conversation after the current conversation ends. If the reason for ending the conversation indicates that the current user has clearly indicated their intention to end the conversation, the pet robot is controlled to enter a deep sleep state that releases the temporary resources corresponding to the current conversation and stops voice and semantic processing. This method can perform targeted state management based on the clear reason for ending the conversation. When it is confirmed that the user will no longer continue chatting, the system resources are released in a timely manner and the operating power consumption is reduced, thereby reasonably reducing the device's computing resource usage.

[0083] In another embodiment provided by the present invention, the interaction habit information includes at least one of the following: the current user's age information; the current user's historical average session duration; the current user's daily short response frequency; and the current user's common methods for ending sessions.

[0084] Specifically, after the pet robot device is started or the user wakes it up, it loads historical profile data corresponding to the current user's identity. By acquiring this data, it can understand the user's usual expression characteristics so as to make targeted adjustments when determining whether the chat has ended.

[0085] For example, if the age information indicates the current user is a young child aged 3 to 5, and their daily short responses are frequent, it suggests that the user habitually uses short words like "um," "oh," or "okay" to answer. In subsequent calculations, the impact of these short responses on the decision to end the conversation will be reduced. Conversely, if the daily short response frequency is low, it indicates the user has strong expressive abilities; therefore, when consecutive short responses occur, the influence of this signal on the decision to end the conversation will be appropriately increased. Furthermore, the historical average conversation duration can be used to assess whether the current chat has reached the user's usual chat duration, or personalized ending phrases can be accurately identified based on common conversation ending methods.

[0086] In another embodiment provided by the present invention, the interaction state information includes at least one of the following: the current user's voice pause information; the change in the distance between the current user and the sound source relative to the pet robot; the current user's attention orientation relative to the pet robot; and the current user's interaction frequency information within a preset time period.

[0087] Specifically, during the chat between the user and the device, sensor components such as microphone arrays and cameras collect data in real time that reflects the user's current physical behavior and level of participation, and use this data as the basis for subsequent comprehensive judgment on whether the chat has ended.

[0088] For example, it can detect whether a user is currently speaking and for how long, thus obtaining information about speech pauses. A microphone array can determine if the user's voice is gradually moving away from the device; if the sound continues to move away, it provides information about changes in sound source distance. A camera can determine if the user has been away from the device for an extended period or if their attention has noticeably shifted, thus providing information about the direction of attention. Furthermore, it can analyze whether the user has proactively asked new questions, responded to the device quickly, and remained inactive for extended periods in recent times, thus providing information about interaction frequency. This information can reflect the user's true interaction status from multiple dimensions, avoiding the risk of a single signal triggering an end to the interaction.

[0089] The conversation control method for pet robots provided by this invention, by acquiring at least one of the following interaction habit information, including the current user's age, historical average conversation duration, daily short response frequency, and common conversation ending methods, and at least one of the following interaction state information, including the current user's voice pause information, sound source distance change information, attention orientation information, and interaction frequency information, can more comprehensively and meticulously depict the user's true interaction intention from both the user's historical personalized expression preferences and current multi-dimensional physical behavior. This provides a rich and reliable data foundation for the subsequent accurate calculation of conversation ending confidence, further reducing the risk of misjudgment caused by single-dimensional judgment.

[0090] Figure 3 This is a schematic diagram of the conversation control device for the pet robot provided by the present invention, as shown below. Figure 3 As shown, it mainly includes, but is not limited to: The information acquisition module 310 is used to acquire the current user's interaction habits and interaction status information.

[0091] The speech analysis module 320 is used to perform speech recognition on the user's input speech in the current session to obtain the speech semantic information of the current user.

[0092] The confidence determination module 330 is used to determine the confidence level of ending the current session based on the current user's interaction habits, interaction status information and voice semantic information.

[0093] The session control module 340 is used to determine the control result of the current session based on the session termination confidence level corresponding to the current session.

[0094] It should be noted that the conversation control device for the pet robot provided by the present invention can execute the conversation control method for the pet robot described in any of the above embodiments during actual operation, which will not be elaborated in this embodiment.

[0095] The conversation control device for pet robots provided by this invention comprehensively assesses the likelihood of a user wanting to end the conversation by considering the user's interaction habits, current interaction status, and the specific content of their speech. This avoids misjudgments caused by relying solely on fixed time intervals, reduces the chance of accidentally ending the conversation, and improves the user experience.

[0096] The present invention also provides a pet robot, comprising: The interactive components include a voice module and a display module. The voice module is used to collect the current user's voice input and output voice interaction information during the conversation; the display module is used to display the pet robot's current sleep state or interactive feedback screen. Sensing components are used to collect data on the current user's physical interaction status. The controller is connected to the interaction component and the sensing component respectively. The controller is used to execute the conversation control method of the pet robot described in any of the above embodiments, which will not be described in detail in this embodiment.

[0097] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 4 As shown, the electronic device may include a processor 410, a communications interface 420, a memory 430, and a communication bus 440. The processor 410, communications interface 420, and memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a conversation control method for the pet robot. This method includes: acquiring the current user's interaction habits and interaction status information; performing speech recognition on the user's input speech in the current conversation to obtain the current user's speech semantic information; determining the confidence level for ending the current conversation based on the current user's interaction habits, interaction status information, and speech semantic information; and determining the control result of the current conversation based on the confidence level for ending the current conversation.

[0098] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0099] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the conversation control method for a pet robot provided by the above methods. The method includes: acquiring the current user's interaction habit information and interaction state information; performing speech recognition on the user's input speech in the current conversation to obtain the current user's speech semantic information; determining the confidence level for ending the conversation corresponding to the current conversation based on the current user's interaction habit information, interaction state information, and speech semantic information; and determining the control result of the current conversation based on the confidence level for ending the conversation corresponding to the current conversation.

[0100] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements a conversation control method for a pet robot provided by the methods described above. The method includes: acquiring current user interaction habit information and interaction state information; performing speech recognition on user input speech in the current conversation to obtain current user speech semantic information; determining the confidence level for ending the current conversation based on the current user interaction habit information, interaction state information, and speech semantic information; and determining the control result of the current conversation based on the confidence level for ending the current conversation.

[0101] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0102] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0103] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A conversation control method for a pet robot, characterized in that, include: Obtain information about the current user's interaction habits and interaction status; Speech recognition is performed on the user's input speech in the current session to obtain the speech semantic information of the current user; Based on the current user's interaction habit information, interaction status information, and voice semantic information, determine the confidence level for ending the current session; The control result of the current session is determined based on the session termination confidence level corresponding to the current session.

2. The conversation control method for a pet robot according to claim 1, characterized in that, The step of determining the session termination confidence level corresponding to the current session based on the current user's interaction habit information, interaction state information, and voice semantic information includes: Extract the implicit end signal from the speech semantic information. The implicit end signal is semantic information that represents a decrease in the user's willingness to interact and does not explicitly indicate the end of the current session. Determine the first session termination score corresponding to the implicit termination signal and the second session termination score corresponding to the interaction state information; Based on the interaction habit information, determine the scoring weight corresponding to the implicit end signal; perform weighted fusion of the first end session score and the scoring weight to obtain a first judgment score; The first judgment score and the second session termination score are fused together to obtain the session termination confidence level corresponding to the current session.

3. The conversation control method for a pet robot according to claim 1, characterized in that, The step of determining the control result of the current session based on the session termination confidence level corresponding to the current session includes: If the confidence level for ending the session is less than the first judgment threshold, then the current session is maintained; If the confidence level for ending the session is greater than or equal to the first judgment threshold and less than the second judgment threshold, then a confirmation prompt message is output. Obtain the current user's feedback on the confirmation prompt, and determine the control result of the current session based on the feedback; If the confidence level for ending the session is greater than or equal to the second judgment threshold, then the current session is ended; Wherein, the first judgment threshold is less than the second judgment threshold.

4. The conversation control method for a pet robot according to claim 3, characterized in that, Determining the control result of the current session based on the feedback includes: If the feedback indicates that the current user has an intention to engage in an interactive session, then the current session is maintained. If the feedback situation is determined to be that no feedback information is received within a preset time period, or if the feedback situation indicates that the current user does not have the intention to have an interactive session, then the current session is terminated.

5. The conversation control method for a pet robot according to claim 3, characterized in that, After the current session ends, the following is also included: Determine the reason for ending the current session; If the reason for ending the session indicates that no explicit intention to end the session has been received from the current user, then the pet robot is controlled to enter a shallow sleep state. The shallow sleep state refers to the state in which the pet robot retains the context information of the current session and opens a short-term session recovery window.

6. The conversation control method for a pet robot according to claim 5, characterized in that, After controlling the pet robot to enter a light sleep state, the following is also included: If a wake-up command is received within the short session recovery window, the current session is restored based on the context information.

7. The conversation control method for a pet robot according to claim 3, characterized in that, After the current session ends, the following is also included: Determine the reason for ending the current session; If the reason for ending the session indicates that the current user has clearly indicated their intention to end the session, then the pet robot is controlled to enter a deep sleep state. The deep sleep state refers to the state in which the pet robot releases the temporary resources corresponding to the current session and stops voice semantic processing.

8. The conversation control method for a pet robot according to claim 1, characterized in that, The interaction habit information includes at least one of the following: The current user's age information; The current user's historical average session duration; The current user's daily short response frequency; The current user's common methods for ending a session.

9. The conversation control method for a pet robot according to claim 1, characterized in that, The interaction status information includes at least one of the following: The current user's voice pause information; The information on the change in the distance between the current user and the sound source of the pet robot; The current user's attention orientation relative to the pet robot; The interaction frequency information of the current user within a preset time period.

10. A conversation control device for a pet robot, characterized in that, include: The information acquisition module is used to acquire information about the current user's interaction habits and interaction status. The speech analysis module is used to perform speech recognition on the user's input speech in the current session to obtain the speech semantic information of the current user; The confidence determination module is used to determine the confidence level of ending the current session based on the current user's interaction habit information, interaction state information, and voice semantic information. The session control module is used to determine the control result of the current session based on the session termination confidence level corresponding to the current session.

11. A pet robot, characterized in that, include: The interactive component includes a voice module and a display module. The voice module is used to collect the current user's input voice and output voice interaction information during the conversation. The display module is used to display the pet robot's current sleep state or interactive feedback screen. Sensing components are used to collect the physical interaction state data of the current user; A controller is connected to both the interaction component and the sensing component, and the controller is used to execute the conversation control method of the pet robot as described in any one of claims 1 to 9.

12. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the conversation control method for the pet robot as described in any one of claims 1 to 9.

13. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the conversation control method for the pet robot as described in any one of claims 1 to 9.

Images