Interaction system, method and control device coupled with artificial intelligence conversation state

Abstract

The application provides an interactive system, method and control device coupled with an artificial intelligence conversation state, the interactive system comprising a control device in communication connection with a host system with an artificial intelligence processing module built-in, the control device comprising: an operation module for detecting a physical operation action of a user; a control unit for sending the physical operation action to a command analysis module; the command analysis module is arranged in the control device or the host system, and is used for determining a conversation state of the artificial intelligence processing module in the host system, and mapping the physical operation into a corresponding control command signal according to the conversation state, wherein the same physical operation action is mapped into different control command signals in different conversation states; the host system is used for receiving the control command signal, and controlling the artificial intelligence processing module to perform a corresponding interactive operation in response to the control command signal, and updating the conversation state of the artificial intelligence processing module.

Description

Technical Field

[0001] This application mainly relates to the field of human-computer interaction technology, and in particular to an interactive system, method and control device coupled with artificial intelligence session state, and especially to a technical solution that introduces session state information in the process of control command parsing to realize dynamic command interpretation and interactive mode control based on session state. Background Technology

[0002] With the popularization of artificial intelligence technology, users hope to engage in multi-round dialogues with AI systems in a natural, efficient, and low-interference manner, enabling the systems to understand user intentions and respond in real time.

[0003] However, current AI interaction methods still primarily rely on keyboards, mice, touchscreens, or fixed voice input devices. These input tools are designed for visually guided fine-grained operations, requiring users to look at the screen and perform actions such as clicking, dragging, or touching to confirm input. In AI interaction scenarios, a complete session often involves multiple operation steps, such as wake-up / trigger (after which input usually starts automatically), end / stop, confirm / send, cancel / retry, mode switching, and task execution. Each of these operations typically corresponds to an independent physical action (button, click, or touch). Even the simplest voice session requires at least three different physical operations: "start," "end," and "confirm." In multi-turn or complex sessions, additional functions such as "delete," "modify," "retry," and "mode switching" may also be involved, often requiring switching or combining inputs between different interfaces. The overall operation path is long, involves many steps, is inefficient, and prone to errors.

[0004] Furthermore, existing input devices are mostly fixed or rely on display interfaces for interaction. Users must look at the screen and perform fine-grained operations to complete the interaction. This interaction mode is not only cumbersome and inefficient, but also difficult to use in scenarios such as mobile, driving, meetings, or outdoors. It lacks portability and blind operation capabilities, making it difficult to meet the efficiency and flexibility requirements of artificial intelligence interaction.

[0005] Therefore, there is an urgent need in this field for a new type of artificial intelligence interaction mechanism that can significantly reduce the types of physical operations required for interaction without changing the conversation logic, and enable on-the-go access and blind operation in a wearable form, thereby comprehensively improving the efficiency, naturalness and ease of use of artificial intelligence human-computer interaction. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide an interactive system, method and control device coupled with the artificial intelligence session state. Through the command interpretation mechanism driven by the session state, the entire artificial intelligence session can be completed with minimal physical operations. It supports blind operation and wearable on-the-go invocation, thereby significantly improving the efficiency, continuity and ease of use of artificial intelligence interaction.

[0007] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0008] In a first aspect, this application provides an interactive system coupled with an artificial intelligence session state, comprising: a control device communicatively connected to a host system with a built-in artificial intelligence processing module; the control device comprising: an operation module for detecting a user's physical operation actions; a control unit for sending the physical operation actions to a command parsing module; a command parsing module, located in the control device or the host system, for determining the session state of the artificial intelligence processing module in the host system, mapping the physical operation to a corresponding control command signal according to the session state, wherein the same physical operation action is mapped to different control command signals in different session states; and a host system for receiving the control command signals, and responding to the control command signals to control the artificial intelligence processing module to execute corresponding interactive operations, and updating the session state of the artificial intelligence processing module.

[0009] Secondly, this application provides an interaction method coupled with an artificial intelligence session state, comprising: a control device detecting a user's physical operation; a command parsing module determining the session state of an artificial intelligence processing module in a host system, and mapping the physical operation to a corresponding control command signal according to the session state, wherein the same physical operation is mapped to different control command signals in different session states; the host system receiving the control command signal, and controlling the artificial intelligence processing module to perform a corresponding interactive operation in response to the control command signal, and updating the session state of the artificial intelligence processing module.

[0010] Thirdly, this application provides a control device coupled with an artificial intelligence session, comprising an operation module for detecting the user's physical operation actions; a communication module for establishing a communication connection with a host system having a built-in artificial intelligence processing module; and a command parsing module for determining the session state of the artificial intelligence processing module in the host system, mapping the physical operation actions to corresponding control command signals based on the session state, and sending the control command signals to the host system through the communication module to drive the host system to respond to the control command signals, control the artificial intelligence processing module to perform corresponding interactive operations, and update the session state of the artificial intelligence processing module; wherein the same physical operation action is mapped to different control command signals in different session states.

[0011] Compared with the prior art, this application has the following advantages:

[0012] 1. This invention binds the semantics of physical operations to the real-time state of the user's interaction with AI voice through a session state-driven command interpretation mechanism. This allows the same physical operation to be interpreted as different command semantics in different session states. Users can complete the entire process from wake-up, end, confirmation, cancellation, and mode switching with minimal physical operations. In typical scenarios, the required types of physical operations can be compressed from 3-8 to 1-3, achieving semantic reuse and automatic logic adaptation under state-driven operation, significantly improving interaction efficiency.

[0013] 2. The control logic of this invention does not rely on a display interface, allowing users to complete all interactions without looking at the screen. Unlike traditional operation methods that rely on visual positioning, such as keyboards, mice, and touchscreens, the control device of this invention can achieve blind operation in screenless environments such as driving, walking, meetings, or outdoors. Users can complete interactive command input solely through tactile or pressure feedback, significantly improving safety and scene adaptability.

[0014] 3. By maintaining the state and mode flags of AI interactive sessions, this invention enables users to naturally switch between text input and command control modes within the same round of conversation, maintaining semantic context consistency and smooth operation, meeting the needs of multi-intent and multi-stage human-computer hybrid dialogue, and improving the naturalness and intelligence level of the overall interactive experience. Attached Figure Description

[0015] The accompanying drawings are included to provide a further understanding of this application. They are incorporated into and constitute a part of this application. The drawings illustrate embodiments of this application and, together with this specification, serve to explain the principles of this application.

[0016] Figure 1 This is a schematic diagram of an interactive system coupled with an artificial intelligence session state according to an embodiment of this application.

[0017] Figure 2 This is a system block diagram of a control device according to an embodiment of this application.

[0018] Figure 3 This is a flowchart of an interaction method coupled with artificial intelligence session state according to an embodiment of this application.

[0019] Figure 4 This is a schematic diagram of the mode switching process in a user-AI voice conversation according to an embodiment of this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0021] This application provides an interactive system, method, and control device coupled with the conversational state of artificial intelligence. Its core concept lies in constructing a closed-loop state coupling between the control device and the host system (with a built-in artificial intelligence module), enabling the operational semantics of the control device to dynamically change according to the real-time state of the artificial intelligence, thereby achieving intelligent, precise, and smooth human-computer interaction control.

[0022] Figure 1 This is a schematic diagram of an interactive system coupled with an artificial intelligence session state according to an embodiment of this application. Figure 1 As shown, the interactive system coupled with the AI ​​session state includes: a control device 100, a host system 200, and a command parsing module 300. The control device 100 is communicatively connected to the host system 200, and the communication connection can be wireless or wired.

[0023] The host system 200 can be a PC, mobile phone, or server, etc. It runs a session management module 210, a speech recognition module 220 (ASR engine), and an artificial intelligence processing module 230 (such as large language model applications).

[0024] Preferably, the control device 100 is a wearable device equipped with a microphone. This allows the user to wear it naturally close to their mouth, facilitating the acquisition of voice signals as a near-field microphone and improving the voice signal-to-noise ratio. The wearable device may take the form of, but is not limited to, a ring, a wristband, or a pendant.

[0025] In some embodiments, the command parsing module 300 is located in the control device 100; in other embodiments, the command parsing module 300 is located in the host system 200.

[0026] Figure 2 This is a system block diagram of a control device according to an embodiment of this application. In this embodiment, the control device 100 is a smart ring. The smart ring includes a body and a ring band 160. The body integrates a control unit 110, an operation module 120, a communication module 130, a microphone 140, a status feedback module 150, and a command parsing module 300. The microphone 140 is an optional component used to collect the user's voice signal; the status feedback module 150 is also an optional component used to generate corresponding physical feedback signals to inform the user of the session or command execution status.

[0027] In one embodiment, the smart ring can collect voice signals via its own microphone 140; in another embodiment, the voice signals can also be collected by the microphone of the host system 200.

[0028] The operation module 120 is configured to detect the user's physical operation actions (such as single click, double click, long press or swipe, etc.).

[0029] Control unit 110 is configured to send the physical operation actions detected by operation module 120 to command parsing module 300.

[0030] The command parsing module 300 is configured to determine the session state of the artificial intelligence processing module 230 in the host system 200, and map physical operations to corresponding control command signals according to the session state. The same physical operation is mapped to different control command signals in different session states. For example, when the artificial intelligence is in the "waiting for confirmation" state, a click operation can be mapped to a "confirm result command"; when the artificial intelligence is in the "idle" state, a click operation is mapped to "wake up voice recognition input".

[0031] The host system 200 is configured to receive control command signals from the command parsing module 300, and in response to the control command signals, control the artificial intelligence processing module 230 to perform corresponding interactive operations (such as waking up voice recognition input, ending voice recognition input, confirming / sending, canceling, modifying or switching modes, etc.), and update the session state of the artificial intelligence processing module.

[0032] Figure 3 This is a flowchart of an interaction method coupled with artificial intelligence session state according to an embodiment of this application. This interaction method is applicable to... Figure 1 The interactive system shown is coupled with the state of an artificial intelligence session. For example... Figure 3As shown, the interaction methods coupled with the AI ​​session state include:

[0033] Step S31: The control device detects the user's physical operation actions.

[0034] User physical actions include, but are not limited to, single-click, double-click, and long-press.

[0035] Step S32: The command parsing module determines the session state of the artificial intelligence processing module in the host system.

[0036] The session states include, but are not limited to, idle state, listening state, speech recognition state, AI processing state, and waiting for confirmation state.

[0037] Optionally, the command parsing module 300 determines the session state between the user and the artificial intelligence processing module in the host system 200 in the following ways:

[0038] 1) Receive status information from host system 200 via communication module 130, and determine session status based on the status information; and / or

[0039] 2) The command parsing module 300 infers the session state based on its internally recorded interaction history and / or operation timing context. This mechanism is key to ensuring that the system can maintain state coupling and provide a reliable user experience even under non-ideal conditions such as unstable communication connections, delayed or lost host state feedback.

[0040] In some embodiments, the command parsing module 300 internally maintains a state inference engine, which relies on an internal state machine, interaction history logs, and predefined timeout thresholds. The command parsing module 300 itself maintains an estimate of the host session state, referred to as the inferred state. This state machine largely corresponds to the host's state machine. Secondly, the control unit caches key interaction events for a period of time. These events include: the last command sent, the timestamp of the last command sent, the last received official status, and local microphone activity detection.

[0041] The following examples illustrate how the inference logic works:

[0042] Scenario 1: If the command parsing module has just sent a voice recognition wake-up command, it can infer that the current state should be a listening state. Even if the user does not see screen feedback or the host status message is lost, the command parsing module unilaterally assumes that the session has started and is ready to interpret subsequent physical operations based on the listening state.

[0043] Scenario 2: The last command sent by the command parsing module is "voice recognition wake-up command", inferring that the state is listening. However, after a long period of time, neither the user's voice is detected through the microphone nor any state update is received from the host. The command parsing module infers that this wake-up failed to start a valid session (the user may have changed their mind, or the host may not have responded successfully). Therefore, it automatically returns to the idle state to avoid the system "getting stuck" in the listening state.

[0044] Scenario 3: If a user performs an operation that is usually only effective when the AI ​​is in a state of waiting for confirmation (such as clicking to confirm), and the command parsing module does not receive this state, it can infer that the user may be in a state of waiting for confirmation based on historical commands (such as the previous command that triggered an AI response). Therefore, the user's click operation will be mapped to a confirmation command instead of other commands.

[0045] Step S33: The command parsing module maps physical operations to corresponding control command signals according to the session state. The same physical operation is mapped to different control command signals in different session states.

[0046] In some embodiments, the mapping step includes: querying a predefined session state-operation action-command mapping table based on the physical operation action and session state; and obtaining the control command signal corresponding to the current session state and physical operation action from the mapping table.

[0047] When the operation module 120 of the control device 100 is a single physical button or capacitive touch area, the detectable physical operation actions include single click, double click, long press, etc., and the corresponding session state-operation action-command mapping table can be Table 1:

[0048] Table 1. Session Status-Operation Action-Command Mapping Table for Single-Key Control Devices

[0049] Physical operation actions Session state Control command signals Click Idle state Voice recognition wake-up command Click Awaiting confirmation Confirm Result Command double click Listening status, speech recognition status, or AI processing status Switch mode command Long press Listening status, speech recognition status, or AI processing status Text editing commands (e.g., deleting the previous sentence of speech recognition output text). Long press Awaiting confirmation Cancel session command

[0050] When the operation module 120 of the control device consists of multiple physical buttons or capacitive touch areas, for example, the operation module 120 includes a main button and secondary buttons, the detectable physical operation actions include short press of the main button, long press of the main button, single click of the secondary button, double click of the secondary button, etc., and the corresponding session state-operation action-command mapping table can be Table 2:

[0051] Table 2 Session State-Operation Action-Command Mapping Table for Dual-Key Control Devices

[0052] Physical operation actions Session state Control command signals short press of main key Idle state Voice recognition wake-up command Secondary key click Awaiting confirmation Confirm Result Command Double-clicking secondary keys Listening status, speech recognition status, or AI processing status Switch mode command Long press the main button Listening status, speech recognition status, or AI processing status Cancel session command

[0053] This application maps physical operation actions (short press or click of the main key) to explicit and reliable voice recognition wake-up commands based on the session state (idle state), solving the problem that AI cannot identify the interaction object in multi-person dialogues; and maps physical operation actions (click or click of the secondary key) to confirmation commands based on the session state (waiting for confirmation state). Physical confirmation is more reliable and efficient than voice confirmation, especially after the user has visually confirmed the voice recognition result.

[0054] Step S34: The host system receives the control command signal and, in response to the control command signal, controls the artificial intelligence processing module to perform the corresponding interactive operation and updates the session state of the artificial intelligence processing module.

[0055] The command parsing module sends control command signals to the host system. The host system responds to the command, performs operations (such as activating voice recognition, confirming the previous result), and updates its own session state, thus forming a closed-loop control.

[0056] A common interaction characteristic in multi-turn conversations with artificial intelligence is the dynamic nature of user intent. Users often need to alternate between two different types of tasks within the same conversation: one is "providing content" (e.g., dictating a text, recording thoughts, or describing a scene), and the other is "issuing instructions" (e.g., asking the AI ​​to perform a task, answer a question, or perform a calculation). This flexible switching of intent is an inherent characteristic of natural human conversation. For example, a user might be dictating a report and suddenly realize they need to check some data, so they might naturally say: "...the market share for this quarter is expected to reach—[By the way, can you calculate the growth rate for the same period last year?]—Okay, back to what we were just saying, the market share for this quarter is expected to reach..." Under the current technological framework, the user must first say an instruction like "stop dictation" to terminate the current dictation session; then reawaken the system and issue the command "calculate the growth rate for the same period last year"; after obtaining the result, they must start a new dictation session and try to regain their previous train of thought, saying "the market share for this quarter is expected to reach...". In some simplified systems, users might try to avoid explicit mode switching and instead speak the instructions aloud during dictation, hoping the AI ​​can intelligently recognize different semantic interpretation patterns. This approach relies heavily on the AI's intent disambiguation capabilities, and the results are often unreliable. The AI ​​may misinterpret instructions within parentheses as text to be dictated, rather than the instructions it needs to execute, thus producing errors.

[0057] To enable the interactive system to dynamically and seamlessly switch semantic interpretation modes within the same session stream based on implicit or explicit user signals, thereby achieving truly natural, pleasant, and intelligent human-computer interaction, in some embodiments, the command parsing module 300 is further configured to send a structured message to the host system 200. The message includes a control command signal (CommandID), a session identifier (Session ID) identifying the current session, and a mode flag (ModeFlag) indicating the semantic interpretation mode. The session management module 210 of the host system 200 is configured to maintain the current session without interruption based on the session identifier and, based on the mode flag, determine the semantic interpretation mode for the speech recognition results within the same session. The semantic interpretation mode includes:

[0058] Text input mode: Inputs the speech recognition results as text information into the target application;

[0059] Command control mode: The speech recognition results are parsed into instructions, which are then sent to the AI ​​processing module to execute the corresponding operations.

[0060] When the control command signal is a mode switching command, the control device is configured to update the mode flag and carry the updated mode flag and timestamp in the message sent to the host system; the host system is configured to process the speech recognition results after the timestamp in the same session according to the semantic interpretation mode indicated by the updated mode flag.

[0061] Optionally, the host system is configured to align the timestamp with the speech stream or recognition result stream processed by the speech recognition module, which contains timing information, to pinpoint the effective point of the mode switch.

[0062] Figure 4 This is a schematic diagram illustrating the mode switching process within a user's voice conversation with artificial intelligence, according to an embodiment of this application. Figure 4 As shown,

[0063] Step S41: The host system is in the process of having a session S1 (Session ID = S1), and the current mode is text input mode (Mode Flag = Text).

[0064] Step S42: While the user is speaking continuously, a physical operation to switch modes is triggered (such as double-clicking the ring's secondary button).

[0065] Step S43: The command parsing module maps the physical operation to a mode switching command (Switch_Mode) based on the current session state and updates the internally maintained mode flag to (Mode Flag = Command). Then, it immediately generates a structured message containing Session ID = S1, Command ID = Switch_Mode, Mode Flag = Command, and the timestamp T1 of the operation, and sends it to the host system.

[0066] Step S44: After receiving the message, the host system does not interrupt the operation of the speech recognition module. It locates the message in the audio stream being processed or the recognition result stream based on the timestamp T1.

[0067] Step S45: Host System Record: Starting from time point T1, the mode of session S1 changes to command control mode. Therefore, speech recognition results belonging to the same session S1 after T1 will be dynamically routed to the AI ​​processing module for instruction parsing and execution, rather than being used as text input.

[0068] In this way, during a user's continuous speech, the first half is recorded as text, while the second half is executed as a command, achieving a truly seamless switch of intent.

[0069] The interactive system of this application realizes smooth intra-session intent switching. Through the mechanism of "session identifier + pattern flag + timestamp", users can seamlessly switch between text and command modes in a single, continuous voice conversation, adapting to the natural needs of human-computer hybrid intent dialogue.

[0070] In some embodiments, the interactive system also supports seamless intra-session intent switching for multimodal input. This embodiment... Figure 4 Building upon this foundation, its application scenarios and input compatibility are further expanded. The core lies in the fact that the host system is enhanced into a multimodal interaction platform, not only handling voice and control commands from the ring, but also seamlessly accepting and processing text, images, and other data from other input devices such as keyboards, touchscreens, and cameras within the same AI interaction session, while maintaining unified intent (text mode / command mode) management.

[0071] Specifically, a multimodal input manager is added to the host system. It serves as the unified entry point and routing center for all input sources, responsible for receiving input data from different channels (such as voice streams and command messages from control devices, text from the keyboard, images from the camera, etc.), and assigning a unified Session ID and Timestamp to all input data. Based on the current Mode Flag and the type of input data, it routes it to the correct processing module.

[0072] For example, a user is writing a weekly report that includes market data.

[0073] First, the user presses the ring briefly to wake up the system and simultaneously generates Session: S1, Initial Mode Flag: Text.

[0074] The user inputs voice, which is recognized as text and flows into the weekly report document.

[0075] The user double-clicks the ring during a pause in conversation. (Mode Flag switches to Command, and the message carries Timestamp:T1).

[0076] User (continuing to speak): "...Please query the best-selling single item in the A series and its specific data." The host system routes subsequent voice commands to the AI ​​command parser based on T1. The AI ​​executes the query and generates the result: "The best-selling single item in the A series is A1, with 5,000 units sold this week, accounting for 40% of total sales."

[0077] The user double-clicks the ring again. (Mode Flag switches back to Text, and the message carries Timestamp: T2). From T2 onwards, voice is again treated as text input.

[0078] When a user wants to directly input text via the keyboard, no switching operation is required; they can simply type "(see attached chart for detailed data)". The keyboard input event carries Session: S1 and the current timestamp T3. The multimodal input manager, based on the current Flag Mode still being Text, directly inserts the text into the document at the cursor position (i.e., time T3).

[0079] All input modalities (voice, physical commands, keyboard text, image files) are associated with a Session ID and aligned on the same timeline using Timestamps. This allows the system to reconstruct the entire session's cross-modal input stream.

[0080] This application constructs a unified multimodal interaction framework centered on conversation and using intent patterns as the routing strategy. Within a continuous conversation S1, users can freely switch between "content creation" (Text mode) and "command execution" (Command mode) roles via a unified physical or software switch. Information input via voice, keyboard, or image is intelligently interpreted and processed according to the current mode, ultimately merging into a coherent and rich output (such as a report, an email, or a proposal). This truly realizes the seamless integration of AI as a collaborator, adapting to the natural needs of future human-machine hybrid intelligent work.

[0081] In some embodiments, the control device 100 further includes a status feedback module 150. The core function of this module is to transmit the abstract session state of the host system 200 to the user in real time through physically perceptible signals (such as light and vibration), thereby constructing a complete perception loop from the digital world (AI state) to the physical world (user perception). This greatly reduces the user's reliance on visual feedback from the host screen, enabling "blind operation" and significantly improving the reliability and immersion of the interaction.

[0082] The status feedback module 150 may include one or more of the following physical feedback units:

[0083] Visual feedback unit: Typically consists of one or more multi-color LED indicators, visible through a light-transmitting hole on the device casing. Its driving circuit is controlled by the control unit 110, and can display different colors (such as red, green, and blue), brightness, and flashing modes (constant light, slow flashing, and fast flashing) according to different session states.

[0084] Tactile feedback unit: This is typically a miniature vibration motor (such as a linear resonant actuator, LRA). Its drive circuit is controlled by control unit 110, which can generate vibration patterns of different intensities, rhythms, and durations.

[0085] The control unit 110 is configured to drive the state feedback module 150 to generate predefined physical feedback signals that highly correspond to the meaning of the state, based on the current session state it determines or receives. For example, in the idle state, the LED is off; in the listening state, the LED steadily lights up blue with a slight short vibration, clearly informing the user that the microphone is on and the system is receiving voice. In the recognition / processing state, the LED softly flashes green, indicating to the user that the voice has been captured and the AI ​​is thinking or processing, please wait. This avoids the user having to repeat operations due to a lack of feedback.

[0086] like Figure 2As shown, this application also provides a control device coupled with an artificial intelligence session, including a control unit 110, an operation module 120, a communication module 130, a microphone 140, a status feedback module 150, and a command parsing module 300. The communication module 130 is configured to establish a communication connection with a host system containing an artificial intelligence processing module; the operation module 120 is configured to detect the user's physical operation actions. The control unit 110 is configured to send the user's physical operation actions to the command parsing module 300. The command parsing module 300 is configured to obtain the session state of the artificial intelligence processing module in the host system through the communication module 130 or infer the session state based on the interaction history; based on the session state, it maps the user's physical operation actions detected by the operation module to corresponding control command signals, wherein the same physical operation action is mapped to different control command signals in different session states; and it sends the control command signals to the host system through the communication module to drive the artificial intelligence processing module to execute corresponding AI interactive operations and update the session state.

[0087] Optionally, the command parsing module 300 is also configured to generate a structured message containing control command signals, a session identifier, and a pattern flag, and send it through the communication module; the session identifier is used to associate the current AI interaction session, and the pattern flag is used to indicate the semantic interpretation pattern of the host system for the speech recognition results.

[0088] The control unit 110 is further configured to drive the state feedback module 150 to generate a physical feedback signal based on the session state.

[0089] This application also provides a computer program product containing instructions. The computer program product may be software or program products containing instructions, capable of running on a network device or stored on any available medium. When the computer program product runs on at least one network device, it causes the at least one network device to perform an interaction method coupled to the artificial intelligence session state.

[0090] This application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that a network device can store, or a data storage device such as a data center that includes one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct the network device to perform an interaction method coupled to the state of an artificial intelligence session.

[0091] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously. Furthermore, other operations may be added to these processes, or one or more steps may be removed from these processes.

[0092] 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 protection scope of the technical solutions of the embodiments of the present invention.

[0093] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0094] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. An interactive system coupled with artificial intelligence conversational state, characterized in that, include: The control device is communicatively connected to a host system with a built-in artificial intelligence processing module. The control device includes: an operation module for detecting the user's physical operation actions; and a control unit for sending the physical operation actions to a command parsing module. A command parsing module, set in the control device or host system, is used to determine the session state of the artificial intelligence processing module in the host system and map the physical operation to the corresponding control command signal according to the session state. The same physical operation is mapped to different control command signals in different session states. The host system is configured to receive the control command signal, and in response to the control command signal, control the artificial intelligence processing module to perform corresponding interactive operations, and update the session state of the artificial intelligence processing module; The command parsing module is further configured to generate a structured message and send the structured message to the host system. The structured message includes at least: the control command signal, a session identifier for identifying the current session, and a pattern flag for indicating the semantic interpretation mode. The host system includes: a session management module, used to maintain the current session without interruption based on the session identifier, and to maintain or update the session state of the artificial intelligence processing module; a speech recognition module, used to recognize input speech and obtain speech recognition results; and an artificial intelligence processing module, used to execute corresponding interactive operations based on the control command signal, and to determine the semantic interpretation mode of the speech recognition result based on the mode flag, wherein the semantic interpretation mode includes: a text input mode, in which the speech recognition result is input as text information into the target application; and a command control mode, in which the speech recognition result is parsed into an instruction and the instruction is sent to the artificial intelligence processing module to execute the corresponding operation.

2. The system of claim 1, wherein, When the control command signal is a mode switching command, the command parsing module is configured to update the mode flag and carry the updated mode flag and timestamp in the message sent to the host system; The artificial intelligence processing module is configured to process, based on the timestamp, the speech recognition results after the timestamp in the same session according to the semantic interpretation mode indicated by the updated mode flag.

3. The system of claim 2, wherein, The artificial intelligence processing module is configured to align the timestamp with the speech stream or recognition result stream with time sequence information processed by the speech recognition module in order to locate the effective point of mode switching.

4. The system of claim 1, wherein, The command parsing module receives session state information from the artificial intelligence processing module of the host system and determines the session state based on the session state information.

5. The system of claim 1, wherein, The command parsing module infers the session state based on its internally recorded interaction history and / or operation timing context.

6. The system of claim 1, wherein, Mapping the physical operation to corresponding control command signals based on the session state includes: Based on the physical operation action and the session state, query the predefined session state-operation action-command mapping table; Retrieve control command signals corresponding to the current session state and physical operation actions from the mapping table.

7. The system of claim 6, wherein, The session state-action-command mapping table contains at least the following mapping relationships: When the session state is idle and the physical operation is a click, it is mapped to an AI voice recognition wake-up command; When the session status is pending confirmation and the physical action is a click, it is mapped to a confirmation result command; When the session state is in voice recognition or artificial intelligence processing state and the physical operation is a double-click, it is mapped to a mode switching command.

8. The system of claim 1, wherein, The control device further includes a status feedback module; and the control device is further configured to drive the status feedback module to generate a corresponding physical feedback signal according to the session status.

9. An interaction method coupled with an artificial intelligence conversation state, characterized in that, include: The control equipment detects the user's physical actions. The command parsing module determines the session state of the artificial intelligence processing module in the host system, and maps the physical operation to the corresponding control command signal according to the session state. The same physical operation is mapped to different control command signals in different session states. The host system receives the control command signal and, in response to the control command signal, controls the artificial intelligence processing module to perform corresponding interactive operations and updates the session state of the artificial intelligence processing module. The command parsing module sends a structured message to the host system, including the control command signal, a session identifier identifying the current session, and a pattern flag indicating the semantic interpretation mode. The host system's session management module maintains the current session without interruption based on the session identifier and maintains or updates the session state of the artificial intelligence processing module. The host system's speech recognition module recognizes the input speech and obtains the speech recognition result. The host system's artificial intelligence processing module executes the corresponding interactive operation based on the control command signal and determines the semantic interpretation mode of the speech recognition result based on the pattern flag. The semantic interpretation mode includes: text input mode: inputting the speech recognition result as text information into the target application; command control mode: parsing the speech recognition result into an instruction and sending the instruction to the artificial intelligence processing module to execute the corresponding operation.

10. The method of claim 9, wherein, When the control command signal is a mode switching command, the control device updates the mode flag and carries the updated mode flag and timestamp in the message; The host system processes the speech recognition results after the timestamp in the same session according to the semantic interpretation mode indicated by the updated mode flag, based on the timestamp.

11. A control device coupled with an artificial intelligence conversation state, characterized in that, include: The operation module is used to detect the user's physical operation actions; The communication module is used to establish a communication connection with the host system that has a built-in artificial intelligence processing module; The control unit is used to send physical operation actions to the command parsing module; The command parsing module is used to determine the session state of the artificial intelligence processing module in the host system, map the physical operation action into a corresponding control command signal according to the session state, and send the control command signal to the host system through the communication module to drive the host system to respond to the control command signal to control the artificial intelligence processing module to perform the corresponding interactive operation and update the session state of the artificial intelligence processing module. The same physical operation is mapped to different control command signals in different session states; The command parsing module is further configured to: generate a structured message, the structured message including at least: the control command signal, a session identifier for identifying the current session, and a mode flag for indicating the semantic interpretation mode; and send the structured message to the host system through the communication module. The host system maintains the current session without interruption based on the session identifier and instructs the artificial intelligence processing module on the semantic interpretation mode of the speech recognition result based on the mode flag. The semantic interpretation mode includes: text input mode: inputting the speech recognition result as text information into the target application; command control mode: parsing the speech recognition result into an instruction and sending the instruction to the artificial intelligence processing module to execute the corresponding operation.

12. The control device according to claim 11, characterized in that, It also includes a status feedback module; the control unit is further configured to drive the status feedback module to generate a physical feedback signal according to the session status.

13. The control device according to claim 11, characterized by The command parsing module is also used to infer the session state based on the interaction history and / or operation timing context recorded internally.

14. The control device according to claim 13, characterized by The command parsing module internally maintains a state inference engine, which relies on an internal state machine, interaction history logs, and predefined timeout thresholds to operate.

15. The control device according to claim 11, characterized by The control device is a wearable device, and the wearable device may be in the form of a ring, a wristband, or a pendant.

Images