Intelligent mouse, adjustment method and system

Abstract

The application relates to the technical field of intelligent input devices, in particular to an intelligent mouse, an adjusting method and a system, which comprise a mouse body; a key assembly arranged on the mouse body, the key assembly comprising at least one function key for generating a key event signal in response to a key operation of a user; a voice collection assembly arranged on the mouse body and used for collecting a voice signal of the user; and a communication assembly arranged on the mouse body and used for establishing a communication connection with a terminal device; by establishing a key mapping relationship between different key operation types and different artificial intelligence service calling types, different operations such as single clicking, double clicking, long pressing, continuous clicking, multi-key combination and composite operation can correspond to different artificial intelligence services, and the corresponding relationship can be visually configured and updated through a desktop artificial intelligence client on the terminal device.

Description

Technical Field

[0001] This invention relates to the field of intelligent input device technology, specifically to an intelligent mouse, adjustment method, and system. Background Technology

[0002] With the development of computer technology, wireless communication technology, and artificial intelligence technology, the mouse has gradually evolved from a traditional cursor positioning and button input device into an intelligent input device integrating functions such as shortcut operation, voice input, and device control. Current smart mice typically have side buttons, top buttons, and other function buttons, allowing users to perform preset operations such as launching applications, adjusting volume, and switching pages. Some smart mice also integrate microphones and other voice acquisition devices to achieve voice input functionality. However, the button functions of current smart mice are generally relatively simple, and their triggering functions are mainly limited to local applications.

[0003] Meanwhile, artificial intelligence technologies such as large language models, multimodal models, speech models, and intelligent agents have developed rapidly. AI services can now be applied to various scenarios, including voice question answering, text generation, screen content analysis, document processing, translation, information retrieval, office assistance, and task execution. However, existing AI services are typically delivered to users through standalone web pages, desktop applications, or other software clients. When users need to invoke AI services during routine mouse operations, they usually need to switch to the corresponding application or service interface and then initiate AI processing requests by clicking, typing, or selecting function menus. This involves multiple steps and can easily interrupt the user's current workflow. Existing methods suffer from the problems of requiring switching desktop clients, multiple clicks, and difficulty in uniformly scheduling different models.

[0004] Furthermore, different AI services may vary in terms of model capabilities, network transmission latency, service node load, and availability. For example, different AI services may be suitable for tasks such as natural language generation, speech recognition, image understanding, document processing, code processing, or tool invocation. If mouse functionality is directly bound to a fixed AI model or service interface, updates to the AI ​​model, changes to the service interface, or changes in network conditions may require corresponding adjustments to the mouse's configuration, hindering flexible access and expansion of different AI services. Therefore, how to rationally select from multiple candidate AI service nodes based on information such as the type of AI service invocation and the service node status is also a problem that needs to be solved in the integration of smart input devices with AI services.

[0005] On the other hand, existing mouse function buttons typically use a one-button-one-fixed-function approach, limiting the number of functions that a limited number of physical buttons can support. When a mouse needs to invoke various AI services such as voice Q&A, text generation, screenshot analysis, translation, document processing, and intelligent agent tasks, expanding functionality by adding more physical buttons not only increases the complexity of the mouse's structure and circuit design but may also negatively impact the user experience. Therefore, it is necessary to further increase the number of functions that a limited number of physical buttons can support and enable users to adjust the mapping between different button operations and AI services according to their actual needs. Customizable button mapping can recognize different operations such as single clicks, double clicks, long presses, continuous clicks, and multi-key combinations, and supports configuring corresponding AI service invocation relationships through a desktop client.

[0006] Furthermore, during the use of a wireless smart mouse, situations may arise such as the terminal device not being ready temporarily, wireless communication malfunctions, or network connection interruptions. If AI service call requests and corresponding voice data are only forwarded instantly, request or data loss may occur if the terminal device cannot respond in time, affecting the continuity of AI service calls. Existing solutions also need to consider the mouse's battery life, reducing the power consumption of wireless communication modules when the mouse is idle or inactive for extended periods, and ensuring timely resumption of normal operation when the user resumes mouse use. This paper addresses this by incorporating receiver buffering and a tiered low-power control mechanism.

[0007] Therefore, how to combine the physical button operation of the mouse, voice input, and contextual data such as screen content, selected text, clipboard data, and current document in the terminal device with the call of artificial intelligence services, and realize flexible configuration of button functions, selection of multiple artificial intelligence service nodes, service status feedback, and reliable handling in abnormal situations, thereby reducing the operation threshold of artificial intelligence services and improving the interaction efficiency and adaptability of smart mice, has become a technical problem that needs to be further solved in this field. Summary of the Invention

[0008] To address the problems in the prior art, this invention provides an intelligent mouse, adjustment method, and system.

[0009] The technical solution adopted by the present invention to solve its technical problem is: an intelligent mouse, including a mouse body; A button assembly is disposed on the mouse body, and the button assembly includes at least one function button for generating a button event signal in response to the user's button operation. A voice acquisition component, located on the mouse body, is used to acquire the user's voice signal; A communication component, located on the mouse body, is used to establish a communication connection with a terminal device; A memory is used to store key mapping information, which is used to represent the correspondence between different key operation types and different artificial intelligence service call types; The processor is connected to the button assembly, the voice acquisition assembly, the communication assembly, and the memory, respectively. The processor is configured as follows: Obtain the key event signal; The corresponding key operation type is determined based on the time sequence characteristics of the key event signal; The target AI service call type is determined based on the key operation type and the key mapping information; Generate a service call request corresponding to the target artificial intelligence service call type, wherein the service call request includes the voice signal and / or a data acquisition instruction for instructing the terminal device to acquire input data; The communication component sends the service call request to the terminal device, so that the terminal device calls the target artificial intelligence service according to the service call request and outputs the processing result returned by the target artificial intelligence service.

[0010] A method for invoking artificial intelligence services based on a smart mouse, applied to a terminal device communicating with a smart mouse, the method comprising: The system receives a service call request sent by the smart mouse. The service call request is determined by the smart mouse based on the key event signals generated by the function keys. The service call request includes an artificial intelligence service call type and voice data and / or data acquisition instructions. Parse the service call request to obtain the AI ​​service call type; If the service call request includes the data acquisition instruction, the data to be processed is acquired from the terminal device according to the data acquisition instruction. Generate a model processing request based on the AI ​​service call type and the data to be processed and / or the voice data; The target AI service node is determined from multiple candidate AI service nodes; Send the model processing request to the target artificial intelligence service node; Receive the processing result returned by the target artificial intelligence service node; The processing result is output through the terminal device.

[0011] An artificial intelligence service invocation system includes a smart mouse, a terminal device, and multiple artificial intelligence service nodes; The smart mouse is communicatively connected to the terminal device, and the terminal device is communicatively connected to the plurality of artificial intelligence service nodes; The smart mouse is used for: In response to key event signals generated by function keys, determine the corresponding key operation type; The type of AI service call is determined based on the type of key operation. And send a service call request, including the type of the artificial intelligence service call, to the terminal device; The terminal device is used for: Obtain the data to be processed based on the service call request; The model processing request is generated based on the AI ​​service call type and the data to be processed. The target artificial intelligence service node is determined from the plurality of artificial intelligence service nodes based on at least one of the artificial intelligence service call type, network status, and running status of the plurality of artificial intelligence service nodes; Send the model processing request to the target artificial intelligence service node; Receive the processing result returned by the target artificial intelligence service node; And output the processing result; The target AI service node is used to perform AI inference processing according to the model processing request and return the processing result to the terminal device.

[0012] The beneficial effects of this invention are: (1) The intelligent mouse, adjustment method and system described in this invention, by setting function buttons, voice acquisition components, communication components and processor on the mouse body, identifies the corresponding key operation type according to the time sequence characteristics of the key event signals generated by the function buttons, and further determines the corresponding artificial intelligence service call type, so that users can directly trigger artificial intelligence services such as voice question answering, text generation, screen content analysis, document processing, translation, information retrieval, office assistance and intelligent agent task execution through the physical buttons of the mouse, reducing the process of switching between different applications or artificial intelligence service interfaces and multiple click operations, and improving the convenience of artificial intelligence service calls and human-computer interaction efficiency.

[0013] (2) The intelligent mouse, adjustment method, and system described in this invention establish a key mapping relationship between different key operation types and different artificial intelligence service call types, enabling different operations such as single click, double click, long press, continuous click, multi-key combination, and compound operation to correspond to different artificial intelligence services. The mapping relationship can be visualized and updated through a desktop artificial intelligence client on the terminal device, thereby expanding the number of functions that the mouse can directly call without increasing the number of physical buttons, and allowing for personalized configuration based on different users' operating habits and usage scenarios. It explicitly supports configuring the mapping relationship on the desktop client and sending the updated mapping configuration to the mouse.

[0014] (3) The intelligent mouse, adjustment method and system described in this invention determine the target artificial intelligence service node from multiple candidate artificial intelligence service nodes through terminal devices or artificial intelligence service routing. The service node selection can be combined with information such as artificial intelligence service call type, model capability, network status, node load and user preferences, so that different types of artificial intelligence processing tasks can be sent to the appropriate artificial intelligence service node. When an artificial intelligence service node is added, deactivated or the model version changes, the service can be extended by updating the node information maintained by the service routing, without changing the hardware structure and button recognition logic of the intelligent mouse, thereby improving the system's compatibility and scalability for different artificial intelligence models, service interfaces and deployment locations. Attached Figure Description

[0015] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0016] Figure 1 A schematic diagram of the overall structure of a smart mouse provided by the present invention; Figure 2 The overall data flow architecture diagram of the system provided by this invention; Figure 3 This is a flowchart of the DPI level cyclic adjustment provided by the present invention; Figure 4 The adaptive calibration flowchart provided by this invention; Figure 5 The flowchart for intelligent power consumption adaptive adjustment provided by the present invention.

[0017] In the image: 1. Smart mouse; 2. Mouse body; 21. Upper shell; 22. Lower shell; 23. Scroll wheel assembly; 3. Status indicator assembly; 4. Button assembly; 41. Side button; 42. Top button; 5. Voice acquisition assembly. Detailed Implementation

[0018] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the following detailed description, in conjunction with the accompanying drawings and specific embodiments, provides a more comprehensive understanding of the smart mouse, the method for invoking artificial intelligence services based on the smart mouse, and the system for invoking artificial intelligence services provided in this application. It should be understood that the following embodiments are merely illustrative of this application and are not intended to limit the scope of protection of this application.

[0020] In the description of this application, unless otherwise expressly specified and limited, the terms "connection," "communication connection," "electrical connection," etc., should be interpreted broadly. For example, "connection" can be a direct connection or an indirect connection through intermediate components; it can be a wired connection or a wireless connection; it can be a connection used to transmit control signals or a connection used to transmit data, audio signals, or status information. Those skilled in the art can understand the specific meaning of the above terms in this application according to specific application scenarios.

[0021] In this application, "processor" can be a microcontroller, central processing unit, digital signal processor, application-specific integrated circuit, field-programmable gate array, system-on-a-chip, or other processing device capable of executing preset programs and control logic. A processor may include a single processing core or multiple processing cores that work together.

[0022] In this application, "memory" may include flash memory, read-only memory, and random access memory within the processor, or it may include a separate memory chip connected to the processor. The memory is used to store program instructions, key mapping information, device configuration parameters, communication parameters, and data to be sent or processed.

[0023] The "artificial intelligence services" in this application include, but are not limited to, services such as voice question answering, text generation, text rewriting, text summarization, screen content analysis, image understanding, document processing, translation, information retrieval, office assistance, code processing, and tool invocation provided by large language models, multimodal models, speech models, image understanding models, retrieval enhancement generation systems, or intelligent agent systems.

[0024] In this application, "artificial intelligence service node" refers to a computing node capable of receiving model processing requests and performing artificial intelligence inference processing. Artificial intelligence service nodes can be deployed locally on terminal devices, on local area network servers, edge servers, cloud servers, or in computing centers distributed across different geographical regions. Different artificial intelligence service nodes can run the same artificial intelligence model, or they can run artificial intelligence models with different capabilities, scales, input modalities, or areas of expertise.

[0025] The "domestic large-scale model service node" in this application can be considered a specific implementation of an artificial intelligence service node. For example, the domestic large-scale model service node can provide services based on models such as Qianwen, DeepSeek, Wenxin Yiyan, or Doubao, but the scope of protection of this application is not limited by the specific model name, model version, service provider, or deployment location.

[0026] Example 1: Overall Structure of a Smart Mouse This embodiment provides a smart mouse 1. The smart mouse 1 includes a mouse body 2, a status indicator component 3, a button component 4, a voice acquisition component 5, a processor, a communication component, and a memory.

[0027] The mouse body 2 may include an upper shell 21, a lower shell 22, and an internal receiving space disposed between the upper shell 21 and the lower shell 22. The upper shell 21 and the lower shell 22 may be connected to each other by means of clips, screws, adhesive structures, or a combination of the aforementioned connection methods.

[0028] The mouse body 2 may also include conventional mouse input components. These components may include a left button, a right button, a scroll wheel assembly 23, and a displacement detection assembly. The displacement detection assembly may include an optical sensor, a laser sensor, or other sensors capable of detecting the displacement of the mouse body 2 relative to a supporting surface.

[0029] The displacement detection component, left button, right button, and scroll wheel component 23 are each connected to the processor. The processor can generate cursor movement information based on the displacement data output by the displacement detection component, and generate regular mouse control information based on the operation signals generated by the left button, right button, and scroll wheel component 23.

[0030] Button assembly 4 includes at least one function button. The function button is independent of the left and right buttons and is used to trigger artificial intelligence services. The function button can be located on the side, top, front, or other convenient location for the user's fingers to operate on the mouse body 2.

[0031] In one specific implementation, the button assembly 4 includes at least one side button 41 and at least one top button 42. The side button 41 is located on the side of the mouse body 2 and can be positioned at the location corresponding to the user's thumb when holding the mouse. The top button 42 is located on the top of the mouse body 2 and can be located behind the scroll wheel assembly 23 or in the area between the left and right buttons.

[0032] Side button 41 and top button 42 generate button event signals via microswitches, membrane switches, touch sensors, or pressure sensors, respectively. When using microswitches or membrane switches, side button 41 and top button 42 can be mounted on a button circuit board, which is connected to the processor via a flexible circuit board, ribbon cable, or wire.

[0033] Side button 41 and top button 42 can be connected to different general-purpose input / output ports of the processor. This allows the processor to identify the specific function key that generated the key event signal. The processor can also identify the pressed and released states of different function keys through a key scanning circuit.

[0034] The voice acquisition component 5 can be located on the side of the mouse body 2. The mouse body 2 can be provided with a pickup hole corresponding to the voice acquisition component 5. The pickup hole is used to transmit external sound to the voice acquisition component 5, and at the same time, a dustproof mesh or acoustic membrane can be used to reduce dust from entering the interior of the mouse body 2.

[0035] The communication component is located within the mouse body 2 and connected to the processor. The communication component is used to establish a communication connection between the smart mouse 1 and the terminal device. The terminal device can be a desktop computer, laptop computer, tablet computer, smart display device, or other electronic device capable of running desktop artificial intelligence clients.

[0036] The memory can be integrated into the processor or placed as a separate device within the mouse body 2. The memory stores key mapping information. This key mapping information represents the correspondence between function keys, key operation types, and artificial intelligence service call types.

[0037] For example, key mapping information may include a key identifier field, an operation type field, and a service call type field. The key identifier field is used to identify side key 41, top key 42, or other function keys; the operation type field is used to identify single click, double click, long press, continuous click, multi-key combination, or compound operation; and the service call type field is used to identify artificial intelligence services such as voice question answering, text generation, screen content analysis, document summarization, translation, or intelligent agent task execution.

[0038] It should be noted that the above fields are merely an example to illustrate key mapping information. Key mapping information can also take the form of lookup tables, key-value pairs, configuration files, state machine parameters, or other data formats that can represent the correspondence between key operations and AI service call types.

[0039] Example 2: Key Event Recognition and Service Call Request Generation The processor can acquire key event signals generated by side key 41 and / or top key 42 through interrupt detection or periodic scanning, and record the press time, release time and key identifier corresponding to each key event.

[0040] The processor determines the key operation type based on the time sequence characteristics of the key event signals. The time sequence characteristics may include at least one of the following: key press duration, key release duration, time interval between adjacent key events, number of key events within a preset time period, trigger order of different function keys, and whether the trigger times of different function keys overlap.

[0041] In one example, when the processor detects that a function key has been pressed and released within a first duration, and the function key is not pressed again within the subsequent double-click wait time, the processor recognizes the key operation as a single click operation.

[0042] When the processor detects that the same function key has completed two valid press and release actions in succession within a preset double-click time window, it can recognize the key operation as a double-click operation.

[0043] When the duration of a function key press reaches or exceeds a preset long-press threshold, the processor can recognize the key press operation as a long press. The processor can generate a long press event immediately upon reaching the long press threshold, or it can confirm the long press event after the function key is released.

[0044] When the processor detects three, four, or more valid clicks on the same function key within a preset continuous click time window, it can recognize the key operation as a continuous click operation.

[0045] When the processor detects that the side button 41 and the top button 42 are pressed simultaneously or partially overlapping in time, it can identify the corresponding operation as a multi-key combination operation. For example, a simultaneous long press of the side button 41 and the top button 42 can correspond to an AI service node switching command or a backup node invocation command.

[0046] A compound operation can consist of at least two basic key operations in a preset order. For example, clicking the side button 41 first and then holding down the side button 41, or double-clicking the top button 42 first and then clicking the side button 41, can both be recognized as different compound operations.

[0047] For example, the effective press time for a single click can be less than or equal to 200 milliseconds; the interval between two consecutive clicks in a double click operation can be between 100 and 500 milliseconds; and the duration of a long press operation can be greater than or equal to 800 milliseconds. These values ​​are only for illustrating one feasible implementation, and specific thresholds can be adjusted according to button hardware characteristics, user habits, and accidental touch control requirements.

[0048] After identifying the key operation type, the processor determines the target artificial intelligence service call type based on the key identifier of the function key, the key operation type, and the key mapping information in the memory.

[0049] For example, the key mapping information can map the single-click operation of the side key 41 to voice question and answer, the double-click operation of the side key 41 to screen content analysis, the long-press operation of the side key 41 to real-time translation, the single-click operation of the top key 42 to text generation, the double-click operation of the top key 42 to document summarization, and the long-press operation of the top key 42 to agent task execution.

[0050] The above mapping is only the default configuration. Different users can change the mapping according to their work habits. For example, designers can configure double-clicking the side button 41 for image analysis, programmers can configure double-clicking the side button 41 for code interpretation, and copywriters can configure the single-click operation of the top button 42 for text rewriting.

[0051] After determining the target AI service call type, the processor generates a corresponding service call request. The service call request may include an AI service call type identifier, as well as voice data and / or data acquisition instructions.

[0052] When the target AI service call type is voice question and answer, voice translation, or voice command execution, the processor can activate the voice acquisition component 5, and after acquiring the user's voice, encapsulate the voice data into the service call request.

[0053] When the target AI service call type requires the use of screen content, selected text, clipboard content, current document, or current application context on the terminal device, the service call request may include the corresponding data retrieval instructions.

[0054] Data acquisition commands may include at least one of the following: screenshot command, selected text acquisition command, clipboard data acquisition command, current document acquisition command, and current application context acquisition command.

[0055] For example, when a user double-clicks side button 41 to trigger screen content analysis, the smart mouse 1 does not need to directly obtain the screen data of the terminal device. Instead, it sends a service call request to the terminal device, including a screenshot command. The desktop AI client responds to the screenshot command, obtains the current screen image, and uses the current screen image as data to be processed.

[0056] In one alternative implementation, the service invocation request can take the form of a data packet. The data packet may include a header, payload, and verification information. The header may include a device identifier, protocol version, request sequence number, key identifier, key operation type identifier, AI service invocation type identifier, and data type identifier.

[0057] The payload may include voice data, control parameters, or data acquisition instructions. Verification information may include checksums, cyclic redundancy check values, or message authentication information. Service call requests may also include at least one of the following: data length information, generation time information, packet sequence number, and retransmission identifier.

[0058] In practical implementation, custom binary protocols, Universal Serial Bus human-machine interface device protocols, Bluetooth Low Energy features, serial communication protocols, or other protocols suitable for transmitting request data between the smart mouse 1 and the terminal device can be used.

[0059] Example 3: Visual configuration and updating of key mapping information The desktop AI client provides a key mapping configuration interface. This interface displays the function keys of the smart mouse 1, the key operation types supported by each function key, and the available AI service call types.

[0060] Users can select the target function key and target key operation type in the configuration interface, and specify the corresponding AI service call type. The desktop AI client generates mapping configuration data based on the user's input configuration.

[0061] The mapping configuration data may include at least one of the following: target key identifier, target key operation type identifier, new service call type identifier, and configuration version information.

[0062] The desktop AI client sends the mapping configuration data to the smart mouse 1 via a communication connection between the terminal device and the smart mouse 1. After receiving the mapping configuration data, the processor can first perform integrity and version verification on the mapping configuration data.

[0063] After the mapping configuration data passes verification, the processor updates the key mapping information in memory. The update method can be to replace all key mapping information or to modify only the mapping record corresponding to the target key identifier and the target key operation type identifier.

[0064] To reduce the risk of key mapping information corruption due to power outages during the update process, a current mapping area and a backup mapping area can be set in the memory. The processor first writes the new key mapping information to the backup mapping area, and after verifying that the written content is correct, switches the backup mapping area to the current mapping area.

[0065] The aforementioned dual-storage area update mechanism is an optional implementation. The mapping between button operations and AI service call requests can be visually configured by the terminal device, and the configuration results can be sent to the smart mouse 1 for updates.

[0066] Example 4: Voice Acquisition and Noise Reduction Processing The voice acquisition component 5 includes a microphone, an analog front-end processing circuit, and a digital signal processing unit. The microphone can be a microelectromechanical system (MEMS) microphone, an electret microphone, or other sound pickup devices suitable for integration into the mouse body 2.

[0067] The microphone converts the captured sound into a raw speech signal and outputs the raw speech signal to the analog front-end processing circuit. The analog front-end processing circuit may include at least one of the following: an amplifier circuit, a low-pass filter circuit, a high-pass filter circuit, a band-pass filter circuit, and an analog-to-digital converter circuit.

[0068] The analog front-end processing circuit performs at least one of the following processes on the original speech signal: amplification, filtering, and analog-to-digital conversion, and outputs the processed digital speech signal to the digital signal processing unit.

[0069] The digital signal processing unit can perform at least one of the following processing on the digital speech signal: background noise suppression, fixed noise cancellation, automatic gain control, echo suppression, or speech activity detection.

[0070] Voice activity detection can be used to determine whether user voice is present in the currently acquired signal. When no valid voice is detected, the processor can choose not to send the corresponding audio segment, thereby reducing the amount of data transmitted wirelessly.

[0071] In one operating mode, voice acquisition is initiated when the user presses and holds the side button 41, and terminated when the user releases the side button 41. The processor encapsulates the voice data acquired during the long press into a service call request and sends it to the terminal device.

[0072] In another operating mode, the user initiates voice acquisition by clicking side button 41. The digital signal processing unit determines the start and end times of the voice recording through voice activity detection. If no valid voice is detected within a preset silence period, the processor terminates the voice acquisition.

[0073] Voice data can be encoded or compressed within the smart mouse 1 before being sent, or it can be sent to the terminal device in the form of pulse code modulation data or other audio data, and then encoded by the desktop artificial intelligence client.

[0074] Example 5: Status Indication and Status Feedback The smart mouse 1 also includes a status indicator component 3. The status indicator component 3 can be disposed on the surface of the upper housing 21, for example, near the scroll wheel assembly 23, at the tail of the mouse, or in a position that can be observed by the user during normal use.

[0075] The status indicator component 3 includes a multi-color light-emitting unit and a driving circuit. The multi-color light-emitting unit can be a red, green, or blue light-emitting diode, multiple light-emitting diodes of different colors, or a light strip that can change color and brightness.

[0076] The driving circuit is connected to both the multi-color light-emitting unit and the processor. The processor controls at least one of the following parameters of the multi-color light-emitting unit: its emission color, brightness, flashing frequency, and illumination duration, through the driving circuit.

[0077] When the smart mouse 1 starts collecting voice data, the processor can control the multi-color light-emitting unit to output a first status prompt; when a service call request has been sent to the terminal device, a second status prompt can be output; when the artificial intelligence service is performing inference processing, a third status prompt can be output; when the processing result is returned, a fourth status prompt can be output; and when a network error, service call failure, or terminal device not being ready occurs, an abnormal status prompt can be output.

[0078] For example, the voice acquisition status can be displayed as solid red, the artificial intelligence service processing status can be displayed as blue breathing, the normal communication connection status can be displayed as solid green, the abnormal status can be displayed as rapidly flashing red, and the low battery status can be displayed as flashing yellow.

[0079] The above correspondence between colors and states is not a necessary limitation. Actual products may use other colors, brightness, or flashing modes depending on user settings, product appearance design, and power consumption requirements.

[0080] The desktop AI client can send service status information to the smart mouse 1 at different processing stages. For example, the desktop AI client sends the first service status information after receiving a service call request, the second service status information after the model successfully processes the request, and the third service status information after receiving the processing result.

[0081] The processor controls the status indicator component 3 to output different status prompts based on different service status information, so that users can understand the current operating stage of the artificial intelligence service without having to constantly observe the display interface of the terminal device.

[0082] Example 6: Dual-mode communication and wireless receiver buffer The communication components may include a Bluetooth communication unit and a 2.4GHz wireless communication unit. A communication mode switching component may be located on the bottom of the mouse body 2. The communication mode switching component may be a toggle switch, a push switch, or a touch switch.

[0083] When the communication mode switching component is in the first position, the processor enables the Bluetooth communication unit and enables the smart mouse 1 to communicate directly with the terminal device via Bluetooth.

[0084] When the communication mode switching component is in the second position, the processor enables the 2.4GHz wireless communication unit and enables the smart mouse 1 to communicate with the terminal device through the wireless receiver.

[0085] The wireless receiver includes a receiver processing unit, a radio frequency communication unit, a buffer unit, and a terminal interface. The radio frequency communication unit is used to receive regular mouse control information, service call requests, voice data, and configuration confirmation information sent by the smart mouse 1.

[0086] The terminal interface can be a Universal Serial Bus (USB) interface. The terminal interface can be USB Type-A, USB Type-C, or other interface types capable of establishing a data connection with the terminal device.

[0087] The cache unit may include flash memory, static random access memory, or other storage devices suitable for temporarily storing data. The cache unit is used to temporarily store service call requests and / or voice data when the terminal device or desktop AI client fails to read the data in a timely manner.

[0088] The receiver processing unit can detect whether the terminal device is in a ready state through the terminal interface. A ready state can be a state where the terminal device has recognized the wireless receiver, the corresponding communication driver has been loaded, and the desktop AI client can read AI service-related data.

[0089] When the terminal device is in a ready state, the receiver processing unit can send the data received through the radio frequency communication unit to the terminal device.

[0090] When the terminal device is in a not-ready state, the receiver processing unit can write the service call request and related voice data into the cache unit and return the not-ready state information to the smart mouse 1.

[0091] The processor controls the status indication component 3 to output a not ready prompt based on the not ready status information, so as to inform the user that the current request has been received but has not yet been processed by the terminal device.

[0092] Once the terminal device returns to the ready state, the receiver processing unit reads the data from the buffer unit according to the data reception order and sends it to the terminal device through the terminal interface.

[0093] In one optional implementation, each service call request has a request sequence number, and the voice data packet has a request sequence number and a packet sequence number corresponding to the service call request. The receiver processing unit reconstructs the correspondence between the request and the voice data based on the request sequence number and the packet sequence number.

[0094] When the cache space is insufficient, the receiver processing unit can prioritize retaining control-type requests and return a cache shortage status information to the smart mouse 1. The specific mechanisms for cache replacement, overflow handling, and data retransmission can be determined based on the storage capacity and communication protocol used by the product.

[0095] A scheme that bypasses the buffer unit and communicates directly with the terminal device in a wired connection state. In one implementation, when the wireless receiver establishes a stable connection with the terminal device through the terminal interface and the terminal device is in a ready state, the receiver processing unit forwards the newly received data directly to the terminal device without first writing it to the buffer unit.

[0096] Example 7: Artificial Intelligence Service Invocation Method Based on Smart Mouse This embodiment provides a method for invoking artificial intelligence services based on a smart mouse. This method can be executed by a desktop AI client running on a terminal device, or it can be executed collaboratively by the desktop AI client and an AI service route set up on a local or cloud server.

[0097] The method may include the following steps.

[0098] Step S101: Receive the service call request sent by the smart mouse 1.

[0099] The service call request is generated by the smart mouse 1 based on key event signals generated by the function keys. The service call request includes the artificial intelligence service call type, as well as voice data and / or data acquisition instructions.

[0100] Terminal devices can receive service call requests directly via Bluetooth communication or via a wireless receiver.

[0101] Step S102: Parse and verify the service call request.

[0102] Desktop AI clients can parse device identifier, request sequence number, key identifier, key operation type identifier, AI service call type identifier, and data type identifier from service call requests.

[0103] The desktop AI client can perform integrity checks on service call requests based on data length and verification information. If a service call request fails the integrity check, it can discard the request, request the smart mouse 1 to resend it, or send an exception status message to the smart mouse 1.

[0104] Step S103: Obtain the data to be processed according to the data acquisition instruction.

[0105] When a service call request includes a screenshot command, the desktop AI client obtains the full screen image or the image of a specified window currently displayed on the terminal device.

[0106] When a service call request includes a command to retrieve selected text, the desktop AI client can retrieve the text selected by the user in the current application through the operating system interface.

[0107] When a service call request includes a clipboard data retrieval instruction, the desktop AI client reads the clipboard content of the terminal device.

[0108] When a service call request includes a command to retrieve the current document, the desktop AI client can retrieve the full content or a specified portion of the currently open document through the office application interface.

[0109] When a service call request includes a command to retrieve the current application context, the desktop AI client can obtain the type of the currently active application, the window title, the user-selected object, or other context data that can be provided by the operating system and application programming interfaces.

[0110] Before accessing data involving user information, the desktop AI client can determine whether it has the appropriate data access permissions based on preset permission settings. If the user does not have the permissions, the client can prompt the user to grant authorization or stop the current access.

[0111] Step S104: Generate a model processing request based on the AI ​​service call type, the data to be processed, and / or the voice data.

[0112] The model processing request may include at least one of the following: task type, input data, prompts, output format requirements, and user preference information.

[0113] For example, when the AI ​​service call type is screen content analysis, the model processing request can include the current screen image and task type information to instruct the model to analyze the screen content.

[0114] When the AI ​​service call type is translation, the model processing request can include the text to be translated, source language information, target language information, and translation style requirements.

[0115] When the AI ​​service call type is document summary, the model processing request can include the current document content, summary length requirements, and summary format requirements.

[0116] When the AI ​​service call type is agent task execution, the model processing request can include text instructions converted from user speech, the scope of tools allowed to be called, and task execution constraints.

[0117] Step S105: Determine the target artificial intelligence service node from multiple candidate artificial intelligence service nodes.

[0118] Candidate AI service nodes can include large model service nodes and intelligent agent service nodes. Different candidate AI service nodes can be distributed in different regions, or have different model capabilities, network latency, load status, and service availability.

[0119] Desktop AI clients or AI service routes can obtain at least one of the following: the geographical location information of the terminal device, the network status information between the terminal device and each candidate AI service node, the load information of each candidate AI service node, capability tags, availability status, and user preference information.

[0120] Geographic location information can be a country, region, or network access area. Network status information can include at least one of the following: connection availability, round-trip time, packet loss rate, and available bandwidth.

[0121] Load information may include at least one of the following: the current request queue length of the candidate AI service node, processor utilization, graphics processor utilization, number of concurrent requests, or expected waiting time.

[0122] Capability tags can be used to represent at least one capability dimension of a candidate AI service node in natural language generation, speech recognition, image understanding, document processing, code processing, translation, information retrieval, and tool invocation.

[0123] Desktop AI client or AI service routing determines the target capability dimension based on the AI ​​service call type. For example, screenshot analysis corresponds to the image understanding capability dimension, document summarization corresponds to the document processing or natural language generation capability dimension, and agent task execution corresponds to the tool call capability dimension.

[0124] In one exemplary implementation, candidate AI service nodes can be classified into different levels based on their capability adaptability, network quality, node load, and user preferences, and node priority can be determined according to each level.

[0125] For example, when multiple candidate AI service nodes support the target capability dimension, the candidate AI service node with lower network latency and lower load can be selected first; when the user pre-specifies the preferred model or preferred node, the user-specified node can be selected first if the corresponding node is available.

[0126] Node selection can be achieved using rule-based sorting, weighted scoring, priority lists, or a combination of the aforementioned methods.

[0127] In one optional weighted scoring implementation, weights can be assigned to capability suitability, network quality, and node load, and node scores can be calculated based on normalized capability parameters, network latency parameters, and load parameters. This calculation method is only an implementation example for node selection and does not constitute a necessary limitation of this application.

[0128] Step S106: Send a model processing request to the target artificial intelligence service node.

[0129] When the AI ​​service routing is set on the terminal device, the desktop AI client can send model processing requests to the target AI service node through the local AI service routing module.

[0130] When the AI ​​service routing is set on a cloud server, the desktop AI client first sends the model processing request to the AI ​​service routing, and then the AI ​​service routing forwards the model processing request to the target AI service node.

[0131] Step S107: Receive the processing result returned by the target artificial intelligence service node.

[0132] The processing results can be text, images, audio, structured data, operation instructions, or task execution results. For streaming output, the desktop AI client can display the results gradually as it receives partial output.

[0133] Step S108: Output the processing result through the terminal device.

[0134] The processing results can be displayed in the desktop AI client window, or written to the clipboard, inserted into the current document, overwritten with selected text, played via voice, or triggered by the corresponding application operation.

[0135] For example, translation results can directly replace selected text, document summary results can be inserted at the end of the current document, screen analysis results can be displayed in a floating window, and the task execution results of the office automation agent can include created files, organized data, or completed application operations.

[0136] Example 8: Abnormal Switching of Artificial Intelligence Service Nodes After sending a model processing request to the target AI service node, the desktop AI client or AI service router can detect whether the target AI service node meets the abnormal switching conditions.

[0137] Abnormal switching conditions may include at least one of the following: the request has not been responded to for more than a preset time, the connection between the target AI service node is interrupted, the load of the target AI service node exceeds a preset load threshold, and the target AI service node returns service unavailable or processing abnormal information.

[0138] When the target AI service node meets the abnormal switching conditions, the AI ​​service routing can mark the target AI service node as temporarily unavailable and determine a backup AI service node from the remaining candidate AI service nodes.

[0139] Backup AI service nodes can be determined based on the same node selection rules as the target AI service node, or based on a pre-set list of node priorities.

[0140] The AI ​​service routing sends model processing requests to backup AI service nodes. To prevent the same task from being executed repeatedly, model processing requests can carry a unique request identifier, which the AI ​​service routing can use to identify duplicate requests.

[0141] When the model processes requests involving tool invocation tasks with irreversible results, such as deleting files or sending data, AI service routing can confirm whether the target node has already performed the corresponding operation before switching to a backup node, or split the tool invocation into an inference phase and an execution confirmation phase to reduce the risk of repeated execution.

[0142] The aforementioned irreversible task control is an optional implementation applicable to agent task execution scenarios. For ordinary text generation, translation, or summarization tasks, the process can be switched directly to a backup node for reprocessing after a request failure.

[0143] Example 9: Local Offline Response to Network Anomalies When a terminal device cannot access multiple candidate AI service nodes, the desktop AI client can determine whether the task corresponding to the model processing request meets the local processing conditions.

[0144] Local processing conditions may include at least one of the following: the terminal device is equipped with a local artificial intelligence processing module capable of handling the task; the amount of data to be processed does not exceed the local processing capacity; the task does not require access to external real-time data; and the user allows local offline processing.

[0145] The local AI processing module can be a lightweight local model inference engine running on the terminal device, or a local question-answering module generated based on preset rules, local knowledge base, and retrieval enhancements.

[0146] When a model processing request meets the local processing conditions, the desktop AI client sends the request to the local AI processing module. The local AI processing module processes the request and generates the local processing result.

[0147] The desktop AI client outputs the local processing results and sends service status information to the smart mouse 1 to indicate the offline response status. The processor then controls the status indicator component 3 to output the corresponding offline status prompt based on this service status information.

[0148] When a model processing request does not meet local processing conditions, the desktop AI client can save the model processing request and resend it after the network is restored; it can also prompt the user that the current task cannot be completed offline.

[0149] Example 10: Artificial Intelligence Service Invocation System This embodiment provides an artificial intelligence service invocation system. The system includes a smart mouse 1, a terminal device, and multiple artificial intelligence service nodes.

[0150] The smart mouse 1 communicates with the terminal device, which in turn communicates with multiple AI service nodes. The AI ​​service routing can be configured on the terminal device or on a cloud server that communicates with the terminal device.

[0151] The smart mouse 1 is used to respond to key event signals generated by function keys, determine the key operation type, determine the artificial intelligence service call type based on the key operation type, and send a service call request including the artificial intelligence service call type to the terminal device.

[0152] The terminal device is used to obtain data to be processed based on the service call request, and to generate a model processing request based on the type of artificial intelligence service call and the data to be processed.

[0153] The terminal device or AI service routing determines the target AI service node from multiple AI service nodes based on at least one of the following: AI service call type, network status, and the operating status of multiple AI service nodes.

[0154] The target AI service node performs AI inference processing based on the model processing request and returns the processing result to the terminal device.

[0155] The terminal device outputs the processing results and can feed back the running status of the artificial intelligence service to the smart mouse 1. It can also provide status prompts to the user through the status indicator component 3.

[0156] The AI ​​service routing can maintain node information for each AI service node. Node information may include at least one of the following: node address, node capability label, node load, node availability status, and node network quality.

[0157] Artificial intelligence service routing can periodically update node information, or query the current status of some or all nodes before each service call.

[0158] When an AI service node is added, deactivated, or its model version changes, the node information maintained by the AI ​​service routing can be updated without altering the hardware structure and button event recognition logic of the smart mouse 1.

[0159] Therefore, the smart mouse 1 can call different artificial intelligence service nodes through a unified service call request interface, reducing the smart mouse 1's dependence on specific model brands, model interfaces, and deployment locations.

[0160] Example 11: Low Power Consumption Control The smart mouse 1 may also include a motion detection component. This motion detection component may include a three-axis accelerometer and be connected to the processor via an integrated circuit bus or a serial peripheral interface.

[0161] The motion detection component is used to detect the motion state and / or posture state of the mouse body 2. The processor can read the three-axis acceleration data at a preset sampling frequency and determine whether the mouse body 2 is stationary based on the three-axis acceleration data.

[0162] In one example, when the change in acceleration is less than or equal to a preset change threshold and the duration reaches a preset static duration, the processor determines that the detected data meets the preset static condition and controls the communication component to enter a first low-power state.

[0163] In the first low-power state, the communication components can reduce the radio frequency transmission frequency, extend the communication polling cycle, or shut down some unnecessary circuits.

[0164] When no valid key event is detected within a longer preset time period, and the detection data continues to meet the preset static conditions, the processor can control the communication component to enter a second low-power state.

[0165] The power consumption of the second low-power state is lower than that of the first low-power state. In the second low-power state, the RF transmitting unit can be turned off, while the processor retains key interrupts and motion detection interrupts.

[0166] When the acceleration change detected by the motion detection component exceeds a preset threshold, or when the processor detects a valid key event, the processor controls the communication component to exit the low-power state and return to the working state.

[0167] For example, the sampling frequency of the motion detection component can be 50 Hz, the preset change threshold can be 0.05g, the preset stillness duration can be 5 seconds, and the waiting time to enter the second low power state can be 30 minutes.

[0168] Example 12: Power Management The mouse body 2 can also house a rechargeable battery and power management components. The rechargeable battery can be a rechargeable lithium battery.

[0169] The power management component is connected to the rechargeable battery, the processor, the communication component, and the status indicator component 3, respectively. The power management component can perform at least one of the following functions: charging management, battery voltage detection, overcurrent protection, and undervoltage protection.

[0170] The processor determines the remaining battery power based on the battery status information output by the power management component. When the remaining battery power is less than or equal to a preset battery power threshold, the processor controls the status indicator component 3 to output a low battery warning.

[0171] For example, the preset power threshold can be 20% of the total battery capacity. When the remaining power is below this threshold, the multi-color light-emitting unit can output a flashing yellow warning.

[0172] The mouse body 2 also includes a scroll wheel assembly 23. The scroll wheel assembly 23 includes a scroll wheel, a rotation axis, a magnetic component, and a position adjustment component.

[0173] The rotating shaft may include a fitting shaft that mates with the roller and a magnetic fitting part connected to the fitting shaft. The roller is fitted onto the fitting shaft and is able to rotate with the rotating shaft.

[0174] The magnetic mating part can be made of a soft magnetic material. The outer periphery of the magnetic mating part has multiple drag-forming structures arranged circumferentially. These drag-forming structures can be periodically arranged teeth, protrusions, grooves, or serrated contours.

[0175] The magnetic component can be a permanent magnet. A position adjustment mechanism is used to move the magnetic component between a first position and a second position. The position adjustment mechanism can be manually operated by the user or driven by a processor-controlled drive assembly.

[0176] When the magnetic component is in the first position, it engages magnetically with the magnetic mating part. As the roller rotates, multiple resistance-forming structures move sequentially relative to the magnetic component, thereby generating a periodically changing magnetic force and applying intermittent rotational resistance to the roller, giving the user a ratchet-like rolling feel.

[0177] When the magnetic component is in the second position, the distance between the magnetic component and the magnetic mating part increases or the two are offset from each other, which reduces the magnetic force, thereby reducing or eliminating the intermittent rotational resistance of the roller, and putting the roller in a free rolling or low-resistance rolling state.

[0178] In one specific structure, the position adjustment component is movably disposed inside the mouse body 2, and can drive the magnetic component to switch between a position abutting the magnetic mating part and a position away from the magnetic mating part.

[0179] The scroll wheel assembly 23 is an optional additional structure of the smart mouse 1. It can improve the scroll wheel operation experience, but does not affect the implementation of core functions such as button event recognition, artificial intelligence service request generation and artificial intelligence service node invocation.

[0180] The technical features described in the various embodiments of this application can be combined with each other as long as there is no contradiction. For example, the wireless receiver caching scheme can be implemented in combination with any one or more of the following: a custom button mapping scheme, an abnormal switching scheme for artificial intelligence service nodes, an offline processing scheme, and a low-power control scheme.

[0181] The above embodiments have provided a relatively detailed description of this application, but do not imply that this application is limited to the specific embodiments described above. Any equivalent substitutions or modifications made by those skilled in the art to the above structures, processes, parameters, and data formats without departing from the concept of this application should fall within the scope of protection claimed in this application.

[0182] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0183] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of protection claimed by the present invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart mouse, comprising: The mouse itself; A button assembly is disposed on the mouse body, and the button assembly includes at least one function button for generating a button event signal in response to the user's button operation. A voice acquisition component, located on the mouse body, is used to acquire the user's voice signal; A communication component, located on the mouse body, is used to establish a communication connection with a terminal device; A memory is used to store key mapping information, which is used to represent the correspondence between different key operation types and different artificial intelligence service call types; The processor is connected to the button assembly, the voice acquisition assembly, the communication assembly, and the memory, respectively. The processor is configured as follows: Obtain the key event signal; The corresponding key operation type is determined based on the time sequence characteristics of the key event signal; The target AI service call type is determined based on the key operation type and the key mapping information; Generate a service call request corresponding to the target artificial intelligence service call type, wherein the service call request includes the voice signal and / or a data acquisition instruction for instructing the terminal device to acquire input data; The communication component sends the service call request to the terminal device, so that the terminal device calls the target artificial intelligence service according to the service call request and outputs the processing result returned by the target artificial intelligence service.

2. The smart mouse according to claim 1, characterized in that, The button assembly includes at least one side button disposed on the side of the mouse body and at least one top button disposed on the top of the mouse body; The side key and the top key are respectively connected to different signal input terminals of the processor.

3. The smart mouse according to claim 1, characterized in that, The time series features include at least one of the following: key press duration, key release duration, time interval between two adjacent key events, number of key events within a preset time period, and trigger order of different function keys. The key operation types include single-click operation, double-click operation, long-press operation, continuous click operation, multi-key combination operation, and at least two of the following: a compound operation consisting of at least two basic key operations.

4. The smart mouse according to claim 1, characterized in that, The processor is also configured to: The communication component receives the mapping configuration data sent by the terminal device. Update the key mapping information in the memory according to the mapping configuration data; The mapping configuration data is generated by a desktop AI client running on the terminal device based on the configuration operations entered by the user in the visual configuration interface.

5. The smart mouse according to claim 1, characterized in that, The voice acquisition component includes a microphone, an analog front-end processing circuit, and a digital signal processing unit; The analog front-end processing circuit is connected to the microphone and the digital signal processing unit respectively, and is used to filter and / or amplify the raw speech signal collected by the microphone; The digital signal processing unit is connected to the processor and is used to perform noise reduction processing on the speech signal processed by the analog front-end processing circuit, and send the noise-reduced speech signal to the processor.

6. The smart mouse according to claim 1, characterized in that, It also includes a status indicator component, which is connected to the processor; The processor is also configured to: The communication component receives service status information sent by the terminal device. Based on the service status information, control the status indication component to output corresponding status prompts; The service status information includes at least one of the following: voice acquisition status, request sending status, artificial intelligence service processing status, processing result return status, network connection status, and abnormal status.

7. The smart mouse according to claim 1, characterized in that, The communication components include a Bluetooth communication unit and a 2.4GHz wireless communication unit; The mouse body is equipped with a communication mode switching component connected to the processor; The processor is configured to, based on the mode switching signal generated by the communication mode switching component, select to communicate directly with the terminal device via the Bluetooth communication unit, or to communicate with the terminal device via the 2.4GHz wireless communication unit and the wireless receiver paired with the smart mouse.

8. The smart mouse according to claim 7, characterized in that, The wireless receiver includes a receiver processing unit, a radio frequency communication unit, a buffer unit, and a terminal interface; The radio frequency communication unit, the buffer unit, and the terminal interface are respectively connected to the receiver processing unit; The receiver processing unit is configured to: The service call request sent by the smart mouse is received through the radio frequency communication unit; When the terminal device is in a not-ready state, the service call request is cached in the cache unit; And after the terminal device is restored to the ready state, the cached service call request is sent to the terminal device through the terminal interface.

9. The smart mouse according to claim 1, characterized in that, It also includes a motion detection component, which is connected to the processor and is used to detect the motion state and / or posture state of the mouse body; The processor is also configured to: When the detection data output by the motion detection component meets the preset static condition, the communication component is controlled to enter a first low-power state. If no valid key event is detected within a preset time period and the detection data continuously meets the preset static condition, the communication component is controlled to enter a second low-power state with power consumption lower than the first low-power state. And when the detection data meets the preset wake-up conditions or a valid key event is detected, the communication component is controlled to return to the working state.

10. The smart mouse according to claim 1, characterized in that, The mouse body includes a scroll wheel assembly; The roller assembly includes a roller, a rotating shaft, a magnetic component, and a position adjusting component. The rotating shaft has a magnetic mating part that magnetically engages with the magnetic component. The outer periphery of the magnetic mating part is provided with a plurality of resistance forming structures arranged circumferentially. The position adjustment component is used to drive the magnetic component to move between a first position and a second position; At the first position, the magnetic element engages magnetically with the magnetic mating part, so that the resistance forming structure generates intermittent rotational resistance during the rotation of the roller; In the second position, the magnetic element and the magnetic mating part are spaced apart to reduce or eliminate the intermittent rotational resistance.

11. A method for invoking artificial intelligence services based on a smart mouse, characterized in that, The method, applied to a terminal device communicating with a smart mouse, includes: The system receives a service call request sent by the smart mouse. The service call request is determined by the smart mouse based on the key event signals generated by the function keys. The service call request includes an artificial intelligence service call type and voice data and / or data acquisition instructions. Parse the service call request to obtain the AI ​​service call type; If the service call request includes the data acquisition instruction, the data to be processed is acquired from the terminal device according to the data acquisition instruction. Generate a model processing request based on the AI ​​service call type and the data to be processed and / or the voice data; The target AI service node is determined from multiple candidate AI service nodes; Send the model processing request to the target artificial intelligence service node; Receive the processing result returned by the target artificial intelligence service node; The processing result is output through the terminal device.

12. The method according to claim 13, characterized in that, The step of determining the target artificial intelligence service node from multiple candidate artificial intelligence service nodes includes: The system acquires at least one of the following: the geographical location information of the terminal device, the network status information between the terminal device and each candidate AI service node, the load information of each candidate AI service node, the capability tags of each candidate AI service node, and user preference information. Based on the information obtained, determine the node evaluation information corresponding to each candidate artificial intelligence service node; And determine the target artificial intelligence service node from the plurality of candidate artificial intelligence service nodes based on the node evaluation information.

13. The method according to claim 12, characterized in that, The capability tags are used to characterize at least one capability dimension of the corresponding candidate artificial intelligence service node in natural language generation, speech recognition, image understanding, document processing, code processing, translation, information retrieval, and tool invocation; The step of determining the node evaluation information corresponding to each candidate artificial intelligence service node based on the acquired information includes: The target capability dimension is determined based on the type of AI service call. Based on at least one of the following factors for each candidate AI service node in the target capability dimension: capability parameters, network transmission latency, and node load, the node evaluation information corresponding to each candidate AI service node is determined.

14. The method according to claim 13, characterized in that, After sending the model processing request to the target AI service node, the method further includes: Detect whether the target AI service node meets the abnormal switching conditions; If the target AI service node meets the abnormal switching conditions, a backup AI service node is determined from the candidate AI service nodes other than the target AI service node. And send the model processing request to the backup artificial intelligence service node; The abnormal switching conditions include at least one of the following: request timeout, connection interruption, node load exceeding a preset load threshold, and node returning abnormal information. Also includes: When it is detected that the terminal device cannot access the multiple candidate artificial intelligence service nodes, it is determined whether the task corresponding to the model processing request meets the local processing conditions. If the local processing conditions are met, the model processing request is sent to the local artificial intelligence processing module set in the terminal device; The local artificial intelligence processing module processes the model processing request to obtain the local processing result; It also outputs the local processing result and sends service status information to the smart mouse to indicate the offline response status.

15. An artificial intelligence service invocation system, characterized in that, This includes smart mice, terminal devices, and multiple artificial intelligence service nodes; The smart mouse is communicatively connected to the terminal device, and the terminal device is communicatively connected to the plurality of artificial intelligence service nodes; The smart mouse is used for: In response to key event signals generated by function keys, determine the corresponding key operation type; The type of AI service call is determined based on the type of key operation. And send a service call request, including the type of the artificial intelligence service call, to the terminal device; The terminal device is used for: Obtain the data to be processed based on the service call request; The model processing request is generated based on the AI ​​service call type and the data to be processed. The target artificial intelligence service node is determined from the plurality of artificial intelligence service nodes based on at least one of the artificial intelligence service call type, network status, and running status of the plurality of artificial intelligence service nodes; Send the model processing request to the target artificial intelligence service node; Receive the processing result returned by the target artificial intelligence service node; And output the processing result; The target AI service node is used to perform AI inference processing according to the model processing request and return the processing result to the terminal device.

Images