Router based on real number fusion of multiple AI agents

Abstract

The invention relates to the technical field of digital routers, in particular to a router based on real number fusion of multiple AI agents. The technical problem is that an existing router cannot realize cross-system intelligent collaborative decision; according to the technical scheme, the router based on multi-AI agent real number fusion comprises a core processing module, a data transmission module, a storage module, a visual gesture recognition module, an AI voice dialogue interaction module, an LED display module, a heat dissipation module, an agent terminal, a tower-type server and a base module. The system coordinates the work of each module through the core processing module, constructs a digital capability dynamic distribution mechanism similar to a network router, achieves the data interaction and cooperative calculation of multiple AI agents, achieves the efficient cooperative work of multiple post agents in the park, breaks through the traditional information island phenomenon, and improves the safety of the park. The system data of the park can be interacted and shared in real time, and a unified intelligent decision-making system is formed.

Description

Technical Field

[0001] The present invention relates to the technical field of digital routers, and in particular to a router based on multi-AI agent real number fusion. Background Art

[0002] In today's digital age, traditional industrial parks and enterprises face many difficulties in the process of digital transformation. Traditional routers have a single function, focusing only on data transmission and maintaining network connections, and are almost blank in the integrated application of multiple AI agents. Multi-AI agent technology can achieve more complex tasks and decisions through the collaboration of multiple agents, but traditional routers are unable to effectively support the collaborative work of multiple AI agents due to the limitations of their architecture and processing power.

[0003] Traditional routers have extremely limited human-computer interaction methods, relying mainly on a text-based configuration interface. This interaction method not only requires operators to have professional network knowledge, but also has cumbersome operating procedures. The existing technical system makes it difficult to form an effective closed loop, and there is a lack of close connection between the data generated by physical equipment in the park and the business processes and decision-making models in the digital system.

[0004] In the campus conference management scenario, the existing conference system lacks deep integration with network equipment and the overall campus digital system; information processing during the meeting is still mainly manual, inefficient and error-prone; task dispatching, intelligent scheduling, task progress tracking and performance evaluation also mostly rely on manual operation and experience judgment, lacking intelligence and automation.

[0005] To sum up, the existing technologies in campus network communication, human-computer interaction, digital-physical integration, and conference management can no longer meet the needs of today's digital development in parks. There is an urgent need for an innovative solution that has multimodal human-computer interaction, multi-AI intelligent agent integration functions, and can effectively promote digital-physical integration and ecological collaboration in parks.

[0006] Therefore, to address the above problems, a router based on the real-number fusion of multiple AI agents is proposed. The core processing module coordinates the work of each module, constructs a dynamic allocation mechanism of digital capabilities similar to that of a network router, realizes data interaction and collaborative computing of multiple AI agents, and realizes efficient collaborative work of multiple positions in the park. It breaks the traditional information island phenomenon, enables real-time interaction and sharing of data from various systems in the park, and forms a unified intelligent decision-making system. Summary of the Invention

[0007] In order to overcome the problem that in the process of digital transformation of traditional industrial parks, existing routers only have basic data transmission functions and are unable to support the collaborative work of multiple AI intelligent bodies, resulting in the formation of information islands among various systems in the park and the inability to achieve cross-system intelligent collaborative decision-making.

[0008] The technical solution of the present invention is: a router based on multi-AI agent real number fusion, comprising: The core processing module adopts a multi-core parallel processor architecture with built-in cache to handle data requests and collaborative computing of multiple AI agents; Data transmission module, including high-speed Ethernet interface and wireless communication module, used for data transmission of equipment within the park; Storage module, including flash memory and random access memory, used to store operating systems, configuration files, and AI model data; A visual gesture recognition module, including an infrared camera array and a depth sensor, is used to capture user gestures and convert them into operational instructions; AI voice dialogue interaction module, including microphone array and speech recognition engine, for voice command recognition and natural language interaction; LED display module, using a high-resolution display screen to display the operating status of the intelligent body and data visualization information; A heat dissipation module, including a top cover exhaust fan and a ventilation base cover, forming an air convection channel; Intelligent terminal, connected via wired or wireless mode, used for user interaction with the router; Tower server with built-in multi-AI agent collaborative algorithm for performing collaborative computing tasks; Base module, including wheels and adjustable feet, for equipment movement and fixed support.

[0009] The core processing module acts as a central scheduler and adopts a three-level processing architecture, including: Access layer: connects various functional modules through the data bus to realize data collection and instruction distribution; Coordination layer: runs the agent coordination algorithm and dynamically allocates computing resources; Decision-making layer: executes cross-agent collaborative decision-making logic.

[0010] The hybrid transmission network is constructed through the data transmission module, including: Connect fixed devices via Gigabit Ethernet interface; Access to mobile terminals via Wi-Fi wireless modules; High-bandwidth device interconnection is achieved through optical fiber interfaces.

[0011] A unified operation interface is provided through the intelligent agent terminal, which supports visual monitoring of the operating status of each intelligent agent, cross-system task scheduling instructions, and abnormal situation warning prompts.

[0012] During the data collection phase, the visual gesture recognition module collects operator gesture commands through an infrared camera; the AI ​​voice dialogue interaction module receives voice commands through a microphone array; and various IoT devices report real-time data through the data transmission module. During the data processing phase, the main processor chip of the core processing module performs general computing tasks, while the coprocessor, dedicated to AI model inference computing, and the GPU accelerator card of the tower server handles deep learning tasks. During the intelligent agent collaboration phase, the internal data exchange matrix is ​​used to achieve the following: The security agent shares personnel location data with the logistics agent; Energy intelligence obtains equipment operation data to optimize power supply strategies; The production agent adjusts the production schedule based on supply chain data; During the decision-making execution stage, the LED display module visually displays the collaborative decision-making results, the intelligent terminal pushes execution instructions to personnel in various positions, and sends control commands to the execution equipment through the data transmission module.

[0013] Preferably, the work of each module is coordinated through the core processing module, and a dynamic allocation mechanism of digital capabilities similar to that of a network router is constructed to realize data interaction and collaborative computing of multiple AI intelligent bodies, thereby achieving efficient collaborative work of intelligent bodies in multiple positions within the park, breaking the traditional information island phenomenon, and enabling data from various systems in the park to interact and share in real time, forming a unified intelligent decision-making system.

[0014] Preferably, the core processing module includes a main processor chip, a coprocessor, a cache and a data bus; the main processor chip adopts a 16-core 32-thread architecture with a main frequency of 3.2GHz~4.5GHz; the coprocessor is dedicated to AI model inference calculations; the cache capacity is 32MB~64MB; the bandwidth of the data bus is 256bit; when in use, different types of computing tasks are assigned to the most suitable processor unit, the computing tasks are processed by the main processor chip, and the AI ​​model inference is specially processed by the coprocessor, the data access delay is reduced through the cache, and the high-bandwidth data bus is used to ensure the data interaction efficiency between the processor units.

[0015] Preferably, the data transmission module includes a Gigabit Ethernet interface, a Wi-Fi wireless module, a fiber optic interface and an internal data exchange matrix; the Gigabit Ethernet interface supports 10 / 100 / 1000Mbps adaptation; the Wi-Fi wireless module supports 2.4GHz and 5GHz dual-bands; the fiber optic interface supports 10Gbps data transmission; the internal data exchange matrix adopts a Crossbar architecture; when in use, the optimal transmission method is selected according to different scenarios, conventional devices are connected through the Ethernet interface, mobile devices are accessed through the Wi-Fi module, and high-bandwidth demand devices use the fiber optic interface. Data exchange between internal modules is achieved through the Crossbar architecture exchange matrix to achieve non-blocking transmission, realizing high-speed data transmission of equipment within the park.

[0016] Preferably, the visual gesture recognition module includes an infrared camera, a depth sensor, an image processing unit and a gesture recognition algorithm module; the infrared camera has a resolution of 1920×1080 and a frame rate of 60fps; the depth sensor has a detection distance of 0.5m~5m and an accuracy of ±1cm; the image processing unit has a built-in convolutional neural network accelerator; the gesture recognition algorithm module supports click, slide, zoom and custom gesture recognition; when in use, the infrared camera and the depth sensor work together to collect the spatial position and motion trajectory of the user's gesture, which is processed in real time by the image processing unit, and the gesture recognition algorithm module converts the gesture action into operation instructions to realize contactless human-computer interaction. Through natural and intuitive gesture interaction, in scenarios such as park inspection and equipment debugging, staff can complete operations without touching the equipment.

[0017] Preferably, the AI ​​voice dialogue interaction module includes a microphone array, a voice preprocessing unit, a voice recognition engine and a natural language understanding module; the microphone array consists of multiple groups of omnidirectional microphones and supports beamforming; the voice preprocessing unit includes noise reduction and echo cancellation functions; the voice recognition engine supports Chinese, English and dialect recognition; the natural language understanding module is based on the Transformer architecture; when in use, the beamforming technology of the microphone array is used to focus on the target sound source, the voice preprocessing unit eliminates environmental noise and echo interference, the voice recognition engine converts voice into text, and the natural language understanding module parses user intentions and generates responses, achieving a natural and smooth voice interaction experience, and realizing high-accuracy voice interaction in a noisy campus environment.

[0018] Preferably, the LED display module includes a main display screen, a status indicator light, a touch control panel and a display driving circuit; the main display screen has a resolution of 3840×2160; the status indicator light uses RGB LED; the touch control panel supports multi-touch; the display driving circuit supports the HDR10 standard; when in use, the main display screen displays the three-dimensional map of the park and the heat map of the intelligent body distribution, the status indicator light reflects the health status of the system in real time, the touch control panel supports gesture operation, and the display driving circuit ensures high-quality visual presentation effects, realizing a multi-dimensional visual display of the park's operating status, so that managers can intuitively understand the working status of each intelligent body and the overall operation of the park.

[0019] Preferably, the heat dissipation module includes a top cover exhaust fan, a ventilation base cover, a heat pipe radiator and a temperature sensor; the top cover exhaust fans are 6 groups in number, distributed in a ring shape, with a rotation speed of 800~2000RPM; the ventilation base cover has an opening rate of 60%~70%, and an aperture of 3mm~5mm; the heat pipe radiator is connected to the CPU and GPU chips; the temperature sensors are distributed inside the device; when in use, the device temperature is monitored in real time through the temperature sensor, and the exhaust fan speed is intelligently adjusted. The heat pipe radiator transfers the core heat to the heat sink, and the ventilation base cover and the top cover exhaust fan form a vertical air duct to achieve efficient heat dissipation, ensuring the stability of the device during high-load operation.

[0020] Preferably, the intelligent terminal includes a touch screen, physical buttons, a network interface and a wireless module; the touch screen resolution is 2560×1600; the physical buttons include a power button and a function key; the network interface supports wired network connection; the wireless module supports Bluetooth 5.0 and Wi-Fi connection; when in use, user input is received through the touch screen and physical buttons, the network interface and wireless module realize data interaction with the core system, and the terminal has a built-in lightweight AI model to achieve fast local processing.

[0021] As a preferred option, the tower server includes a motherboard, memory, storage array and GPU accelerator card; the motherboard supports dual-core CPUs; the memory capacity is 128GB~256GB, and the frequency is 3200MHz; the storage array uses NVMe SSD with a total capacity of 8TB~16TB; the GPU accelerator card is used for AI model inference; when in use, general computing power is provided by the dual-core CPU, large-capacity high-speed memory supports multi-task parallel processing, the NVMe storage array ensures fast data access, and the GPU accelerator card specifically handles AI computing tasks to achieve optimized allocation of computing resources.

[0022] Preferably, the base module includes rollers, feet, anti-collision strips and aviation plugs; there are multiple sets of rollers with a diameter of 75mm and a brake device; there are multiple sets of feet with an adjustable height range of 30mm~50mm; the anti-collision strips are 5mm thick and made of rubber; the aviation plug is used for power and data connections; when in use, the rollers are used to easily move the device, and after reaching the designated position, the feet provide stable support, the anti-collision strips protect the edges of the device, and the aviation plug ensures stable and reliable power and data connections.

[0023] Beneficial effects of the present invention: By coordinating the work of each module through the core processing module, a dynamic allocation mechanism for digital capabilities similar to that of a network router is constructed to realize data interaction and collaborative computing among multiple AI agents. This enables efficient collaborative work among multiple intelligent agents in the park, breaking the traditional information island phenomenon and enabling real-time interaction and sharing of data from various systems in the park to form a unified intelligent decision-making system. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 Shown is a schematic diagram of the router assembly based on multi-AI agent real number fusion of the present invention; Figure 2 Shown is a schematic diagram of the installation of a router device based on multi-AI agent real number fusion of the present invention; Figure 3 Shown is a schematic diagram of the top view of the router based on multi-AI agent real number fusion of the present invention; Figure 4 Shown is a schematic diagram of the router system architecture based on multi-AI agent real number fusion of the present invention; Explanation of the accompanying symbols: 1. Main display screen; 2. Infrared camera array; 3. Microphone array; 4. Top cover exhaust fan; 5. Ventilation base cover; 6. Intelligent terminal; 7. Tower server; 8. Roller; 9. Support foot; 10. Status indicator light; 11. Heat pipe radiator; 12. Anti-collision strip; 13. Aviation plug. DETAILED DESCRIPTION

[0025] The present invention will be further described below with reference to the accompanying drawings and examples.

[0026] See also Figure 1 The present invention provides an embodiment: a router based on multi-AI agent real number fusion, comprising: The core processing module adopts a multi-core parallel processor architecture with built-in cache to handle data requests and collaborative computing of multiple AI agents; Data transmission module, including high-speed Ethernet interface and wireless communication module, used for data transmission of equipment within the park; Storage module, including flash memory and random access memory, used to store operating systems, configuration files, and AI model data; A visual gesture recognition module, including an infrared camera array 2 and a depth sensor, is used to capture user gestures and convert them into operation instructions; AI voice dialogue interaction module, including a microphone array 3 and a speech recognition engine, for voice command recognition and natural language interaction; LED display module, using a high-resolution display screen to display the operating status of the intelligent body and data visualization information; The heat dissipation module includes a top cover exhaust fan 4 and a ventilation base cover 5 to form an air convection channel; Intelligent terminal 6, connected via wired or wireless mode, for user interaction with the router; Tower Server 7, with built-in multi-AI agent collaborative algorithm for performing collaborative computing tasks; The base module includes rollers 8 and adjustable feet 9, which are used for moving and fixing the equipment.

[0027] The core processing module acts as a central scheduler and adopts a three-level processing architecture, including: Access layer: connects various functional modules through the data bus to realize data collection and instruction distribution; Coordination layer: runs the agent coordination algorithm and dynamically allocates computing resources; Decision-making layer: executes cross-agent collaborative decision-making logic.

[0028] The hybrid transmission network is constructed through the data transmission module, including: Connect fixed devices via Gigabit Ethernet interface; Access to mobile terminals via Wi-Fi wireless modules; High-bandwidth device interconnection is achieved through optical fiber interfaces.

[0029] A unified operation interface is provided through the intelligent terminal 6, which supports visual monitoring of the operating status of each intelligent agent, the issuance of cross-system task scheduling instructions, and early warning prompts of abnormal situations.

[0030] During the data collection phase, the visual gesture recognition module collects operator gesture commands through an infrared camera; the AI ​​voice dialogue interaction module receives voice commands through the microphone array 3; and various IoT devices report real-time data through the data transmission module. During the data processing phase, the main processor chip of the core processing module performs general computing tasks, while the coprocessor is dedicated to AI model inference computing. The GPU accelerator card of the tower server 7 handles deep learning tasks. During the intelligent agent collaboration phase, the internal data exchange matrix is ​​used to achieve the following: The security agent shares personnel location data with the logistics agent; Energy intelligence obtains equipment operation data to optimize power supply strategies; The production agent adjusts the production schedule based on supply chain data; During the decision execution phase, the LED display module visually displays the collaborative decision results, the intelligent terminal 6 pushes execution instructions to personnel in various positions, and issues control commands to the execution equipment through the data transmission module.

[0031] See also Figure 2 and Figure 3In this embodiment, the core processing module includes a main processor chip, a coprocessor, a cache and a data bus; the main processor chip adopts a 16-core 32-thread architecture with a main frequency of 3.2GHz~4.5GHz; the coprocessor is dedicated to AI model inference calculation; the cache capacity is 32MB~64MB; the bandwidth of the data bus is 256bit; when in use, different types of computing tasks are assigned to the most suitable processor unit, the computing tasks are processed by the main processor chip, and the AI ​​model inference is specially processed by the coprocessor. The data access delay is reduced by the cache, and the high-bandwidth data bus is used to ensure the data interaction efficiency between the processor units. The data transmission module includes a Gigabit Ethernet interface and a Wi-Fi wireless module , fiber optic interface and internal data exchange matrix; Gigabit Ethernet interface supports 10 / 100 / 1000Mbps adaptive; Wi-Fi wireless module supports 2.4GHz and 5GHz dual-band; fiber optic interface supports 10Gbps data transmission; internal data exchange matrix adopts Crossbar architecture; when in use, the optimal transmission method is selected according to different scenarios, conventional devices are connected through Ethernet interface, mobile devices are connected through Wi-Fi module, high bandwidth demand devices use fiber optic interface, data exchange between internal modules is achieved through Crossbar architecture exchange matrix to achieve non-blocking transmission, realizing high-speed data transmission of equipment in the park, visual gesture recognition module includes infrared camera, depth sensor, image Processing unit and gesture recognition algorithm module; infrared camera resolution 1920×1080, frame rate 60fps; depth sensor, detection distance 0.5m~5m, accuracy ±1cm; image processing unit built-in convolutional neural network accelerator; gesture recognition algorithm module, supports click, slide, zoom and custom gesture recognition; when in use, the infrared camera and depth sensor work together to collect the spatial position and motion trajectory of the user's gesture, which is processed in real time by the image processing unit. The gesture recognition algorithm module converts gesture actions into operation instructions to achieve contactless human-computer interaction. Through natural and intuitive gesture interaction, in scenarios such as park inspection and equipment debugging, staff can complete operations without touching the equipment. The AI ​​voice dialogue interaction module includes a microphone array 3, a voice preprocessing unit, a voice recognition engine and a natural language understanding module; the microphone array 3 consists of multiple groups of omnidirectional microphones and supports beamforming; the voice preprocessing unit includes noise reduction and echo cancellation functions; the voice recognition engine supports Chinese, English and dialect recognition; the natural language understanding module is based on the Transformer architecture; when in use, the beamforming technology of the microphone array 3 is used to focus on the target sound source, the voice preprocessing unit eliminates environmental noise and echo interference, the voice recognition engine converts voice into text, and the natural language understanding module parses user intentions and generates responses to achieve a natural and smooth voice interaction experience, and achieves high-accuracy voice interaction in a noisy campus environment.

[0032] See also Figure 4In this embodiment, the LED display module includes a main display screen 1, a status indicator light 10, a touch control panel and a display driving circuit; the main display screen 1 has a resolution of 3840×2160; the status indicator light 10 uses RGB LED; the touch control panel supports multi-touch; the display driving circuit supports the HDR10 standard; when in use, the main display screen 1 displays a three-dimensional map of the park and a heat map of the intelligent body distribution, the status indicator light 10 reflects the health status of the system in real time, the touch control panel supports gesture operation, and the display driving circuit ensures high-quality visual presentation effects, realizing a multi-dimensional visualization of the park's operating status, so that managers can intuitively understand the working status of each intelligent body and the overall operation of the park. The heat dissipation module includes a top cover exhaust fan 4, a ventilation base cover 5, a heat pipe radiator 11 and a temperature sensor; the top cover exhaust fan 4, the number is 6 groups, distributed in a ring, the speed is 800~2000RPM; the ventilation base cover 5 has an opening rate of 60%~70%, and an aperture of 3mm~5mm; the heat pipe radiator 11 is connected to the CPU and GPU chips; the temperature sensor is distributed inside the device; when in use, the device is monitored in real time through the temperature sensor Temperature, intelligently adjust the exhaust fan speed, the heat pipe radiator 11 conducts the core heat to the heat sink, the ventilation base cover 5 and the top cover exhaust fan 4 form a vertical air duct to achieve efficient heat dissipation, ensuring the stability of the equipment during high-load operation, the intelligent terminal 6 includes a touch screen, physical buttons, a network interface and a wireless module; the touch screen resolution is 2560×1600; the physical buttons include a power button and a function button; the network interface supports wired network connection; the wireless module supports Bluetooth 5.0 and Wi-Fi connection; when in use, user input is received through the touch screen and physical buttons, and the network interface and wireless module realize data interaction with the core system. The terminal has a built-in lightweight AI model to achieve fast local processing. The tower server 7 includes a motherboard, memory, storage array and GPU accelerator card; the motherboard supports dual-core CPU; the memory capacity is 128GB~256GB, the frequency is 3200MHz; the storage array uses NVMe SSD, total capacity 8TB~16TB; GPU accelerator card is used for AI model inference; when in use, general computing power is provided by dual-core CPU, large-capacity high-speed memory supports multi-task parallel processing, NVMe storage array ensures fast data access, GPU accelerator card specializes in processing AI computing tasks, and realizes optimized allocation of computing resources. The base module includes rollers 8, feet 9, anti-collision strips 12 and aviation plugs 13; rollers 8 are provided in multiple groups, with a diameter of 75mm and a brake device; feet 9 are provided in multiple groups, with an adjustable height range of 30mm~50mm; anti-collision strips 12 are 5mm thick and made of rubber; aviation plugs 13 are used for power and data connections; when in use, the device can be easily moved through the rollers 8, and after reaching the designated position, the feet 9 provide stable support, the anti-collision strips 12 protect the edges of the device, and the aviation plugs 13 ensure stable and reliable power and data connections.

[0033] When working, the operator first moves the digital router to the designated location of the campus control center using the rollers 8 of the base module, adjusts the adjustable feet 9 to place the device firmly, and uses the anti-collision strips 12 to effectively protect the edges of the device from collision damage. The power cord and the campus backbone network optical fiber are connected through the aviation plug 13. After starting the device, the top cover exhaust fan 4 and the ventilation base cover 5 of the heat dissipation module automatically form a vertical air duct. The heat pipe radiator 11 quickly dissipates heat from the CPU and GPU chips. The temperature sensor monitors in real time and intelligently adjusts the fan speed to ensure stable operation of the device at an ambient temperature of 25-35°C. The main processor chip and coprocessor of the core processing module complete self-tests, the flash memory and random access memory of the storage module load the operating system and basic AI models, and the Gigabit Ethernet interface, Wi-Fi wireless module and optical fiber interface of the data transmission module simultaneously establish communication links with various devices in the campus, forming a hybrid transmission network. Users can operate the intelligent terminal 6 through the touch screen or physical buttons, or achieve contactless interaction through the infrared camera array 2 and depth sensor of the visual gesture recognition module: when the operator makes a "click" gesture, the infrared camera captures the gesture trajectory at a frame rate of 60fps, the depth sensor accurately locates the spatial coordinates, and the image processing unit recognizes the gesture in real time through the convolutional neural network accelerator and converts it into an operation instruction; at the same time, the microphone array 3 of the AI ​​voice dialogue interaction module focuses on the user's voice through beamforming technology. After noise reduction and echo cancellation processing, the voice recognition engine converts the voice into text, and the natural language understanding module analyzes the user's intention based on the Transformer architecture; the core processing module distributes the collected multimodal data through the cache and data bus to the main processor chip to handle general computing tasks, the coprocessor to execute AI model inference, and the GPU accelerator card of the tower server 7 to handle deep learning tasks. The NVMe SSD array of the storage module achieves fast data access with a capacity of 8TB to 16TB and a frequency of 3200MHz. The core processing module acts as a central dispatcher, coordinating various intelligent agents through a three-level processing architecture: the access layer collects visual, voice, and IoT device data via the data bus; the coordination layer runs the intelligent agent coordination algorithm and dynamically allocates computing resources; the decision-making layer executes cross-agent collaborative logic; the main display screen 1 of the LED display module displays the three-dimensional map of the campus and the intelligent agent distribution heat map at 4K resolution, and the status indicator light 10 reflects the system health status in real time through RGB LED; finally, the collaborative decision results are pushed to personnel in various positions through the intelligent agent terminal 6, and the data transmission module issues control commands to the execution equipment, forming a complete closed loop of "data collection-processing-collaboration-execution".

[0034] Through the above steps, the core processing module is used to coordinate the work of each module, build a dynamic allocation mechanism for digital capabilities similar to that of a network router, realize data interaction and collaborative computing of multiple AI agents, and achieve efficient collaborative work of multiple positions in the park. This breaks the traditional information island phenomenon and enables real-time interaction and sharing of data from various systems in the park to form a unified intelligent decision-making system.

[0035] The embodiments of the present invention are described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Various changes can be made within the scope of knowledge of those skilled in the art without departing from the spirit of the present invention.

Claims

1. A router based on multi-AI agent real number fusion, characterized by: include: The core processing module adopts a multi-core parallel processor architecture with built-in cache to handle data requests and collaborative computing of multiple AI agents; Data transmission module, including high-speed Ethernet interface and wireless communication module, used for data transmission of equipment within the park; Storage module, including flash memory and random access memory, used to store operating systems, configuration files, and AI model data; A visual gesture recognition module, including an infrared camera array (2) and a depth sensor, for capturing user gestures and converting them into operation instructions; AI voice dialogue interaction module, including microphone array (3) and speech recognition engine, for voice command recognition and natural language interaction; LED display module, using a high-resolution display screen to display the operating status of the intelligent body and data visualization information; The heat dissipation module includes a top cover exhaust fan (4) and a ventilation base cover (5) to form an air convection channel; Intelligent terminal (6), connected via wired or wireless means, for user interaction with the router; Tower server (7), with built-in multi-AI agent collaborative algorithm for performing collaborative computing tasks; The base module includes rollers (8) and adjustable feet (9) for moving and fixing the equipment.

2. The router based on multi-AI agent real number fusion according to claim 1, characterized in that: The core processing module includes a main processor chip, coprocessor, cache and data bus; the main processor chip adopts a 16-core 32-thread architecture with a main frequency of 3.2GHz~4.5GHz; the coprocessor is dedicated to AI model inference calculations; the cache capacity is 32MB~64MB; and the data bus bandwidth is 256bit.

3. The router based on multi-AI agent real number fusion according to claim 1, characterized in that: The data transmission module includes Gigabit Ethernet interface, Wi-Fi wireless module, optical fiber interface and internal data exchange matrix; Gigabit Ethernet interface supports 10 / 100 / 1000Mbps adaptive; The Wi-Fi wireless module supports 2.4GHz and 5GHz dual-bands; the optical fiber interface supports 10Gbps data transmission; and the internal data exchange matrix adopts a crossbar architecture.

4. The router based on multi-AI agent real number fusion according to claim 1, characterized in that: The visual gesture recognition module includes an infrared camera, a depth sensor, an image processing unit, and a gesture recognition algorithm module; the infrared camera has a resolution of 1920×1080 and a frame rate of 60fps; the depth sensor has a detection range of 0.5m~5m and an accuracy of ±1cm; the image processing unit has a built-in convolutional neural network accelerator; and the gesture recognition algorithm module supports click, slide, zoom, and custom gesture recognition.

5. The router based on multi-AI agent real number fusion according to claim 1, characterized in that: The AI ​​voice dialogue interaction module includes a microphone array (3), a voice preprocessing unit, a voice recognition engine, and a natural language understanding module; the microphone array (3) consists of multiple groups of omnidirectional microphones and supports beamforming; the voice preprocessing unit includes noise reduction and echo cancellation functions; the voice recognition engine supports Chinese, English, and dialect recognition; the natural language understanding module is based on the Transformer architecture.

6. The router based on multi-AI agent real number fusion according to claim 1, characterized in that: The LED display module includes a main display screen (1), a status indicator light (10), a touch control panel and a display drive circuit; the main display screen (1) has a resolution of 3840×2160; the status indicator light (10) uses RGB LED; the touch control panel supports multi-touch; The display driver circuit supports the HDR10 standard.

7. The router based on multi-AI agent real number fusion according to claim 1, characterized in that: The heat dissipation module includes a top cover exhaust fan (4), a ventilation base cover (5), a heat pipe radiator (11) and a temperature sensor; the top cover exhaust fans (4) are 6 groups in number and are distributed in a ring shape, with a rotation speed of 800~2000RPM; the ventilation base cover (5) has an opening rate of 60%~70% and an aperture of 3mm~5mm; the heat pipe radiator (11) is connected to the CPU and GPU chips; and the temperature sensors are distributed inside the device.

8. The router based on multi-AI agent real number fusion according to claim 1, characterized in that: The intelligent terminal (6) includes a touch screen, physical buttons, a network interface and a wireless module; the touch screen has a resolution of 2560×1600; the physical buttons include a power button and a function button; the network interface supports wired network connection; and the wireless module supports Bluetooth 5.0 and Wi-Fi connection.

9. The router based on multi-AI agent real number fusion according to claim 1, characterized in that: The tower server (7) includes a motherboard, memory, storage array and GPU accelerator card; the motherboard supports dual-channel CPU; the memory capacity is 128GB~256GB, the frequency is 3200MHz; the storage array uses NVMe SSD, with a total capacity of 8TB~16TB; the GPU accelerator card is used for AI model inference.

10. The router based on multi-AI agent real number fusion according to claim 1, characterized in that: The base module includes rollers (8), legs (9), anti-collision strips (12) and aviation plugs (13); the rollers (8) are provided in multiple groups, with a diameter of 75 mm and a brake device; the legs (9) are provided in multiple groups, with a height adjustable range of 30 mm to 50 mm; the anti-collision strips (12) are 5 mm thick and made of rubber; the aviation plugs (13) are used for power and data connection.

Images