A router based on real number fusion of multiple AI agents

Abstract

The utility model relates to digital router technical field especially, more particularly to a kind of router based on multi-AI intelligent agent real number fusion;Technical problem: the existing router cannot realize the intelligent collaborative decision of cross-system;Technical scheme: a kind of router based on multi-AI intelligent agent real number fusion, including core processing module, data transmission module, storage module, visual gesture recognition module, AI voice dialogue interaction module, LED display module, heat dissipation module, intelligent agent terminal, tower server and base module;The utility model works by core processing module coordination each module, constructs the digital ability dynamic distribution mechanism similar to network router, realizes the data interaction and collaborative computing of multi-AI intelligent agent, realizes the efficient collaborative work of multi-post intelligent agent in park, breaks the traditional information island phenomenon, so that park each system data can real-time interaction and sharing, forms unified intelligent decision system.

Description

Technical Field

[0001] This utility model relates to the field of digital router technology, and in particular to a router based on the real data fusion of multiple AI intelligent agents. Background Technology

[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 network connection maintenance, 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 cannot effectively support the collaborative work of multiple AI agents due to the limitations of architecture and processing power.

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

[0004] In the context of park meeting management, the existing meeting system, network equipment and overall park digital system lack deep integration; information processing during the meeting is still mainly done manually, which is inefficient and prone to errors; task assignment, intelligent scheduling, task progress tracking and performance evaluation also mostly rely on manual operation and experience judgment, lacking intelligence and automation.

[0005] In summary, existing technologies in areas such as campus network communication, human-computer interaction, data-real integration, and meeting management are no longer sufficient to meet the needs of today's digital development in campuses. There is an urgent need for an innovative solution that can effectively promote data-real integration and ecosystem collaboration in campuses by enabling multimodal human-computer interaction, multi-AI intelligent agent integration, and multimodal human-computer interaction.

[0006] Therefore, to address the above issues, we propose a router based on the real-data fusion of multiple AI agents. By coordinating the work of each module through a core processing module, a dynamic allocation mechanism for digital capabilities similar to a network router is constructed to realize data interaction and collaborative computing among multiple AI agents. This enables efficient collaborative work among multiple AI agents within the park, breaks down the traditional information silos, and allows data from various systems within the park to interact and share in real time, forming a unified intelligent decision-making system. Utility Model Content

[0007] In order to overcome the problem that existing routers only have basic data transmission functions and cannot support the collaborative work of multiple AI agents during the digital transformation of traditional industrial parks, resulting in information silos among various systems in the park and the inability to achieve intelligent collaborative decision-making across systems.

[0008] The technical solution of this utility model is: a router based on real-data fusion of multiple AI agents, comprising: The core processing module adopts a multi-core parallel processor architecture and has a built-in high-speed cache for processing data requests and collaborative computing from multiple AI agents; The data transmission module includes a high-speed Ethernet interface and a wireless communication module, used for data transmission of devices within the park; The storage module, including flash memory and random access memory, is used to store the operating system, configuration files, and AI model data; The visual gesture recognition module, including an infrared camera array and a depth sensor, is used to capture user gestures and convert them into operation commands; The AI ​​voice dialogue interaction module includes a microphone array and a voice recognition engine for voice command recognition and natural language interaction; The LED display module uses a high-resolution display screen to show the operating status and data visualization information of the intelligent agent; The heat dissipation module includes a top-mounted exhaust fan and a ventilation base cover, forming an air convection channel; Intelligent terminal, connected via wired or wireless means, is used for user interaction with router; Tower servers with built-in multi-AI agent collaborative algorithms for performing collaborative computing tasks; The base module, including casters and adjustable feet, is used for moving and securing the equipment.

[0009] The core processing module acts as the central scheduler, employing a three-tier processing architecture, specifically including: Access layer: Connects various functional modules through a data bus to realize data acquisition and command distribution; Coordination layer: Runs agent coordination algorithms and dynamically allocates computing resources; Decision layer: Executes collaborative decision-making logic across agents.

[0010] A hybrid transmission network is constructed through data transmission modules, specifically including: Connect to fixed devices via a gigabit Ethernet interface; Access to the mobile terminal via a Wi-Fi wireless module; High-bandwidth device interconnection is achieved through fiber optic interfaces.

[0011] The intelligent agent terminal provides a unified operating interface, supporting visual monitoring of the operating status of each intelligent agent, issuance of cross-system task scheduling instructions, and early warning prompts for abnormal situations.

[0012] During the data acquisition phase, the visual gesture recognition module captures operator gesture commands via an infrared camera; the AI ​​voice dialogue interaction module receives voice commands via a microphone array; 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 acceleration card of the tower server processes deep learning tasks; during the intelligent agent collaboration phase, it is achieved through an internal data exchange matrix. Security intelligent agents share personnel location data with logistics intelligent agents; Energy intelligent agents acquire equipment operation data to optimize power supply strategies; The production intelligence agent adjusts production scheduling based on supply chain data; During the decision-making and execution phase, the LED display module visually displays the collaborative decision-making results, the intelligent agent 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 core processing module coordinates the work of each module, and a dynamic allocation mechanism for digital capabilities similar to a network router is constructed to realize data interaction and collaborative computing among multiple AI agents. This enables efficient collaborative work among multiple AI agents within the park, breaks down traditional information silos, and allows data from various systems in the park to interact and share in real time, forming a unified intelligent decision-making system.

[0014] As a preferred 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 to 4.5GHz. The coprocessor is dedicated to AI model inference calculations. The cache capacity is 32MB to 64MB. The data bus bandwidth is 256bit. In use, different types of computing tasks are assigned to the most suitable processor units. The computing tasks are handled by the main processor chip, while AI model inference is handled by the coprocessor. Data access latency is reduced through the cache, and the high-bandwidth data bus ensures efficient data interaction between 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 auto-negotiation; the Wi-Fi wireless module supports dual-band 2.4GHz and 5GHz; the fiber optic interface supports 10Gbps data transmission; and the internal data exchange matrix adopts a crossbar architecture. 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 devices with high bandwidth requirements use the fiber optic interface. Data exchange between internal modules is achieved through a crossbar architecture exchange matrix to realize non-blocking transmission, enabling high-speed data transmission for devices 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, swipe, zoom, and custom gesture recognition. In use, the infrared camera and depth sensor work together to collect the spatial position and motion trajectory of the user's gestures. The image processing unit processes the data in real time, and the gesture recognition algorithm module converts the gestures into operation commands, enabling contactless human-computer interaction. Through natural and intuitive gesture interaction, staff can complete operations without touching the equipment in scenarios such as park inspection and equipment debugging.

[0017] As a preferred embodiment, 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 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. In use, the microphone array's beamforming technology focuses on the target sound source, the voice preprocessing unit eliminates environmental noise and echo interference, the voice recognition engine converts speech into text, and the natural language understanding module parses the user's intent and generates a response, achieving a natural and fluent voice interaction experience. This results in highly accurate voice interaction even in noisy park environments.

[0018] Preferably, the LED display module includes a main display screen, status indicator lights, a touch control panel, and a display driver circuit; the main display screen has a resolution of 3840×2160; the status indicator lights use RGB LEDs; the touch control panel supports multi-touch; the display driver circuit supports the HDR10 standard; in use, the main display screen displays a 3D map of the park and a heat map of the distribution of intelligent agents, the status indicator lights reflect the system's health status in real time, the touch control panel supports gesture operation, and the display driver circuit ensures high-quality visual presentation, realizing a multi-dimensional visualization of the park's operating status. Managers can intuitively understand the working status of each intelligent agent 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 fan consists of 6 groups arranged in a ring, with a speed of 800~2000 RPM; the ventilation base cover has an opening ratio of 60%~70% and an aperture of 3mm~5mm; the heat pipe radiator connects the CPU and GPU chips; the temperature sensor is distributed inside the device; during use, the temperature sensor monitors the device temperature in real time, intelligently adjusts the exhaust fan speed, the heat pipe radiator conducts core heat to the heat sink, and the ventilation base cover and top cover exhaust fan form a vertical airflow channel to achieve efficient heat dissipation and ensure the stability of the device under high load operation.

[0020] As a preferred embodiment, the intelligent terminal includes a touchscreen, physical buttons, a network interface, and a wireless module; the touchscreen has a resolution of 2560×1600; the physical buttons include a power button and function buttons; the network interface supports wired network connections; the wireless module supports Bluetooth 5.0 and Wi-Fi connections; during use, user input is received through the touchscreen and physical buttons, and data interaction with the core system is achieved through the network interface and wireless module. The terminal has a built-in lightweight AI model for rapid localization processing.

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

[0022] Preferably, the base module includes casters, feet, anti-collision strips, and an aviation connector; multiple sets of casters are provided, each 75mm in diameter, and equipped with a brake; multiple sets of feet are provided, with an adjustable height range of 30mm~50mm; the anti-collision strips are 5mm thick and made of rubber; the aviation connectors are used for power and data connections; during use, the casters allow for easy movement of the device, and once it reaches the designated position, the feet provide stable support, the anti-collision strips protect the edges of the device, and the aviation connectors ensure stable and reliable power and data connections;

[0023] The beneficial effects of this utility model are: By coordinating the work of each module through the core processing module, a dynamic allocation mechanism for digital capabilities similar to a network router is constructed to realize data interaction and collaborative computing among multiple AI agents. This enables efficient collaborative work among multiple AI agents within the park, breaks down traditional information silos, and allows data from various systems within the park to interact and share in real time, forming a unified intelligent decision-making system. Attached Figure Description

[0024] Figure 1 The diagram shown is a schematic of the router assembly based on real-data fusion of multiple AI agents according to this utility model. Figure 2 The diagram shown is an installation schematic of the router device based on real-data fusion of multiple AI intelligent agents according to this utility model. Figure 3 The diagram shown is a top view of the router based on real-data fusion of multiple AI agents according to this utility model. Figure 4 The diagram shown is a schematic of the router system architecture based on real-data fusion of multiple AI agents according to this utility model. Explanation of reference numerals in the attached diagram: 1. Main display screen; 2. Infrared camera array; 3. Microphone array; 4. Top cover exhaust fan; 5. Ventilation base cover; 6. Smart terminal; 7. Tower server; 8. Casters; 9. Support feet; 10. Status indicator light; 11. Heat pipe radiator; 12. Anti-collision strip; 13. Aviation connector. Detailed Implementation

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

[0026] Please see Figure 1 This utility model provides an embodiment: a router based on real-data fusion of multiple AI agents, comprising: The core processing module adopts a multi-core parallel processor architecture and has a built-in high-speed cache for processing data requests and collaborative computing from multiple AI agents; The data transmission module includes a high-speed Ethernet interface and a wireless communication module, used for data transmission of devices within the park; The storage module, including flash memory and random access memory, is used to store the operating system, configuration files, and AI model data; The visual gesture recognition module includes an infrared camera array 2 and a depth sensor, used to capture user gestures and convert them into operation commands; AI voice dialogue interaction module, including microphone array 3 and voice recognition engine, is used for voice command recognition and natural language interaction; The LED display module uses a high-resolution display screen to show the operating status and data visualization information of the intelligent agent; The heat dissipation module includes a top cover exhaust fan 4 and a ventilation base cover 5, forming an air convection channel; The intelligent terminal 6 is connected via wired or wireless means and is used for user interaction with the router; Tower server 7, with built-in multi-AI agent collaborative algorithm, is used to perform collaborative computing tasks; The base module, including casters 8 and adjustable feet 9, is used for moving and securing the equipment.

[0027] The core processing module acts as the central scheduler, employing a three-tier processing architecture, specifically including: Access layer: Connects various functional modules through a data bus to realize data acquisition and command distribution; Coordination layer: Runs agent coordination algorithms and dynamically allocates computing resources; Decision layer: Executes collaborative decision-making logic across agents.

[0028] A hybrid transmission network is constructed through data transmission modules, specifically including: Connect to fixed devices via a gigabit Ethernet interface; Access to the mobile terminal via a Wi-Fi wireless module; High-bandwidth device interconnection is achieved through fiber optic interfaces.

[0029] The intelligent agent terminal 6 provides a unified operation interface, supporting visual monitoring of the operating status of each intelligent agent, cross-system task scheduling command issuance, and early warning prompts for abnormal situations.

[0030] During the data acquisition phase, the visual gesture recognition module captures operator gesture commands via an infrared camera; the AI ​​voice dialogue interaction module receives voice commands via microphone array 3; various IoT devices report real-time data via 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 acceleration card of the tower server 7 processes deep learning tasks; during the intelligent agent collaboration phase, it is achieved through an internal data exchange matrix. Security intelligent agents share personnel location data with logistics intelligent agents; Energy intelligent agents acquire equipment operation data to optimize power supply strategies; The production intelligence agent adjusts production scheduling based on supply chain data; During the decision-making and execution phase, the LED display module visually displays the collaborative decision-making results, the intelligent agent terminal 6 pushes execution instructions to personnel in various positions, and sends control commands to the execution equipment through the data transmission module.

[0031] Please see 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 to 4.5GHz. The coprocessor is dedicated to AI model inference calculations. The cache capacity is 32MB to 64MB. The data bus bandwidth is 256 bits. In use, different types of computing tasks are assigned to the most suitable processor units. The computing tasks are processed by the main processor chip, while AI model inference is handled by the coprocessor. The cache reduces data access latency, and the high-bandwidth data bus ensures efficient data interaction between processor units. The data transmission module includes a gigabit Ethernet interface and a Wi-Fi wireless module. The system includes fiber optic interfaces and an internal data exchange matrix; the Gigabit Ethernet interface supports 10 / 100 / 1000Mbps auto-negotiation; the Wi-Fi module supports dual-band 2.4GHz and 5GHz; the fiber optic interface supports 10Gbps data transmission; the internal data exchange matrix adopts a Crossbar architecture; during use, the optimal transmission method is selected according to different scenarios. Conventional devices connect via Ethernet interfaces, mobile devices access via Wi-Fi modules, and devices with high bandwidth requirements use fiber optic interfaces. Data exchange between internal modules is achieved through a Crossbar architecture exchange matrix, enabling high-speed data transmission for devices within the park. The visual gesture recognition module includes an infrared camera, depth sensor, and image sensor. The system includes a processing unit and a gesture recognition algorithm module; an infrared camera with a resolution of 1920×1080 and a frame rate of 60fps; a depth sensor with a detection distance of 0.5m~5m and an accuracy of ±1cm; an image processing unit with a built-in convolutional neural network accelerator; and a gesture recognition algorithm module that supports click, swipe, zoom, and custom gesture recognition. During use, the infrared camera and depth sensor work together to capture the spatial position and motion trajectory of the user's gestures. The image processing unit processes this data in real time, and the gesture recognition algorithm module converts the gestures into operation commands, enabling contactless human-computer interaction. Through natural and intuitive gesture interaction, staff can complete operations without touching the equipment in scenarios such as park inspections and equipment debugging. 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 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. In use, the beamforming technology of the microphone array 3 focuses on the target sound source, the voice preprocessing unit eliminates environmental noise and echo interference, the voice recognition engine converts speech into text, and the natural language understanding module interprets the user's intent and generates a response, achieving a natural and fluent voice interaction experience. It achieves high-accuracy voice interaction even in noisy park environments.

[0032] Please see Figure 4In this embodiment, the LED display module includes a main display screen 1, status indicator lights 10, a touch control panel, and a display driver circuit. The main display screen 1 has a resolution of 3840×2160. The status indicator lights 10 use RGB LEDs. The touch control panel supports multi-touch. The display driver circuit supports the HDR10 standard. During use, the main display screen 1 displays a 3D map of the park and a heat map of the distribution of intelligent agents. The status indicator lights 10 reflect the system's health status in real time. The touch control panel supports gesture operation. The display driver circuit ensures high-quality visual presentation, realizing a multi-dimensional visualization of the park's operating status. Managers can intuitively understand the working status of each intelligent agent 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 consists of 6 sets arranged in a ring, with a speed of 800~2000 RPM. The ventilation base cover 5 has an opening rate of 60%~70% and a hole diameter of 3mm~5mm. The heat pipe radiator 11 connects the CPU and GPU chips. The temperature sensor is distributed inside the device. During use, the temperature sensor monitors the device in real time. Temperature control and intelligent adjustment of exhaust fan speed; heat pipe radiator 11 conducts core heat to heat sink; ventilation base cover 5 and top cover exhaust fan 4 form a vertical airflow channel for efficient heat dissipation, ensuring the stability of the equipment under high load. The intelligent terminal 6 includes a touchscreen, physical buttons, a network interface, and a wireless module; the touchscreen resolution is 2560×1600; physical buttons include a power button and function buttons; the network interface supports wired network connection; the wireless module supports Bluetooth 5.0 and Wi-Fi connection; during use, user input is received through the touchscreen and physical buttons, and data interaction with the core system is achieved through the network interface and wireless module. The terminal has a built-in lightweight AI model for rapid localized processing. The tower server 7 includes a motherboard, memory, storage array, and GPU accelerator card; the motherboard supports dual CPUs; memory capacity is 128GB~256GB, frequency 3200MHz; the storage array uses NVMe. SSDs with total capacities ranging from 8TB to 16TB; GPU accelerator cards for AI model inference; during use, dual-CPU systems provide general-purpose computing power, large-capacity high-speed memory supports multi-task parallel processing, NVMe storage arrays ensure fast data access, and GPU accelerator cards are dedicated to AI computing tasks, achieving optimized allocation of computing resources. The base module includes casters 8, feet 9, anti-collision strips 12, and aviation connectors 13; multiple sets of casters 8, 75mm in diameter, with brakes; multiple sets of feet 9, with an adjustable height range of 30mm to 50mm; anti-collision strips 12, 5mm thick, made of rubber; and aviation connectors 13 for power and data connections. During use, the casters 8 allow for easy movement of the device, and the feet 9 provide stable support when the device reaches the designated position. The anti-collision strips 12 protect the edges of the device, and the aviation connectors 13 ensure stable and reliable power and data connections.

[0033] During operation, the operator first moves the digital router to the designated location in the park control center using the casters 8 of the base module, adjusts the adjustable feet 9 to ensure stable placement of the equipment, and the anti-collision strips 12 effectively protect the edges of the equipment from collision damage. The power cord and the park's main network fiber optic cable are connected via the aviation plug 13. After the equipment is started, 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, and the temperature sensor monitors in real time and intelligently adjusts the fan speed to ensure stable operation of the equipment in an ambient temperature of 25~35℃. 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 fiber optic interface of the data transmission module simultaneously establish communication links with various devices in the park to form a hybrid transmission network. Users can operate the intelligent terminal 6 via its touchscreen 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 and converts it into an operation command through a convolutional neural network accelerator; at the same time, the microphone array 3 of the AI ​​voice dialogue interaction module focuses the user's voice through beamforming technology, and after noise reduction and echo cancellation processing, the voice recognition engine converts the voice into text, and the natural language understanding module interprets the user's intent based on the Transformer architecture; the core processing module distributes the collected multimodal data to the main processor chip to process general computing tasks through high-speed cache and data bus, the coprocessor to execute AI model inference, and the GPU accelerator card of the tower server 7 to process deep learning tasks; the NVMe SSD array of the storage module achieves fast data access with a capacity of 8TB~16TB and a frequency of 3200MHz; The core processing module acts as a central scheduler, coordinating various intelligent agents through a three-tiered processing architecture: the access layer collects visual, voice, and IoT device data via a data bus; the coordination layer runs intelligent agent coordination algorithms and dynamically allocates computing resources; the decision-making layer executes cross-intelligent agent collaborative logic; the main display screen 1 of the LED display module displays a 4K resolution 3D map of the park and a heat map of intelligent agent distribution, while the status indicator 10 reflects the system's health status in real time via RGB LEDs; finally, the collaborative decision-making results are pushed to personnel at each post through the intelligent agent terminal 6, and the data transmission module issues control commands to the execution devices, forming a complete closed loop of "data acquisition-processing-coordination-execution".

[0034] Through the above steps, the core processing module coordinates the work of each module, constructs a dynamic allocation mechanism for digital capabilities similar to a network router, realizes data interaction and collaborative computing among multiple AI agents, enables efficient collaborative work among multiple AI agents in the park, breaks down the traditional information silos, and allows data from various systems in the park to interact and share in real time, forming a unified intelligent decision-making system.

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

Claims

1. A router based on real-data fusion of multiple AI agents, characterized in that: include: The core processing module adopts a multi-core parallel processor architecture and has a built-in high-speed cache for processing data requests and collaborative computing from multiple AI agents; The data transmission module includes a high-speed Ethernet interface and a wireless communication module, used for data transmission of devices within the park; The storage module, including flash memory and random access memory, is used to store the operating system, configuration files, and AI model data; The visual gesture recognition module includes an infrared camera array (2) and a depth sensor for capturing user gestures and converting them into operation commands; The AI ​​voice dialogue interaction module includes a microphone array (3) and a voice recognition engine for voice command recognition and natural language interaction; The LED display module uses a high-resolution display screen to show the operating status and data visualization information of the intelligent agent; The heat dissipation module includes a top cover exhaust fan (4) and a ventilation base cover (5) to form an air convection channel; The intelligent agent terminal (6) is connected via wired or wireless means and is used 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 casters (8) and adjustable feet (9) for moving and securing the equipment.

2. A router based on real-data fusion of multiple AI agents according to claim 1, characterized in that: 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 to 4.5GHz; the coprocessor is dedicated to AI model inference calculation; the cache capacity is 32MB to 64MB; and the bandwidth of the data bus is 256bit.

3. A router based on real-data fusion of multiple AI agents according to claim 1, characterized in that: 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 auto-negotiation. The Wi-Fi module supports dual-band 2.4GHz and 5GHz; the fiber optic interface supports 10Gbps data transmission; and the internal data exchange matrix adopts a Crossbar architecture.

4. A router based on real-data fusion of multiple AI agents 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 distance 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, swipe, zoom, and custom gesture recognition.

5. A router based on real-data fusion of multiple AI agents 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 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; and the natural language understanding module is based on the Transformer architecture.

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

7. A router based on real-data fusion of multiple AI agents 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 fan (4) consists of 6 sets arranged in a ring, with a rotation speed of 800~2000RPM; the ventilation base cover (5) has an opening rate of 60%~70% and a hole diameter of 3mm~5mm; the heat pipe radiator (11) connects the CPU and GPU chips; the temperature sensor is distributed inside the device.

8. A router based on real-data fusion of multiple AI agents 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; the wireless module supports Bluetooth 5.0 and Wi-Fi connection.

9. A router based on real-data fusion of multiple AI agents 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 CPUs; the memory capacity is 128GB~256GB and the frequency is 3200MHz; the storage array uses NVMe SSD and the total capacity is 8TB~16TB; the GPU accelerator card is used for AI model inference.

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

Images