AI edge computing controller

Abstract

The utility model belongs to the technical field of edge computing and artificial intelligence, and discloses an AI edge computing controller which comprises a controller body, a main chip module is arranged in the controller body, the main chip module adopts an NVIDIA Jetson chip, an all-aluminum heat dissipation backboard is arranged on the outer wall of the controller body, a communication interface module is arranged on the outer wall of the controller body, and the communication interface module is arranged on the outer wall of the controller body. The communication interface module comprises a data transmission interface, a clock synchronization module and a trigger signal interface, a wireless communication module is further arranged in the controller body, and a detachable antenna is connected to the outside of the wireless communication module. According to the utility model, the problems of low real-time reasoning efficiency, insufficient multi-sensor data synchronization precision, poor hardware environment adaptability and the like of an AI model in an edge scene are solved, AI computing power up to 275TOPS can be provided, multiple industrial protocols and high-precision clock synchronization can be supported, and timestamp alignment and accurate triggering of multi-sensor data are ensured.

Description

Technical Field

[0001] This utility model belongs to the field of edge computing and artificial intelligence technology, specifically relating to an edge computing controller hardware architecture integrating AI algorithms and its real-time data processing method. Background Technology

[0002] With the rapid development of artificial intelligence technology, AI applications have gradually expanded from the cloud to the edge. As an important component of the Internet of Things (IoT) architecture, edge computing enables data processing and analysis closer to the data source, thereby reducing data transmission latency, improving system response speed, and protecting data privacy.

[0003] However, edge computing devices still face many challenges in practical applications:

[0004] 1) Low AI model inference efficiency: Traditional edge devices have limited computing power, making it difficult to meet the real-time inference needs of complex AI models;

[0005] 2) Insufficient synchronization accuracy of multi-sensor data: In application scenarios involving multi-sensor fusion, the lack of a unified clock reference leads to inconsistent data timestamps, affecting the accuracy of data fusion;

[0006] 3) Poor adaptability to hardware environment: The industrial environment is complex and changeable, and factors such as temperature, humidity and electromagnetic interference place higher demands on the stability of equipment;

[0007] 4) It is difficult to balance power consumption and performance: High-performance computing is usually accompanied by high power consumption, while edge devices often need to operate under limited power consumption conditions.

[0008] Existing edge computing devices struggle to address the aforementioned issues simultaneously, limiting the widespread application of AI technology in edge scenarios. Therefore, there is an urgent need for an edge computing controller capable of efficiently executing AI algorithms in edge environments while simultaneously ensuring multi-sensor data synchronization, environmental adaptability, and power consumption control. Utility Model Content

[0009] The purpose of this invention is to provide an AI edge computing controller to solve the problems mentioned in the background art.

[0010] To achieve the above objectives, this utility model provides the following technical solution: an AI edge computing controller, comprising a controller body, a main chip module disposed within the controller body, the main chip module employing an NVIDIA Jetson chip, an all-aluminum heat dissipation backplate disposed on the outer wall of the controller body, a communication interface module disposed on the outer wall of the controller body, the communication interface module including a data transmission interface, a clock synchronization module and a trigger signal interface, and a wireless communication module disposed within the controller body, the wireless communication module being externally connected to a detachable antenna.

[0011] Preferably, the main chip module includes an Ampere architecture GPU, a 12-core Arm A78 CPU, a deep learning accelerator (DLA), and a programmable vision accelerator (PVA), and the main chip module has a 15W-60W adaptive power consumption control system.

[0012] Preferably, the data transmission interface supports a maximum single-channel transmission rate of 6Gbps and employs differential signal transmission and a physical anti-interference design using coaxial cable / shielded twisted pair.

[0013] Preferably, the clock synchronization module adopts a high-precision clock synchronization protocol to provide a unified clock reference for cameras, lidar, and IMU devices.

[0014] Preferably, the trigger signal interface supports synchronous trigger signals for EtherCAT and CAN bus industrial protocols.

[0015] Preferably, the controller body adopts an all-metal structure and industrial-grade electronic components, supporting a wide operating temperature range of -20℃ to 60℃.

[0016] Preferably, the detachable antenna includes a WiFi antenna and a 4G / 5G antenna.

[0017] Preferably, the communication interface module is capable of simultaneously transmitting 4K / 60fps video, audio, and control signals.

[0018] Compared with the prior art, the beneficial effects of this utility model are:

[0019] 1. The AI ​​edge computing controller proposed in this utility model adopts a high-performance NVIDIA Jetson chip, which integrates an Ampere architecture GPU, a 12-core Arm A78 CPU, a deep learning accelerator and a programmable vision accelerator, providing up to 275 TOPS of AI computing power, effectively solving the problem of low real-time inference efficiency of AI models in edge scenarios; at the same time, through the 15W-60W adaptive power consumption design, the relationship between performance and power consumption is balanced.

[0020] 2. The AI ​​edge computing controller of this utility model is designed with a high-precision clock synchronization module and trigger signal interface, which provides a unified clock reference for multiple sensors, supports synchronous trigger signals of industrial protocols, solves the problem of insufficient data synchronization accuracy of multiple sensors, and significantly improves the accuracy of data fusion.

[0021] 3. The AI ​​edge computing controller of this utility model adopts an all-aluminum heat dissipation backplate to achieve a fanless heat dissipation design. Combined with an all-metal body structure and industrial-grade electronic components, it supports a wide temperature range of -20℃ to 60℃, which solves the problem of poor hardware environmental adaptability and significantly improves the stability and reliability of the equipment in complex industrial environments. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the appearance structure of the AI ​​edge computing controller of this utility model from one perspective.

[0023] Figure 2 This is a schematic diagram of the appearance structure of the AI ​​edge computing controller of this utility model from another perspective;

[0024] Figure 3 This is a system architecture block diagram of the AI ​​edge computing controller of this utility model;

[0025] Figure 4 This is a flowchart of the multi-sensor data synchronization process for the AI ​​edge computing controller of this utility model.

[0026] In the diagram: 1. Controller body; 2. Main chip module; 3. All-aluminum heat dissipation backplate; 4. Communication interface module; 5. Data transmission interface; 6. Clock synchronization module; 7. Trigger signal interface; 8. Wireless communication module; 9. Detachable antenna; 21. Ampere architecture GPU; 22. 12-core Arm A78 CPU; 23. Deep learning accelerator (DLA); 24. Programmable vision accelerator (PVA); 91. WiFi antenna; 92. 4G / 5G antenna. Detailed Implementation

[0027] The technical solutions of the present utility model will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present utility model, and not all embodiments. Based on the embodiments of the present utility model, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of the present utility model.

[0028] Example 1

[0029] Please see Figure 1This utility model provides a technical solution: an AI edge computing controller, including a controller body 1, a main chip module 2 is provided inside the controller body 1, the main chip module 2 adopts an NVIDIA Jetson chip, an all-aluminum heat dissipation backplate 3 is provided on the outer wall of the controller body 1, a communication interface module 4 is provided on the outer wall of the controller body 1, the communication interface module 4 includes a data transmission interface 5, a clock synchronization module 6 and a trigger signal interface 7, a wireless communication module 8 is also provided inside the controller body 1, and a detachable antenna 9 is externally connected to the wireless communication module 8.

[0030] Main chip module 2 includes an Ampere architecture GPU21, a 12-core Arm A78 CPU22, a deep learning accelerator DLA23, and a programmable vision accelerator PVA24. Main chip module 2 features a 15W-60W adaptive power control system. The NVIDIA Jetson chip can provide up to 275 TOPS of AI computing power, effectively addressing the issue of low real-time inference efficiency for AI models in edge computing scenarios. The adaptive power control system dynamically adjusts power consumption based on the actual computing load, effectively controlling energy consumption while ensuring performance.

[0031] The all-aluminum heat sink backplate 3 design enables fanless heat dissipation, avoiding problems such as noise, dust accumulation, and mechanical failure that may be caused by fans, thus improving the reliability and stability of the system. At the same time, the all-aluminum heat sink backplate 3, together with the all-metal structure of the controller body 1, forms a highly efficient heat dissipation system, which can maintain the normal operating temperature of the chip even under high load conditions.

[0032] Example 2

[0033] Please see Figure 1-2 Based on Embodiment 1, the data transmission interface 5 of the communication interface module 4 supports a maximum single-channel transmission rate of 6Gbps, employing differential signal transmission and a physical anti-interference design using coaxial cable / shielded twisted pair. The data transmission interface 5 can simultaneously transmit 4K / 60fps video, audio, and control signals, meeting the real-time data requirements of multiple cameras and LiDAR.

[0034] The clock synchronization module 6 adopts a high-precision clock synchronization protocol to provide a unified clock reference for cameras, lidar, and IMU devices, ensuring the timestamp alignment of multi-sensor data and solving data fusion errors caused by clock drift.

[0035] Trigger signal interface 7 supports synchronous trigger signals for EtherCAT and CAN bus industrial protocols, providing nanosecond-level trigger pulses to ensure synchronized movements between the robotic arm and the vision system. These three modules collectively address the issue of insufficient synchronization accuracy of multi-sensor data, significantly improving the system's collaborative working capabilities in complex environments.

[0036] The controller body 1 adopts an all-metal structure and industrial-grade electronic components, supporting a wide operating temperature range of -20℃ to 60℃, thus solving the problem of poor hardware environmental adaptability. This design enables the AI ​​edge computing controller to operate stably in various harsh industrial environments, greatly expanding its application scenarios.

[0037] Example 3

[0038] Please see Figure 1-3 Based on Embodiment 2, the detachable antenna 9 externally connected to the wireless communication module 8 includes a WiFi antenna 91 and a 4G / 5G antenna 92. Through these wireless communication interfaces, the AI ​​edge computing controller can achieve seamless connectivity between local computing and cloud data exchange.

[0039] In practical applications, the AI ​​edge computing controller first performs local data processing and AI model inference through the main chip module 2, transmitting only necessary information to the cloud via the wireless communication module 8. This significantly reduces redundant information transmission and improves overall system efficiency. Meanwhile, the detachable antenna 9 design solves the installation difficulties that may arise from wiring in complex environments.

[0040] Figure 3 The process of multi-sensor data synchronization is demonstrated. Clock synchronization module 6 first provides a unified clock reference for all connected sensor devices, and then trigger signal interface 7 sends precise trigger signals to ensure that all sensors collect data at the same time. Data transmission interface 5 receives data from multiple sensors and transmits it to main chip module 2 for processing. Main chip module 2, through its powerful computing capabilities, performs real-time fusion analysis of the multi-source data, ultimately outputting high-quality decision results.

[0041] The working principle and usage process of this utility model are as follows: After the AI ​​edge computing controller is powered on, the main chip module 2 begins system initialization, and the clock synchronization module 6 establishes a unified clock reference to provide synchronization signals for connected sensor devices. Users can connect various sensor devices, such as high-definition cameras, LiDAR, and IMUs, through the data transmission interface 5. The trigger signal interface 7 ensures that the data acquisition actions of all devices are synchronized. The Ampere architecture GPU 21, 12-core Arm A78 CPU 22, deep learning accelerator DLA 23, and programmable vision accelerator PVA 24 within the main chip module 2 work together to process and perform AI inference on the acquired multi-source data in real time. The all-aluminum heat dissipation backplate 3 ensures good heat dissipation throughout the entire operation, maintaining stable system operation even under high load. The wireless communication module 8 maintains communication with cloud services through the detachable antenna 9, realizing the collaboration between local computing and cloud services. The adaptive power consumption control system dynamically adjusts power consumption according to the actual computing load, minimizing energy consumption while ensuring performance.

[0042] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the present invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An AI edge computing controller, comprising a controller body (1), characterized in that: The controller body (1) is equipped with a main chip module (2), which uses an NVIDIA Jetson chip. The outer wall of the controller body (1) is equipped with an all-aluminum heat dissipation backplate (3). The outer wall of the controller body (1) is equipped with a communication interface module (4), which includes a data transmission interface (5), a clock synchronization module (6), and a trigger signal interface (7). The controller body (1) is also equipped with a wireless communication module (8), which is externally connected to a detachable antenna (9).

2. The AI ​​edge computing controller according to claim 1, characterized in that: The main chip module (2) includes an Ampere architecture GPU (21), a 12-core Arm A78 CPU (22), a deep learning accelerator DLA (23), and a programmable vision accelerator PVA (24). The main chip module (2) has a 15W-60W adaptive power consumption control system.

3. The AI ​​edge computing controller according to claim 1, characterized in that: The data transmission interface (5) supports a maximum single-channel transmission rate of 6Gbps and adopts differential signal transmission and physical anti-interference design with coaxial cable / shielded twisted pair.

4. The AI ​​edge computing controller according to claim 1, characterized in that: The clock synchronization module (6) adopts a high-precision clock synchronization protocol to provide a unified clock reference for cameras, lidar, and IMU devices.

5. The AI ​​edge computing controller according to claim 1, characterized in that: The trigger signal interface (7) supports synchronous trigger signals for EtherCAT and CAN bus industrial protocols.

6. The AI ​​edge computing controller according to claim 1, characterized in that: The controller body (1) adopts an all-metal structure and industrial-grade electronic components, supporting a wide temperature range of -20℃ to 60℃.

7. The AI ​​edge computing controller according to claim 1, characterized in that: The detachable antenna (9) includes a WiFi antenna (91) and a 4G / 5G antenna (92).

8. The AI ​​edge computing controller according to claim 1, characterized in that: The communication interface module (4) can simultaneously transmit 4K / 60fps video, audio and control signals.

Images