Hardware architecture of learning machine

Abstract

The utility model discloses a hardware architecture of a learning machine. The hardware architecture comprises a multi-core processor, a memory, a data storage module, a display, an input device, a context processing unit, a data retrieval unit, a strategy adjustment unit and a loudspeaker, the context processing unit is connected to the multi-core processor, is integrated with an inertial measurement unit, an ambient light sensor and a microphone array, and is used for collecting environment and state information of a user; the data retrieval unit is connected to the multi-core processor and comprises an SSD (Solid State Disk) and a hardware accelerator; the strategy adjusting unit is connected to the multi-core processor and comprises a programmable logic controller or a microcontroller.

Description

Technical Field

[0001] The utility model relates to a hardware structure of a multifunctional learning machine, belonging to the technical field of electronic equipment. Background Art

[0002] Existing learning machines mainly rely on preset learning content and simple human-computer interaction functions. Although they have improved students' learning efficiency to a certain extent, they are obviously insufficient in processing long conversations and complex interactions. In particular, when memorizing and retracing the content of long conversations, they are prone to information forgetting, low storage efficiency, and insufficient processing performance. In order to solve these problems, some people have proposed improvements in algorithms, but few people have focused on hardware improvements. Utility Model Content

[0003] To this end, the utility model provides a hardware foundation for the learning machine to realize the anti-forgetting function of long conversation by introducing high-speed RAM and large-capacity SSD storage, multi-core high-performance processors and hardware accelerators (such as NVIDIA Jetson modules), and context processing units integrated with multiple sensors such as accelerometers and gyroscopes, so as to help make up for the defects and shortcomings of intelligent learning machines in long conversations.

[0004] A hardware architecture of a learning machine includes a multi-core processor, a memory, a data storage module, a display, an input device, a context processing unit, a data retrieval unit, a policy adjustment unit and a speaker; the context processing unit is connected to the multi-core processor, and is integrated with an inertial measurement unit, an ambient light sensor and a microphone array for collecting user environment and status information; the data retrieval unit is connected to the multi-core processor, and includes an SSD hard disk and a hardware accelerator; the policy adjustment unit is connected to the multi-core processor, and includes a programmable logic controller or a microcontroller.

[0005] Furthermore, the multi-core processor is an ARM Cortex-A series processor or an Intel x86 processor.

[0006] Furthermore, the memory is connected to the data retrieval unit and includes RAM and flash memory.

[0007] Furthermore, the RAM adopts DDR4, and the flash memory adopts eMMC or NVMe SSD.

[0008] Furthermore, the data storage module is connected to the data retrieval unit and includes a solid state drive or a mechanical hard drive.

[0009] Furthermore, the capacity of the solid state drive or the mechanical hard drive is greater than 1 TB.

[0010] Furthermore, the input device is connected to the multi-core processor and includes a microphone and a camera.

[0011] Furthermore, the SSD hard disk of the data retrieval unit is an SSD hard disk equipped with an NVMe interface, and the hardware accelerator is an NVIDIA Jetson module.

[0012] Furthermore, the inertial measurement unit includes an integrated accelerometer and a gyroscope.

[0013] The beneficial effects of the present invention are embodied in: by optimizing the hardware design and integrating multiple high-performance hardware modules, the hardware architecture of the learning machine can be optimized and the hardware performance can be improved, so that the learning machine can be significantly improved in terms of dialogue memory and backtracking functions on the basis of the hardware architecture provided by the present invention, thereby improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic diagram of the hardware architecture of the learning machine of the embodiment of the utility model. DETAILED DESCRIPTION

[0015] The utility model is further described below in conjunction with the accompanying drawings and specific implementation methods and examples. The purpose of providing the examples is only to illustrate, and not to impose any limitation.

[0016] The present invention provides a hardware architecture of a learning machine. Figure 1, the hardware architecture includes a CPU, a data retrieval unit, a policy adjustment unit, a context processing unit, a speaker, a camera, a display, a RAM and a ROM. Among them, the CPU adopts a high-performance multi-core processor (such as an ARM Cortex-A series or an Intel x86 processor) so that the learning machine using the hardware architecture of this embodiment can efficiently process conversation data and perform learning tasks. The data retrieval unit is connected to the CPU, which is equipped with an SSD hard disk with an NVMe interface and a hardware accelerator (such as an NVIDIA Jetson module) so that the learning machine using the hardware architecture of this embodiment can quickly retrieve and access stored data, helping it to improve the data processing efficiency in long conversations. The context processing unit integrates an inertial measurement unit, a light sensor and a microphone array, wherein the light sensor is used to collect the ambient light information of the user, the microphone array is used to pick up the user's voice, and the inertial measurement unit is used to detect the movement and tilt of the learning machine; this enables the context processing unit to obtain the user's environment and state information, and then combined with the existing circuit for processing the user's environment and state information, to help the learning machine using the hardware architecture of this embodiment improve its ability to understand the conversation context, as well as its memory and backtracking ability in long conversations. The strategy adjustment unit uses a programmable logic controller (such as Siemens S7-1200) or a microcontroller (such as Arduino or Raspberry Pi), which mainly provides hardware support for the learning machine to realize the function of "dynamically adjusting the learning strategy and dialogue content according to the user's learning progress and feedback".

[0017] RAM is used as the memory in this hardware architecture, and high-speed RAM such as DDR4 and large-capacity flash memory (such as eMMC or NVMe SSD) are preferably used, so that the hardware architecture of this embodiment can be used in a learning machine to quickly access and save learning content and conversation data, thereby helping to improve storage efficiency and data reading speed.

[0018] As the data storage module in this hardware architecture, ROM preferably uses a large-capacity solid-state drive (such as a 1TB NVMeSSD) or a high-performance mechanical hard drive (such as a 2TB HDD) for long-term storage of conversation history records and learning data for quick retrieval.

[0019] In some preferred embodiments, the display uses a high-resolution touch screen display (such as 1080p or higher resolution), which is connected to the CPU and can be used to display learning content and dialogue interfaces to provide a high-quality visual experience.

[0020] Microphones and cameras are used as input devices, preferably high-sensitivity MEMS microphones (such as the Knowles series) and high-definition cameras (such as Logitech C920) can be used to receive user input and capture user video and audio data, thereby providing a reliable data basis for high-quality human-computer interaction. As a device for outputting audio, speakers are equipped with high-quality speakers (such as Bose or JBL modules) to output clear audio, ensuring that users can obtain high-quality sound feedback and interactive experience during use.

[0021] In the context processing unit, the inertial measurement unit can preferably use the BMI160 inertial measurement unit launched by Bosch Sensortec, which integrates a 16-bit 3-axis gravity accelerometer and an ultra-low power 3-axis gyroscope in a single package, using a 14-pin LGA package with a size of 2.5×3.0×0.8mm3. The light sensor is used to detect the light intensity around the device so that the learning machine can adjust the screen brightness accordingly to provide a better visual experience. The TAOS TSL2561 ambient light sensor can be used.

[0022] In summary, the hardware architecture provided by the embodiment of the utility model, through optimization and improvement in hardware configuration and coordination, enables the learning machine based on the hardware architecture to improve the anti-forgetting performance of long conversations and improve the user experience without considering algorithm improvements.

[0023] Example 1

[0024] A hardware architecture of a learning machine includes the following hardware components:

[0025] 1. Processor:

[0026] -Adopts ARM Cortex-A72 quad-core processor with a main frequency of 1.8GHz and supports multi-tasking, ensuring the efficient performance of the device when handling long conversations and complex calculations.

[0027] 2. Memory:

[0028] -Equipped with 8GB DDR4 memory for fast access and processing of learning content and conversation data, improving the system's response speed.

[0029] - 128GB NVMe SSD storage is used for efficient storage and fast access to large amounts of conversation history and learning data.

[0030] 3. Display:

[0031] -Equipped with a 10.1-inch 1080p full HD touch screen display, providing clear and delicate display effects, ensuring users a good visual experience during use.

[0032] 4. Input devices:

[0033] -High-sensitivity MEMS microphone (Knowles SPH0645LM4H) with high signal-to-noise ratio and wide frequency response, capable of clearly picking up user voice.

[0034] - HD webcam (Logitech C920) with 1080p video recording support, ensuring high-quality video capture and image processing.

[0035] 5. Data storage module:

[0036] -Adopts 1TB NVMe SSD to ensure long-term storage and fast access of large-capacity data, supporting the storage needs of large amounts of conversation history and learning data.

[0037] 6.Context processing unit:

[0038] -Integrated Bosch BMI160 accelerometer and gyroscope to detect device movement and tilt, helping to understand user operating habits and posture changes.

[0039] -Equipped with TAOS TSL2561 ambient light sensor, it detects the surrounding light intensity and automatically adjusts the screen brightness to provide a comfortable visual experience.

[0040] -Uses multiple MEMS microphone arrays to achieve directional voice pickup and background noise elimination to ensure the accuracy of voice recognition.

[0041] 7.Data Retrieval Unit:

[0042] -1TB SSD hard drive and NVIDIA Jetson Xavier NX module equipped with NVMe interface for fast retrieval and access to stored data, significantly improving the device's data processing efficiency during long conversations.

[0043] 8. Strategy Adjustment Unit:

[0044] - Using Siemens S7-1200 programmable logic controller, the learning strategy and conversation content are adjusted in real time according to the user's learning progress and feedback to ensure the personalization and optimization of the learning process.

[0045] 9. Speaker:

[0046] -Equipped with Bose SoundLink speakers, it provides clear and rich audio output, ensuring high-quality sound feedback and interactive experience.

[0047] Example 2

[0048] A hardware architecture of a learning machine adopts the following hardware configuration:

[0049] 1. Processor:

[0050] -Uses Intel i5-8265U quad-core processor, with a main frequency of 1.6GHz and support for turbo acceleration up to 3.9GHz, ensuring the device's efficient performance when processing large amounts of conversation data and complex calculations.

[0051] 2. Memory:

[0052] -Equipped with 16GB DDR4 memory for fast processing and storage of learning content and conversation data, significantly improving system response speed.

[0053] - 256GB NVMe SSD storage is used for efficient storage and fast access to large amounts of conversation history and learning data.

[0054] 3. Display:

[0055] -Adopts a 13.3-inch 4K ultra-high-definition touch screen display to provide excellent display effects and ensure the best visual experience for users during use.

[0056] 4. Input devices:

[0057] -High-sensitivity MEMS microphone (Knowles SPH0645LM4H) with high signal-to-noise ratio and wide frequency response, capable of clearly picking up user voice.

[0058] - HD camera (Logitech Brio) with support for 4K video recording, ensuring high-quality video capture and image processing.

[0059] 5. Data storage module:

[0060] -Uses 2TB HDD to ensure long-term storage and fast access to large-capacity data, supporting the storage needs of large amounts of conversation history and learning data.

[0061] 6.Context processing unit:

[0062] -Integrated Bosch BMA456 accelerometer and gyroscope to detect device movement and tilt, helping to understand user operating habits and posture changes.

[0063] -Equipped with TAOS TSL2561 ambient light sensor, it detects the surrounding light intensity and automatically adjusts the screen brightness to provide a comfortable visual experience.

[0064] -Uses multiple MEMS microphone arrays to achieve directional voice pickup and background noise elimination to ensure the accuracy of voice recognition.

[0065] 7.Data Retrieval Unit:

[0066] -256GB SSD hard drive and NVIDIA Jetson AGX Xavier module equipped with NVMe interface for rapid retrieval and access to stored data, significantly improving the device's data processing efficiency during long conversations.

[0067] 8. Strategy Adjustment Unit:

[0068] - Using Arduino Mega 2560 microcontroller, the learning strategy and conversation content are adjusted in real time according to the user's learning progress and feedback to ensure the personalization and optimization of the learning process.

[0069] 9. Speaker:

[0070] -Equipped with JBL Charge 4 speakers, it provides clear and full audio output, ensuring high-quality sound feedback and interactive experience.

[0071] The above contents are further detailed descriptions of the present invention in combination with specific preferred implementations, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For technicians in the technical field to which the present invention belongs, several equivalent substitutions or obvious variations can be made without departing from the concept of the present invention, and the performance or use is the same, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. A hardware architecture of a learning machine, characterized in that: It includes a multi-core processor, a memory, a data storage module, a display, an input device, a context processing unit, a data retrieval unit, a policy adjustment unit and a speaker; the context processing unit is connected to the multi-core processor, which integrates an inertial measurement unit, an ambient light sensor and a microphone array for collecting user environment and status information; the data retrieval unit is connected to the multi-core processor, which includes an SSD hard disk and a hardware accelerator; the policy adjustment unit is connected to the multi-core processor, which includes a programmable logic controller or a microcontroller.

2. The hardware architecture according to claim 1, characterized in that: The multi-core processor is an ARM Cortex-A series processor or an Intel x86 processor.

3. The hardware architecture according to claim 1, wherein: The memory is connected to the data retrieval unit and includes a RAM and a flash memory.

4. The hardware architecture according to claim 3, characterized in that: The RAM adopts DDR4, and the flash memory adopts eMMC or NVMe SSD.

5. The hardware architecture according to claim 1, wherein: The data storage module is connected to the data retrieval unit and includes a solid state hard disk or a mechanical hard disk.

6. The hardware architecture according to claim 5, characterized in that: The capacity of the solid state drive or the mechanical hard drive is greater than 1 TB.

7. The hardware architecture of claim 1, wherein: The input device is connected to the multi-core processor and includes a microphone and a camera.

8. The hardware architecture of claim 1, wherein: The SSD hard disk of the data retrieval unit is an SSD hard disk equipped with an NVMe interface, and the hardware accelerator is an NVIDIA Jetson module.

9. The hardware architecture of claim 1, wherein: The inertial measurement unit includes an integrated accelerometer and gyroscope.

Images