AI calculation device of vehicle intelligent cabin, initialization method of AI calculation device, coordination control method of AI calculation device and computer program product

Abstract

The invention discloses an AI computing device of a vehicle intelligent cabin and an initialization method, a coordination control method and a computer program product thereof, the AI computing device is connected with an intelligent cabin host and works cooperatively with the intelligent cabin host, and the AI computing device comprises an AI chip, a connection interface module, a power management module and a data storage module; the connection interface module is used for being connected with an intelligent cabin host, receiving an AI calculation task allocated by the intelligent cabin host and feeding back an AI calculation result; the data storage module is used for storing the AI calculation task, the calculation result and the trained end side large model; the AI chip adopts a storage and calculation integrated framework and further comprises a calculation unit for executing the AI calculation task, a storage unit for calculation and storage and a control unit for regulating and controlling the AI calculation task. According to the method, under the condition that original intelligent cabin host hardware is not modified, AI computing power expansion is brought to the host, and the getting-on requirement of the AI function of a mass-produced and low-computing-power host chip is met.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle electronic appliances, and particularly relates to an AI computing device for a vehicle intelligent cockpit, an initialization method, a coordination control method, and a computer program product thereof. Background Art

[0002] With the rapid development of Artificial Intelligence (AI) technology, the AI large model technology has become a key development trend in the field of intelligent vehicles. The application of AI large models in intelligent vehicles has significantly promoted the improvement of the human-vehicle interaction experience. Its core advantages are reflected in two important dimensions: one is in the aspect of in-vehicle voice assistants. Through the deep optimization of semantic understanding and the refined processing of corpus generation, the AI large model can accurately analyze the user's intentions. Even in the face of complex sentence structures and diverse contexts, it can generate natural, logical, and anthropomorphic responses, effectively meeting the user's emotional interaction needs; the other is in the multi-modal interaction dimension. With the multi-modal characteristics of the large model, it can comprehensively process various types of data such as voice, vision, and touch, enabling the intelligent cockpit system to fully perceive the user's needs and then provide highly professional intelligent services.

[0003] According to the differences in deployment methods, AI large models can be divided into edge large models, cloud large models, and edge-cloud fusion large models. Among them, edge large models are highly favored by automobile manufacturing enterprises due to their significant characteristics such as low latency, low cost, support for operation in a network-free environment, strong data privacy protection, and relatively simple technical architectures. Currently, there are mainly the following two hardware solutions for implementing edge large models:

[0004] One is to upgrade the intelligent cockpit host SOC chip, using a chip with higher CPU / NPU computing power, and at the same time using the CPU to assist the NPU to meet the computing power requirements of the AI large model for the edge side. However, upgrading the intelligent cockpit host SOC chip requires re-designing the intelligent cockpit hardware, which has the characteristics of a long development and verification cycle, high development cost, and high difficulty. Moreover, the available SOC chips in the market are limited and cannot meet the development needs of AI.

[0005] The other is to install a dedicated AI chip on the intelligent cockpit host. Through the von Neumann architecture, it meets the computing power requirements of the AI large model for the edge side. Similarly, installing a dedicated AI chip on the intelligent cockpit host requires re-designing the intelligent cockpit hardware, which has the characteristics of a long development and verification cycle, high development cost, and high difficulty. Due to the limited bandwidth of the data transmission bus, the data transmission speed is restricted, and a large amount of data accumulates at the bus, resulting in low computing power utilization and waste; at the same time, more than 90% of the energy consumption is wasted in data transfer, with high overall power consumption and large heat generation. Summary of the Invention

[0006] The technical problem to be solved by the embodiments of the present invention is to provide an AI computing device for a vehicle intelligent cockpit, its initialization method, coordination control method, and computer program product, so as to expand the AI computing power at a low cost and low price, and meet the on-vehicle requirements of the AI function of a low-computing-power host chip.

[0007] To solve the above technical problem, the present invention provides an AI computing device for a vehicle intelligent cockpit. The AI computing device is connected to and works in cooperation with the intelligent cockpit host, and includes:

[0008] An AI chip, a connection interface module, a power management module, and a data storage module;

[0009] The connection interface module is used to connect to the intelligent cockpit host, receive the AI computing tasks allocated by the intelligent cockpit host, and feedback the AI computing results;

[0010] The data storage module is used to store the AI computing tasks and computing results, as well as the end-side large model that has been trained;

[0011] The AI chip adopts a memory-computation integrated architecture, and further includes: a computing unit for executing the AI computing tasks, a storage unit for operation and storage, and a control unit for regulating the AI computing tasks.

[0012] Preferably, multiple of the computing units, storage units, and control units are combined according to a predetermined rule inside the AI chip to form multiple operators, and each operator serves as a basic unit for executing the AI computing tasks.

[0013] Preferably, the connection interface module is used to provide at least one of a USB interface, an Ethernet interface, and a board-to-board interface.

[0014] The present invention also provides an initialization method for an AI computing device of a vehicle intelligent cockpit, including the following steps:

[0015] The intelligent cockpit host identifies the connected device and determines whether it is an AI computing device;

[0016] If it is determined to be an AI computing device, the intelligent cockpit host requests the AI computing device to obtain its key parameters;

[0017] The intelligent cockpit host configures the parameters of the AI computing device according to the obtained key parameters;

[0018] The intelligent cockpit host receives the configuration completion status feedback by the AI computing device. If the configuration is not completed, the configuration operation is re-performed according to a preset rule.

[0019] Preferably, before the intelligent cockpit host identifies the connected device, it further includes:

[0020] The intelligent cockpit host powers the connected devices and receives the feedback power supply status information;

[0021] If the power supply status is normal, the intelligent cockpit host identifies the connected devices;

[0022] If the power supply status is abnormal, the intelligent cockpit host cuts off the power supply and then powers on again.

[0023] Preferably, the key parameters include: the computing power of the AI chip, the memory size, the size of the data storage module, and the register address.

[0024] Preferably, the intelligent cockpit host configures the parameters of the AI computing device according to the obtained key parameters, including at least: setting the transmission rate, working mode, and compatibility of the AI computing device.

[0025] The present invention also provides a cooperative control method for an AI computing device and an intelligent cockpit host, including the following steps:

[0026] The system-on-chip (SOC) of the intelligent cockpit host monitors and identifies the current task type of the intelligent cockpit host in real time;

[0027] When the identified current task type is an AI computing task, the AI computing task is issued to the AI computing device connected to the intelligent cockpit host;

[0028] The AI computing device receives and processes the AI computing task, and feeds back the AI computing result to the system-on-chip (SOC) of the intelligent cockpit host.

[0029] Preferably, when the AI computing device processes the AI computing task, the system-on-chip (SOC) of the intelligent cockpit host executes other non-AI computing tasks in parallel;

[0030] When the identified current task type is a non-AI computing task, the system-on-chip (SOC) of the intelligent cockpit host independently completes the non-AI computing task.

[0031] The present invention also provides a computer program product, characterized by including computer instructions, and the computer instructions direct the computer device to perform the operations corresponding to the method.

[0032] Implementing the present invention has the following beneficial effects: Through the AI computing device with an in-memory computing architecture, both complex and simple computing tasks can be efficiently processed, significantly improving the computing efficiency. Without modifying the original intelligent cockpit host hardware, the present invention connects the AI computing device through various interfaces to achieve the expansion of AI computing power, meeting the on-vehicle requirements of AI functions for low-computing-power host chips at low cost and with low cost, greatly enhancing the AI capabilities of the intelligent cockpit. In addition, the cooperative working mode of the intelligent cockpit host and the AI computing device reasonably distributes tasks, parallelly processes AI tasks, and the host independently completes non-AI tasks, improving the overall system operation efficiency and response speed, and optimizing the user experience of the intelligent cockpit. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0034] Figure 1 It is a schematic diagram of the composition of an AI computing device for a vehicle intelligent cockpit in Embodiment 1 of the present invention.

[0035] Figure 2 It is a schematic diagram of the cooperative work of the intelligent cockpit host and the AI computing device in the embodiment of the present invention.

[0036] Figure 3 It is a schematic diagram of the B2B connection between the AI computing daughter board and the main board of the intelligent cockpit host in the embodiment of the present invention.

[0037] Figure 4 It is a schematic diagram of the composition of an AI chip in the embodiment of the present invention.

[0038] Figure 5 It is a schematic flowchart of an initialization method for an AI computing device of a vehicle intelligent cockpit in Embodiment 2 of the present invention.

[0039] Figure 6 It is a specific flowchart of an initialization method for an AI computing device of a vehicle intelligent cockpit in Embodiment 2 of the present invention.

[0040] Figure 7 It is a schematic flowchart of a cooperative control method between an AI computing device of a vehicle intelligent cockpit and the intelligent cockpit host in Embodiment 3 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] The following descriptions of the embodiments refer to the accompanying drawings to illustrate specific embodiments in which the present invention can be implemented.

[0042] Please refer to Figure 1 As shown, Embodiment 1 of the present invention provides an AI computing device for a vehicle intelligent cockpit, which is connected to and works in cooperation with the intelligent cockpit host. The AI computing device includes:

[0043] An AI chip, a connection interface module, a power management module, and a data storage module;

[0044] The connection interface module is used to connect to the intelligent cockpit host, receive the AI computing tasks assigned by the intelligent cockpit host, and feedback the AI computing results;

[0045] The data storage module is used to store the AI computing tasks and results, as well as the end-side large model that has been trained;

[0046] The AI chip adopts a memory-computation integrated architecture, which further includes: a computing unit for executing the AI computing tasks, a storage unit for arithmetic storage, and a control unit for regulating the AI computing tasks.

[0047] Through the above settings, it can be seen that Embodiment 1 of the present invention achieves the effect of efficiently expanding the AI computing ability for the original cockpit host without changing the hardware structure of the original cockpit host, solves the problem of insufficient computing power of the host chip in mass-produced vehicles in the current market, enables these vehicles originally limited by low computing power to be instantly upgraded, and supports the on-vehicle application of complex AI functions.

[0048] Specifically, in Embodiment 1 of the present invention, please combine with Figure 2 As shown, the connection interface module establishes a connection channel between the AI computing device and the intelligent cockpit host, thereby ensuring that data can be smoothly exchanged between the two. For example, the intelligent cockpit host can transmit the data of the AI computing tasks to be processed to the AI computing device through the connection interface module, and after the AI computing device completes the calculation, it can also transmit the result data back to the intelligent cockpit host through this interface. The connection interface module can adopt various interfaces, such as USB 3.0 interface, Ethernet interface, board-to-board B2B (Board-to-Board) interface, etc., as long as the transmission rate ≥ 50GB / s. As Figure 3 As shown, when using the B2B connection method, the AI computing device can be used as a PC daughter board, and a B2B interface is reserved on the PC main board of the intelligent cockpit host, and the AI computing daughter board can be connected before the host leaves the factory. For the convenience of description, Embodiment 1 of the present invention takes the USB 3.0 interface as an example for subsequent description.

[0049] The power management module is mainly responsible for the power management of the entire AI computing device. It coordinates and controls all power-related operations, from the initial power-on to the subsequent initialization of the intelligent cockpit host. It ensures a stable and appropriate power supply for the AI computing device at different working stages, guaranteeing the normal operation of all components.

[0050] The data storage module is mainly responsible for storing and managing data related to AI computing, including data for AI computing tasks, results generated after AI computing, and end-side large models trained on the vehicle side. For example, when the intelligent cockpit host sends an AI image recognition computing task, the relevant image data is first temporarily stored here. After the AI chip completes the calculation, the final recognition result data is also stored here. At the same time, the data of the pre-trained image recognition model on the vehicle side is also saved by the data storage module. In terms of specific implementation, the data storage module can use a Universal Flash Storage (UFS) chip.

[0051] Please refer to Figure 4 As shown, the AI chip with the in-memory computing architecture includes three types of units: a computing unit, a storage unit, and a control unit. The units communicate with each other at the internal chip level.

[0052] The core responsibility of the computing unit is to execute AI computing tasks, such as performing various arithmetic operations on the input data that meet the requirements of AI algorithms, like convolution calculations and matrix operations in neural networks. It is the core computing part for the AI chip to implement AI functions.

[0053] The storage unit is mainly responsible for storing data during the computing process. It stores all types of data involved in AI computing tasks, including original input data, intermediate calculation data, and final result data, to ensure the read and write requirements of data during the computing process.

[0054] The control unit determines the start, pause, and priority arrangement of AI computing tasks. By reasonably allocating computing resources and coordinating the working rhythm between the computing unit and the storage unit, it ensures the orderly and efficient collaborative work of all parts of the chip.

[0055] In the in-memory computing architecture, multiple computing units, storage units, and control units are combined according to certain rules to form an operator. An operator can be regarded as the basic unit module within the AI chip that undertakes specific AI computing tasks.

[0056] In terms of the task allocation mechanism, when facing simple computing tasks, different operators are each responsible for different computing tasks. They have a clear division of labor and can process multiple relatively simple and independent operations in parallel, improving the overall computing efficiency. When encountering complex tasks, different operators will cooperate with each other to jointly complete the same complex computing task, giving full play to their respective advantages and overcoming complex computing problems through mutual cooperation.

[0057] It should be noted that the operators adopt a synchronous operation method, and they keep consistent in time beats and jointly promote the operation process. This method can avoid time losses such as waiting and coordination caused by asynchrony, so it can greatly improve the computing speed of the entire AI chip and optimize the computing process, making the AI chip more efficient and smooth when processing various AI computing tasks.

[0058] Compared with the von Neumann architecture, in the case of the same AI computing power, the energy efficiency ratio of the memory-compute integrated architecture is increased by 2-3 times, the computing power density is increased by 30%-50%, the latency is reduced by 3-5 times, and the power consumption is reduced by 100%.

[0059] It should also be noted that there can be various specifications for the AI computing device in the embodiments of the present invention, and users can select and purchase according to their needs. For example:

[0060] Specification 1: AI computing power of 30 TOPS + 64G UFS, realizing a 7B text-to-text AI large model;

[0061] Specification 2: AI computing power of 70 TOPS + 128G UFS, realizing a 13B text-to-text and text-to-image AI large model;

[0062] Specification 3: AI computing power of 200 TOPS + 256G UFS, realizing a 33B text-to-text, text-to-image, and text-to-video AI large model.

[0063] After connecting the AI computing device to the intelligent cockpit host through, for example, a USB 3.0 interface, the intelligent cockpit host and the AI computing device complete initialization. As Figure 5 shown, Embodiment 2 of the present invention provides an initialization method for an AI computing device of a vehicle intelligent cockpit, including the following steps:

[0064] The intelligent cockpit host identifies the connected device to determine whether it is an AI computing device;

[0065] If it is determined to be an AI computing device, the intelligent cockpit host requests the AI computing device to obtain its key parameters;

[0066] The intelligent cockpit host configures the parameters of the AI computing device according to the obtained key parameters;

[0067] The intelligent cockpit host receives the configuration completion status feedback from the AI computing device. If the configuration is not completed, the configuration operation will be restarted according to the preset rules.

[0068] Specifically, please refer to Figure 6 As shown, first is step a. The intelligent cockpit host powers the connected AI computing device through the USB 3.0 interface.

[0069] In step b, the intelligent cockpit host continues with the initialization work only after receiving the response that the power supply to the AI computing device is normal; if the AI computing device feedbacks abnormal power supply, the intelligent cockpit host will cut off the power supply to the AI computing device and then power it on again.

[0070] In step c, it is judged whether the device connected to the intelligent cockpit host is an AI computing device. It can be understood that the embodiment of the present invention expands the AI computing power of the existing intelligent cockpit host. Therefore, the AI computing device is an optional product. After the user purchases it, it is directly connected through the external USB3.0 interface of the intelligent cockpit host. This interface is originally configured for the user to connect a USB flash drive to play music, videos, images, etc. stored in the USB flash drive. So when a new device - the AI computing device is connected, the intelligent cockpit host needs to confirm the product type connected to the USB3.0 interface with the AI computing device. It should be noted that if a USB flash drive is connected, the intelligent cockpit host cannot obtain an affirmative reply. When the intelligent cockpit host waits for a preset time, such as 500 ms (this parameter can be adjusted and optimized in combination with the host power-on timing), it will be processed as if a common USB flash drive is connected, and no subsequent steps are required.

[0071] In step d, after receiving the product type confirmation request from the intelligent cockpit host, the AI computing device will feedback an affirmative reply, that is, confirm that the device connected to the USB3.0 interface of the intelligent cockpit host is an AI computing device.

[0072] In step e, the intelligent cockpit host prepares the configuration parameters and requests the AI computing device to feedback the basic parameters and register addresses of the AI chip and the data storage module. Since the AI computing device has multiple specifications according to different AI chip specifications and the sizes of the data storage modules, the intelligent cockpit host needs to obtain the key parameters of the AI computing device before configuration.

[0073] In step f, the AI computing device feedbacks to the intelligent cockpit host the key parameters such as the computing power, memory size, size of the data storage module, and register address of the AI chip.

[0074] Step g, the intelligent cockpit host configures the AI computing device according to the key parameters fed back by the AI computing device, including transmission rate (i.e., the speed or frequency of data transmission), working mode (i.e., the operating state or functional mode of the AI computing device), and other initialization settings (such as compatibility settings). It can be understood that to ensure the seamless integration and compatibility between the AI computing device and the intelligent cockpit host, specific communication protocols, data format conversion rules, or interface adaptation layers need to be configured during initialization, so that the AI computing device can correctly receive and process the AI computing tasks from the intelligent cockpit host and return the calculation results.

[0075] Step h, the intelligent cockpit host receives the configuration completion signal fed back by the AI computing device. If the configuration completion signal is not received, it attempts to reconfigure. When the reconfiguration fails after reaching a predetermined number of times (such as 5 times, which can be adjusted and optimized in combination with the host power-on timing), the power supply of the AI computing device is cut off and the initialization process is restarted.

[0076] After the initialization is completed, the AI computing device can work in coordination with the intelligent cockpit host. Another example is Figure 2 As shown, there is a task allocation module inside the intelligent cockpit host SOC (System on Chip). This module can achieve rapid allocation of host tasks. When it recognizes that the current task of the intelligent cockpit host is an AI computing task, it publishes the task to the AI computing device through the USB 3.0 interface; after the AI computing device completes the calculation, it feeds back the calculation result to the intelligent cockpit host SOC through the USB 3.0. The SOC then proceeds with subsequent work based on the fed-back calculation result. For non-AI tasks, they are independently completed by the intelligent cockpit host SOC chip. It should be noted that for AI computing tasks, the intelligent cockpit host SOC and the AI computing device are in a parallel relationship.

[0077] In an embodiment of the present invention, the AI computing device is designed in a box-like form structure. This box-type AI computing device is convenient for installation and plugging operations in the limited space of the cockpit. When the intelligent cockpit host has a need to expand AI computing power, the user only needs to insert this AI computing device into the corresponding interface of the intelligent cockpit host (such as the USB3.0 interface) like inserting a USB flash drive, and then it can quickly establish a connection with the intelligent cockpit host, automatically enter the initialization process and start working in coordination. The whole process does not require complex tool assistance or the operation of professional technicians, greatly improving the convenience and flexibility of user use, and effectively adapting to the application environment and user operation habits in the cockpit.

[0078] As Figure 7 shown, Embodiment 3 of the present invention also provides a cooperative control method for an AI computing device and an intelligent cockpit host, including the following steps:

[0079] The system on chip (SOC) of the intelligent cockpit host monitors and identifies the current task type of the intelligent cockpit host in real time;

[0080] When it is identified that the current task type is an AI computing task, the AI computing task is published to the AI computing device connected to the intelligent cockpit host;

[0081] The AI computing device receives and processes the AI computing task, and feeds back the AI computing result to the system on chip (SOC) of the intelligent cockpit host.

[0082] Preferably, when the AI computing device processes the AI computing task, the system on chip (SOC) of the intelligent cockpit host executes other non-AI computing tasks in parallel;

[0083] When it is identified that the current task type is a non-AI computing task, the non-AI computing task is independently completed by the system on chip (SOC) of the intelligent cockpit host.

[0084] Corresponding to the initialization method described in the second embodiment of the present invention and the cooperative control method described in the third embodiment of the present invention, the fourth embodiment of the present invention further provides a computer program product, including computer instructions, and the computer instructions instruct a computer device to execute the operations corresponding to the method.

[0085] Preferably, the computer instructions can be divided into one or more modules / units (such as computer program 1, computer program 2,...), and the one or more modules / units are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer program in the device.

[0086] Preferably, the processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or the processor can also be any conventional processor. The processor is the control center of the device and connects various parts of the device through various interfaces and lines.

[0087] The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function, etc., and the data storage area can store relevant data, etc. In addition, the memory can be a high-speed random access memory, or a non-volatile memory, such as a plug-in hard disk, a SmartMedia Card (SMC), a Secure Digital (SD) card, a Flash Card, etc., or the memory can also be other volatile solid-state storage devices.

[0088] It should be noted that the above device may include but is not limited to a processor and a memory, which can be understood by those skilled in the art.

[0089] Regarding the working principle and process of the above embodiments, refer to the description of Embodiment 1 of the present invention above, and details will not be repeated here.

[0090] It can be seen from the above description that, compared with the prior art, the beneficial effects of the present invention are as follows: Through the AI computing device with an in-memory computing architecture, both complex and simple computing tasks can be efficiently processed, significantly improving the computing efficiency. Without modifying the hardware of the original intelligent cockpit host, the present invention connects the AI computing device through various interfaces to achieve the expansion of AI computing power, meeting the on-vehicle requirements of the AI function of low-computing-power host chips at low cost and with low cost, greatly enhancing the AI ability of the intelligent cockpit. In addition, the cooperative working mode of the intelligent cockpit host and the AI computing device reasonably distributes tasks, processes AI tasks in parallel, and the host independently completes non-AI tasks, improving the overall system operation efficiency and response speed, and optimizing the user experience of the intelligent cockpit.

[0091] The above-disclosed are only the preferred embodiments of the present invention, and of course, the scope of the rights of the present invention cannot be limited thereby. Therefore, equivalent changes made according to the claims of the present invention still fall within the scope covered by the present invention.

Claims

1. An AI computing device for a vehicle intelligent cockpit, characterized in that: The AI ​​computing device is connected to and works in coordination with the smart cockpit host, including: AI chip, connection interface module, power management module and data storage module; The connection interface module is used to connect to the smart cockpit host, receive the AI ​​computing tasks assigned by the smart cockpit host and feed back the AI ​​computing results; The data storage module is used to store the AI ​​computing tasks and computing results, as well as the trained end-side large model; The AI ​​chip adopts a storage and computing integrated architecture, which further includes: a computing unit for executing the AI ​​computing tasks, a storage unit for computing storage, and a control unit for regulating the AI ​​computing tasks.

2. The AI ​​computing device according to claim 1, characterized in that: The AI ​​chip internally combines multiple computing units, storage units, and control units according to predetermined rules to form multiple operators, and each operator serves as a basic unit for executing the AI ​​computing task.

3. The AI ​​computing device according to claim 1, characterized in that: The connection interface module is used to provide at least one of a USB interface, an Ethernet interface, and a board-to-board interface.

4. A method for initializing an AI computing device for a vehicle intelligent cockpit according to any one of claims 1 to 3, characterized in that: The following steps are involved: The smart cockpit host identifies the connected device and determines whether it is an AI computing device; If it is determined to be an AI computing device, the intelligent cockpit host requests the AI ​​computing device to obtain its key parameters; The intelligent cockpit host configures the parameters of the AI ​​computing device based on the key parameters obtained; The intelligent cockpit host receives the configuration completion status feedback from the AI ​​computing device. If the configuration is not completed, the configuration operation is re-performed according to the preset rules.

5. The method according to claim 4, characterized in that Before the smart cockpit host recognizes the connected device, it also includes: The smart cockpit host supplies power to the connected devices and receives feedback on the power supply status; If the power supply status is normal, the smart cockpit host will identify the connected device; If the power supply status is abnormal, the smart cockpit host will cut off the power supply and then re-supply it.

6. The method according to claim 4, characterized in that The key parameters include: the computing power of the AI ​​chip, the memory size, the size of the data storage module, and the register address.

7. The method according to claim 4, characterized in that The smart cockpit host performs parameter configuration on the AI ​​computing device according to the acquired key parameters, which at least includes: setting the transmission rate, working mode and compatibility of the AI ​​computing device.

8. A collaborative control method of an AI computing device and a smart cockpit host according to any one of claims 1 to 3, characterized in that: The following steps are involved: The system-on-chip (SOC) of the smart cockpit host monitors and identifies the current task type of the smart cockpit host in real time; When the current task type is identified as an AI computing task, the AI ​​computing task is published to the AI ​​computing device connected to the smart cockpit host; The AI ​​computing device receives and processes the AI ​​computing tasks, and feeds back the AI ​​computing results to the system-on-chip SOC of the smart cockpit host.

9. The method according to claim 8, characterized in that When the AI ​​computing device processes AI computing tasks, the system-on-chip (SOC) of the smart cockpit host performs other non-AI computing tasks in parallel; When it is identified that the current task type is a non-AI computing task, the system on chip SOC of the smart cockpit host independently completes the non-AI computing task.

10. A computer program product, characterized in that The method comprises computer instructions, wherein the computer instructions instruct a computer device to execute an operation corresponding to the method as claimed in any one of claims 4 to 7, or to execute an operation corresponding to the method as claimed in any one of claims 8 to 9.

Images