Update method and device of intelligent cabin system, control method and device of intelligent cabin system, automobile and medium

By extracting the sentence information of the smart cockpit system and using artificial intelligence models to generalize data, the problem of manually obtaining sample data in the update of the smart cockpit system is solved, efficient update and iteration is achieved, and the system update speed and quality is improved.

CN119025622APending Publication Date: 2024-11-26GUANGZHOU XIAOPENG MOTORS TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411211497.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

During the update process of the existing smart cockpit system, due to manual acquisition of sample data, the update speed and quality are poor, and a lot of manpower and material resources are consumed.

Method used

By extracting the sentence meaning information of the initial sample data, using artificial intelligence models to generalize and expand data, generate high-quality training data, and realize the rapid update and iteration of the intelligent cockpit system.

Benefits of technology

It saves manpower and material costs, improves the speed and quality of the smart cockpit system update, and realizes rapid response to functional iteration and continuous optimization of training data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119025622A_ABST
    Figure CN119025622A_ABST
Patent Text Reader

Abstract

The invention provides an updating method and device of an intelligent cabin system, a control method and device of the intelligent cabin system, an automobile and a medium. The updating method comprises the steps of obtaining initial sample data; extracting sentence meaning information of the initial sample data; inputting the initial sample data and the sentence meaning information into a preset first artificial intelligence model, and obtaining extended training data based on the output of the first artificial intelligence model; and updating the intelligent cabin system by using the extended training data. Compared with a manual generalization mode, the training data can be efficiently obtained, and a large amount of manpower and material resource cost is saved; besides, abstract thinking is carried out on the initial sample data and sentence meaning information thereof through the artificial intelligence model so as to expand the sample data, the obtained training data is higher in quality, and rapid response to intelligent cabin system function iteration and continuous optimization of the training data can be realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of smart cockpits, and in particular to an updating method, a control method, a device, a car and a medium for a smart cockpit system. Background Art

[0002] At present, the smart cockpit system of the car integrates various types of functional tasks such as voice reminders, navigation and in-car equipment control. With the update and upgrade of the car machine functions, the smart cockpit system often faces the iteration of functional points and poor quality input data. Poor quality input data may affect the quality of the next update of the smart cockpit system. At present, the smart cockpit products on the market mainly obtain some samples manually, and input the samples as training data into the smart cockpit products to realize the functional update of the smart cockpit products, and realize the correction and iteration of the functions. However, this method requires a lot of manpower and material resources, which will affect the update speed and quality of the smart cockpit system to a certain extent. Summary of the invention

[0003] The present invention application provides an update method, control method, device, automobile and medium of an intelligent cockpit system, which can improve the update speed and quality of the intelligent cockpit system.

[0004] In order to solve or partially solve the above technical problems, the present invention provides an updating method of a smart cockpit system, comprising:

[0005] Get initial sample data;

[0006] Extracting sentence meaning information of the initial sample data;

[0007] Inputting the initial sample data and the sentence meaning information into a preset first artificial intelligence model, and obtaining expanded training data based on the output of the first artificial intelligence model;

[0008] The intelligent cockpit system is updated using the expanded training data.

[0009] The present invention application extracts the sentence meaning information of the initial sample data and uses the artificial intelligence model to generalize and expand the initial sample data. Compared with the manual generalization method, it can efficiently obtain training data and save a lot of manpower and material costs. In addition, by abstractly thinking about the initial sample data and its sentence meaning information through the artificial intelligence model to expand the sample data, the obtained training data is of higher quality, which can achieve rapid response to the functional iteration of the intelligent cockpit system and continuous optimization of the training data.

[0010] As a preferred solution, the smart cockpit system is provided with a smart cockpit model, and the method of updating the smart cockpit system by using the expanded training data includes:

[0011] Inputting the expanded training data into the smart cockpit model, and obtaining a target sentence of the expanded training data based on an output of the smart cockpit model;

[0012] When the preset semantic conditions are met, target training data is obtained according to the target sentence;

[0013] The smart cockpit system is updated using the target training data.

[0014] As a preferred solution, when the preset semantic conditions are met, obtaining target training data according to the target sentence includes:

[0015] Converting the target sentence into a target sentence vector;

[0016] Calculating a first semantic distance according to a preset description vector and the target sentence vector;

[0017] When the first semantic distance is less than a preset semantic threshold, the target sentence is set as the target training data.

[0018] As a preferred solution, the target clause includes a conditional clause and an action clause;

[0019] When the preset semantic condition is met, target training data is obtained according to the target sentence, including:

[0020] Convert the conditional clause into a conditional clause vector, and convert the action clause into an action clause vector;

[0021] Calculating a second semantic distance based on a preset condition description vector and the conditional sentence vector; calculating a third semantic distance based on a preset action description vector and the action sentence vector;

[0022] When both the second semantic distance and the third semantic distance are smaller than a preset semantic threshold, the conditional clause and the action clause are set as target training data.

[0023] As a preferred solution, the sentence meaning information includes scene information, sentence pattern information and keyword information;

[0024] The extracting sentence meaning information of the initial sample data includes:

[0025] Inputting the initial sample data and the pre-configured sentence meaning template into a preset second artificial intelligence model, wherein the sentence meaning template includes scene template information, sentence pattern template information and keyword template information;

[0026] Based on the output of the second artificial intelligence model, the scene information, sentence information and keyword information are obtained.

[0027] The present invention application also provides a control method for an intelligent cockpit system, wherein the intelligent cockpit system is updated in advance by the updating method for the intelligent cockpit system described in any one of the above solutions;

[0028] The control method includes: obtaining a voice request or an operation request; inputting the voice request or the operation request into the smart cockpit system, obtaining a control instruction of a target vehicle based on the smart cockpit system, and controlling the target vehicle using the control instruction.

[0029] The present invention application also provides an update device for an intelligent cockpit system, including a data acquisition module, an extraction module, an expansion module and an update module; wherein:

[0030] The data acquisition module is used to acquire initial sample data;

[0031] The extraction module is used to extract sentence meaning information of the initial sample data;

[0032] The expansion module is used to input the initial sample data and the sentence meaning information into a preset first artificial intelligence model, and obtain expanded training data based on the output of the first artificial intelligence model;

[0033] The updating module is used to update the smart cockpit system by using the expanded training data.

[0034] The present invention application also provides a control device for an intelligent cockpit system, wherein the intelligent cockpit system is updated in advance by the updating method for the intelligent cockpit system described in any one of the above solutions;

[0035] The control device includes a request acquisition module and a control module; wherein,

[0036] The request acquisition module is used to acquire a voice request or an operation request;

[0037] The control module is used to input the voice request or operation request into the smart cockpit system, obtain a control instruction of the target vehicle based on the smart cockpit system, and control the target vehicle using the control instruction.

[0038] The present application also provides a car, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the updating method of the smart cockpit system described in any one of the above-mentioned schemes, or the control method of the smart cockpit system described in any one of the above-mentioned schemes is implemented.

[0039] The present application also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the updating method of the smart cockpit system described in any of the above-mentioned schemes, or the control method of the smart cockpit system described in any of the above-mentioned schemes. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 : A flow chart of an implementation method of the updating method of the smart cockpit system provided in the present application.

[0041] Figure 2 : A flow chart of an implementation method of the control method of the intelligent cockpit system provided in the present application.

[0042] Figure 3 : A structural schematic diagram of an embodiment of an updating device for a smart cockpit system provided in the present application.

[0043] Figure 4 : A structural schematic diagram of an embodiment of a control device for an intelligent cockpit system provided in the present application. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] Embodiment 1

[0046] According to relevant technical records, mainstream smart cockpit products on the market integrate various types of functional tasks such as voice reminders, navigation, and in-car device control. With the update and upgrade of vehicle computer functions, smart cockpit systems often face frequent iterations of functional points and poor quality input data.

[0047] At present, some companies are considering setting up large models in smart cockpit systems to cope with the endless new needs of users and the update and iteration of functions. The performance of large models depends on the amount and quality of training data. Poor quality input data will affect the update efficiency of the smart cockpit system and the quality of the next update.

[0048] The smart cockpit products of some companies mainly obtain sample data manually, and input the samples into the smart cockpit products as training data to update the functions of the smart cockpit products and realize the correction and iteration of the functions. However, this method requires a lot of manpower and material resources, and is limited by people's limited energy and subjective differences between different processors. This method of sample acquisition will affect the update speed and quality of the smart cockpit system to a certain extent.

[0049] For one or more of the above technical issues, please refer to Figure 1 , Figure 1 A flowchart of an implementation method of an update method for a smart cockpit system provided in the present application.

[0050] The updating method of the smart cockpit system described in some embodiments of the present application can be applied to a computer device, which may be a vehicle-mounted terminal.

[0051] The updating method of the smart cockpit system includes steps S101 to S104; wherein the specific contents of each step are as follows:

[0052] Step S101, obtaining initial sample data.

[0053] In this step, the initial sample data may be in the form of a sentence, and the sentence may be a bad case.

[0054] Badcase refers to errors or undesirable results that occur in the actual application of the model. These errors may be caused by data problems, problems with the model itself, or problems with the application environment. Common badcases include inaccurate model output, misunderstanding of the model, and inappropriate model-generated content.

[0055] The sentences or sample data involved in the badcase described in this application mainly correspond to the functions to be updated in the smart cockpit system. Data is the basis of the model. The subsequent steps of this application can help improve the performance of the smart cockpit system by processing the badcase and improving the quality of the samples.

[0056] Step S102: extracting sentence meaning information of the initial sample data.

[0057] In this step, the sentence meaning information includes the scene information, sentence structure information and keyword information of the sentence.

[0058] The extracting of sentence meaning information of the initial sample data includes: inputting the initial sample data and a pre-configured sentence meaning template into a preset second artificial intelligence model, wherein the sentence meaning template includes scene template information, sentence pattern template information and keyword template information; and obtaining the scene information, sentence pattern information and keyword information based on the output of the second artificial intelligence model.

[0059] Taking into account that some embodiments are mainly for training models in smart cockpit systems, and models can better understand how models think. Therefore, the preset second artificial intelligence model described in some implementation methods of the present invention can adopt a super-large model (such as Wenxin Yiyan / GPT4 and other models above 200B) to refine the sentence meaning information of the initial sample data. And then extract some prompts about sentence meaning information. The prompt described in this application refers to the input of the model, that is, these prompts will be used as the input of other models in subsequent steps for further processing.

[0060] For example, for the above scenario information, the second artificial intelligence model is input with {assuming that you are a product manager of an in-vehicle smart cockpit, I will give you a sentence, and your task is to extract the user scenario information of this sentence.}. Then, several examples are input as the above scenario template information, for example:

[0061] {Example:

[0062] Sentence: Remind me not to forget to check in at 8 o'clock tonight.

[0063] Scene information: Voice reminders triggered based on time information.}

[0064] The rest of the examples are similar and will not be described in detail. Based on the output of the second artificial intelligence model, the scene information is obtained. For example, the second artificial intelligence model outputs {automatic operation instructions triggered based on location information}. At this time, the scene prompt can be extracted.

[0065] For example, for the above sentence information, the following content (including several examples as sentence template information) can be input into the second artificial intelligence model, specifically:

[0066] {Suppose you are a language expert model, I will give you a sentence, your task is to think from a more abstract perspective and extract the sentence structure of this sentence. The sentence structure needs to be clear, easy to disassemble and generalize.

[0067] Example:

[0068] Sentence: Remind me not to forget to check in at 8 o'clock tonight.

[0069] Sentence structure: (date) + (time point) + remind me + (remind me of something).

[0070] Sentence: Next time you pass by here, please turn on the air conditioner to 22 degrees.}

[0071] The rest of the examples are similar and will not be described in detail. Based on the output of the second artificial intelligence model, the sentence information is obtained. For example, the second artificial intelligence model outputs {(time adverbial, such as "next time when I pass by here") + help me + (execute action, such as "turn on the air conditioner") + to + (specific degree / state, such as "twenty-two degrees")}. At this time, the sentence prompt can be extracted.

[0072] For another example, for the above keyword information, the following content (including several examples as keyword template information) may be input into the second artificial intelligence model, specifically:

[0073] {Suppose you are a product manager for an in-vehicle smart cockpit. I will give you a sentence. Your task is to extract the keywords (excluding numerical values) of this sentence, no more than 2 words.

[0074] Example:

[0075] Sentence: Help me adjust the seat position to memory position one every morning when starting the car.

[0076] Keywords: every morning, seat position.}

[0077] The rest of the examples are similar and will not be described in detail. Based on the output of the second artificial intelligence model, the keyword information is obtained. For example, the second artificial intelligence model outputs {passing by here, air conditioning}. At this time, the keyword prompt can be extracted.

[0078] Based on the scene prompt, sentence prompt and keyword prompt, the sentence meaning information in step S102 is obtained.

[0079] Step S103: input the initial sample data and the sentence meaning information into a preset first artificial intelligence model, and obtain expanded training data based on the output of the first artificial intelligence model.

[0080] In this step, the first artificial intelligence model can also be the above-mentioned super large model, such as Wenxinyiyan / GPT4 and other models with a size of more than 200B, which is used to expand and enhance the initial sample data.

[0081] Specifically, based on the above-mentioned refined sentence meaning information, the first artificial intelligence model can perform abstract thinking on the "sentence meaning", thereby realizing the expansion of sample data. Compared with the manual method of obtaining sample data, the training data of the intelligent cockpit system can be obtained efficiently, saving manpower and material costs.

[0082] Furthermore, the sentence meaning information may be the above-mentioned scene prompt, sentence prompt and keyword prompt. The initial sample data, the above-mentioned scene prompt, sentence prompt and keyword prompt are input into the first artificial intelligence model, for example:

[0083] {Suppose you are a data generation expert. Now I am training a model in the field of automotive smart cockpit, but the model cannot give the correct output for this sentence. I will give you a sentence and some of its meaning information. Please generate some generalized sentences based on the sentence meaning information as my training data to help the model learn this sentence.

[0084] Badcase: Next time you pass by here, please turn on the air conditioner to 22 degrees

[0085] Scene information: automatic operation instructions triggered based on location information.

[0086] Sentence structure: (time adverbial, such as "next time I pass by here") + help me + (perform an action, such as "turn on the air conditioner") + to + (specific degree / state, such as "twenty-two degrees").

[0087] Keywords: passing by here, air conditioning.

[0088] Generate a generalized sentence:

[0089] Based on the output of the first artificial intelligence model, the generalized sentence obtained is, for example:

[0090] {Please keep my car windows closed every time I pass this gas station.

[0091] Next time, please help me adjust the sound system to a comfortable volume before entering the city.

[0092] Please turn on the heated seats for me on my way home from get off work every day.

[0093] Please help me set the navigation to avoid congested routes when I leave tomorrow morning.

[0094] Help me empty the trunk and check tire pressure before a weekend trip.

[0095] It helps me automatically check whether the car doors are locked every time I park.

[0096] Help me set the vehicle suspension to off-road mode when we are about to enter the mountainous road section.

[0097] Help me brighten my headlights and turn on the high beam assist function when driving at night.

[0098] Help me set the cruise control to 100 km / h next time I enter the highway.

[0099] Help me adjust the temperature in my car to 25 degrees before going to work every day.}

[0100] Based on the output of the first artificial intelligence model, several sentences are obtained, namely, the expanded training data. These sentences are obtained based on the initial sample data, which fully understands the sentence meaning information of the bad case and generalizes the initial bad case. The data quality of the obtained expanded training data is higher, and the sample size is sufficient to support the update of the target function of the intelligent cockpit system.

[0101] Step S104: using the expanded training data to update the smart cockpit system.

[0102] In this step, the expanded training data is input into the smart cockpit model of the smart cockpit system, and the target sentence of the expanded training data is obtained based on the output of the smart cockpit model; when the preset semantic conditions are met, the target training data is obtained according to the target sentence; and the smart cockpit system is updated using the target training data.

[0103] For example, the augmented training data {help me turn on the air conditioner to 22 degrees next time I pass by here} is input, and then the target sentence such as {the next time I pass by here} or {turn on the air conditioner to 22 degrees} can be obtained.

[0104] The semantics of the target sentence are analyzed, and when the target sentence meets the preset semantic conditions, target training data are obtained. These target training data are used as training data for the smart cockpit system to update the smart cockpit system, so that the functions of the smart cockpit system can be iteratively upgraded.

[0105] Furthermore, the step of determining whether the preset semantic condition is met includes:

[0106] The target sentence is converted into a target sentence vector; a first semantic distance is calculated according to a preset description vector and the target sentence vector; when the first semantic distance is less than a preset semantic threshold, the target sentence is set as the target training data.

[0107] Among them, the conversion of target sentences into target sentence vectors can be achieved through the m3e model. m3e is the abbreviation of Moka MassiveMixed Embedding. Moka means that the model is trained, open source and evaluated by MokaAI. Massive means that this model is trained on a dataset of tens of millions of Chinese sentences. Mixed means that this model supports functions such as homogeneous text similarity calculation in Chinese and English, heterogeneous text retrieval, and obtaining sentence vectors.

[0108] After obtaining the target sentence vector through the m3e model, the first semantic distance can be calculated based on the preset description vector and the target sentence vector.

[0109] The preset description vector may be pre-stored in a description list, and each target sentence corresponds to at least one description list. Based on the similarity between the target sentence vector and each description vector of an item in the description list, several description vectors with the highest similarity may be screened out.

[0110] Specifically, the first semantic distance can be calculated using faiss. Faiss is a text content similarity calculation tool produced by Meta AI. It calculates the Euclidean distance between the preset description vector and the target sentence vector in a given dimension to obtain the first semantic distance, which can be understood as the similarity between the two.

[0111] When it is determined that the first semantic distance is less than a preset semantic threshold, the target sentence is set as the target training data and put into a training data set for updating the smart cockpit system.

[0112] Further, the target clause includes a conditional clause and an action clause. When the preset semantic condition is met, the target training data is obtained according to the target clause, including: converting the conditional clause into a conditional sentence vector, converting the action clause into an action sentence vector; calculating a second semantic distance according to a preset condition description vector and the conditional sentence vector; calculating a third semantic distance according to a preset action description vector and the action sentence vector; and setting the conditional clause and the action clause as target training data when both the second semantic distance and the third semantic distance are less than a preset semantic threshold.

[0113] For example, the {next time I pass by here} is a conditional clause, and the {turn on the air conditioner to 22 degrees} is an action clause.

[0114] Through the above m3e model, the conditional clause can be converted into a conditional sentence vector, and the action clause can be converted into an action sentence vector. Then, faiss is used to calculate the second semantic distance according to the preset condition description vector and the conditional sentence vector; the third semantic distance is calculated according to the preset action description vector and the action sentence vector, and when the second semantic distance and the third semantic distance are both less than the preset semantic threshold, the conditional clause and the action clause are set as target training data.

[0115] By dividing the target sentence into conditional clauses and action clauses, the semantic quality of the target sentence can be effectively grasped based on the second semantic distance and the third semantic distance. That is, when both are less than the preset semantic threshold, the target sentence is included in the target training data, which can ensure that the data quality of the target training data is higher, and further improve the iteration quality and efficiency of the intelligent cockpit system. The target performance can be achieved through iterative optimization with the smallest number of iterations possible.

[0116] Furthermore, if Figure 2 As shown, based on the above-mentioned updating method of the smart cockpit system, an embodiment of the present invention further provides a control method of the smart cockpit system, wherein the smart cockpit system is updated in advance by the above-mentioned updating method of the smart cockpit system.

[0117] The control method of the smart cockpit system includes steps S201 to S202. The specific contents of each step are as follows:

[0118] Step S201, obtaining a voice request or an operation request.

[0119] Step S202: Input the voice request or operation request into the smart cockpit system, obtain the control instruction of the target vehicle based on the smart cockpit system, and use the control instruction to control the target vehicle. In this embodiment, the user inputs the voice request or operation request to the smart cockpit system by voice or operation. The smart cockpit system recognizes the request content and obtains the control instruction of the target vehicle, so that the target vehicle can adapt to the updated and iterative smart cockpit system function.

[0120] In addition, if Figure 3 As shown, the present invention also provides an updating device 300 for a smart cockpit system, including a data acquisition module 301, an extraction module 302, an expansion module 303 and an updating module 304; wherein,

[0121] The data acquisition module 301 is used to acquire initial sample data;

[0122] The extraction module 302 is used to extract sentence meaning information of the initial sample data;

[0123] The expansion module 303 is used to input the initial sample data and the sentence meaning information into a preset first artificial intelligence model, and obtain expanded training data based on the output of the first artificial intelligence model;

[0124] The updating module 304 is used to update the smart cockpit system by using the expanded training data.

[0125] In some preferred embodiments, the update module 304 includes a first update submodule, the smart cockpit system is provided with a smart cockpit model, and the first update unit is used to: input the expanded training data into the smart cockpit model, and obtain the target sentence of the expanded training data based on the output of the smart cockpit model; when the preset semantic conditions are met, obtain the target training data according to the target sentence; and use the target training data to update the smart cockpit system.

[0126] In some preferred embodiments, the first updating submodule includes a first setting unit, and the first setting subunit is used to: convert the target sentence into a target sentence vector; calculate a first semantic distance based on a preset description vector and the target sentence vector; when the first semantic distance is less than a preset semantic threshold, set the target sentence as the target training data.

[0127] In some preferred embodiments, the target sentence includes a conditional sentence and an action sentence; the first setting unit includes a first setting sub-unit, and the first setting sub-unit is used to: convert the conditional sentence into a conditional sentence vector, and convert the action sentence into an action sentence vector; calculate a second semantic distance based on a preset condition description vector and the conditional sentence vector; calculate a third semantic distance based on a preset action description vector and the action sentence vector; when the second semantic distance and the third semantic distance are both less than a preset semantic threshold, set the conditional sentence and the action sentence as target training data.

[0128] In some preferred implementations, the sentence meaning information includes scene information, sentence pattern information, and keyword information;

[0129] The extraction module 302 includes an extraction submodule, which is used to: input the initial sample data and a pre-configured sentence meaning template into a preset second artificial intelligence model, wherein the sentence meaning template includes scene template information, sentence pattern template information and keyword template information; and obtain the scene information, sentence pattern information and keyword information based on the output of the second artificial intelligence model.

[0130] like Figure 4 As shown, the present application further provides a control device 400 of a smart cockpit system, wherein the smart cockpit system is updated in advance by the updating method of the smart cockpit system described in any one of the above solutions;

[0131] The control device includes a request acquisition module 401 and a control module 402; wherein,

[0132] The request acquisition module 401 is used to acquire a voice request or an operation request;

[0133] The control module 402 is used to input the voice request or operation request into the smart cockpit system, obtain a control instruction of the target vehicle based on the smart cockpit system, and control the target vehicle using the control instruction.

[0134] The present application also provides a car, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the updating method of the smart cockpit system described in any one of the above-mentioned schemes, or the control method of the smart cockpit system described in any one of the above-mentioned schemes is implemented.

[0135] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal, and various interfaces and lines are used to connect various parts of the entire terminal.

[0136] The memory can be used to store the computer program, and the processor realizes various functions of the terminal by running or executing the computer program stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0137] The present application also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the updating method of the smart cockpit system described in any of the above-mentioned schemes, or the control method of the smart cockpit system described in any of the above-mentioned schemes.

[0138] Wherein, if the update device / terminal of the smart cockpit system or the module integrated by the control update device / terminal of the smart cockpit system is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0139] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0140] The present invention application provides an update method, control method, device, automobile and medium for a smart cockpit. The present invention application extracts sentence meaning information of initial sample data and uses an artificial intelligence model to generalize and expand the initial sample data. Compared with the manual generalization method, the training data can be obtained efficiently, saving a lot of manpower and material costs. In addition, the initial sample data and its sentence meaning information are abstractly thought about by the artificial intelligence model to expand the sample data, and the obtained training data is of higher quality, which can achieve rapid response to the iteration of smart cockpit system functions and continuous optimization of training data.

[0141] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for updating a smart cockpit system, characterized in that: include: Get initial sample data; Extracting sentence meaning information of the initial sample data; Inputting the initial sample data and the sentence meaning information into a preset first artificial intelligence model, and obtaining expanded training data based on the output of the first artificial intelligence model; The intelligent cockpit system is updated using the expanded training data.

2. The method for updating the smart cockpit system according to claim 1, characterized in that: The smart cockpit system is provided with a smart cockpit model, and the use of the expanded training data to update the smart cockpit system includes: Inputting the expanded training data into the smart cockpit model, and obtaining a target sentence of the expanded training data based on an output of the smart cockpit model; When the preset semantic conditions are met, target training data is obtained according to the target sentence; The smart cockpit system is updated using the target training data.

3. The method for updating the smart cockpit system according to claim 2, characterized in that: When the preset semantic condition is met, target training data is obtained according to the target sentence, including: Converting the target sentence into a target sentence vector; Calculating a first semantic distance according to a preset description vector and the target sentence vector; When the first semantic distance is less than a preset semantic threshold, the target sentence is set as the target training data.

4. The method for updating the smart cockpit system according to claim 2, characterized in that: The target clause includes a conditional clause and an action clause; When the preset semantic condition is met, target training data is obtained according to the target sentence, including: Convert the conditional clause into a conditional clause vector, and convert the action clause into an action clause vector; Calculating a second semantic distance based on a preset condition description vector and the conditional sentence vector; calculating a third semantic distance based on a preset action description vector and the action sentence vector; When both the second semantic distance and the third semantic distance are smaller than a preset semantic threshold, the conditional clause and the action clause are set as target training data.

5. The method for updating the smart cockpit system according to claim 1, characterized in that: The sentence meaning information includes scene information, sentence pattern information and keyword information; The extracting sentence meaning information of the initial sample data includes: Inputting the initial sample data and the pre-configured sentence meaning template into a preset second artificial intelligence model, wherein the sentence meaning template includes scene template information, sentence pattern template information and keyword template information; Based on the output of the second artificial intelligence model, the scene information, sentence information and keyword information are obtained.

6. A control method for an intelligent cockpit system, characterized in that: The smart cockpit system is updated in advance by the smart cockpit system updating method according to any one of claims 1 to 5; The control method comprises: Get voice request or operation request; The voice request or operation request is input into the smart cockpit system, a control instruction of a target vehicle is obtained based on the smart cockpit system, and the target vehicle is controlled using the control instruction.

7. An updating device for a smart cockpit system, characterized in that: It includes data acquisition module, extraction module, expansion module and update module; among which, The data acquisition module is used to acquire initial sample data; The extraction module is used to extract sentence meaning information of the initial sample data; The expansion module is used to input the initial sample data and the sentence meaning information into a preset first artificial intelligence model, and obtain expanded training data based on the output of the first artificial intelligence model; The updating module is used to update the smart cockpit system by using the expanded training data.

8. A control device for an intelligent cockpit system, characterized in that: The smart cockpit system is updated in advance by the smart cockpit system updating method according to any one of claims 1 to 5; The control device includes a request acquisition module and a control module; wherein, The request acquisition module is used to acquire a voice request or an operation request; The control module is used to input the voice request or operation request into the smart cockpit system, obtain a control instruction of the target vehicle based on the smart cockpit system, and control the target vehicle using the control instruction.

9. A car, characterized in that: It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the updating method of the smart cockpit system as described in any one of claims 1 to 5, or the control method of the smart cockpit system as described in claim 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein, when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the updating method of the smart cockpit system as described in any one of claims 1 to 5, or the control method of the smart cockpit system as described in claim 6.