Vehicle control processing method, device, equipment, medium and product

Abstract

The invention provides a vehicle control processing method and device, equipment, a medium and a product, and the method comprises the steps: carrying out the driving scene classification of a received user request, obtaining a driving scene type, and enabling the driving scene type to represent the current driving environment of a user; acquiring a current state parameter of a vehicle control object related to the driving scene category; and determining control recommendation information of the vehicle control object according to the driving scene category and the current state parameter of the vehicle control object. The intelligent degree of vehicle control can be improved.

Description

Technical Field

[0001] The present application relates to the field of vehicle control, and in particular to a vehicle control processing method, device, equipment, medium and product. Background Art

[0002] The car computer system is an important part of the car, integrating multiple functions such as vehicle information display, entertainment, navigation and vehicle control, providing drivers with a convenient and intelligent driving experience. With the development of technology, the car computer system has gradually developed from simple information display to highly intelligent and personalized.

[0003] At present, the vehicle computer system mainly manages and adjusts the vehicle's status control functions through sensors and basic control logic, such as automatically closing windows through rain sensors, adjusting air conditioning through temperature sensors, adjusting lighting through light sensors, and intelligently reducing the volume through noise recognition devices. However, the intelligence and personalization levels of these functions are low. How to further improve the intelligence level of vehicle status control functions has become an urgent problem to be solved. Summary of the invention

[0004] Based on the above technical status, the present application provides a vehicle control processing method, device, equipment, medium and product, which can improve the intelligence level of vehicle status control function.

[0005] In order to achieve the above technical objectives, this application specifically proposes the following technical solutions:

[0006] According to a first aspect of an embodiment of the present application, a vehicle control processing method is provided, comprising: performing driving scenario classification on a received user request to obtain a driving scenario category, wherein the driving scenario category represents the user's current driving environment; obtaining current state parameters of a vehicle control object related to the driving scenario category; and determining control recommendation information for the vehicle control object based on the driving scenario category and the current state parameters of the vehicle control object.

[0007] In some implementations, the driving scene classification of the received user request to obtain the driving scene category includes: inputting the received user request into a scene classification model, and classifying the received user request by the scene classification model to obtain the driving scene category; wherein the scene classification model is obtained by training a bidirectional encoder representation model according to user request samples under each driving scene category and the corresponding labeled driving scene category.

[0008] In some implementations, the user request samples under each driving scenario category are obtained by the following steps: for each driving scenario category in the driving scenario categories, an original user request sample under the driving scenario category is obtained; the original user request samples under the driving scenario category are subjected to corpus generalization processing to obtain generalized user request samples under the driving scenario category, wherein the generalized user request samples are different expressions of the original user request samples; and the user request samples under the driving scenario category are obtained according to the original user request samples and the generalized user request samples under the driving scenario category.

[0009] In some implementations, obtaining the current state parameters of the vehicle control object related to the driving scenario category includes: determining the vehicle control object corresponding to the driving scenario category based on the driving scenario category and the correspondence between the driving scenario category and the vehicle control object; and obtaining the current state parameters of the vehicle control object corresponding to the driving scenario category.

[0010] In some implementations, determining the control recommendation information for the vehicle control object based on the driving scenario category and the current state parameters of the vehicle control object includes: inputting the driving scenario category and the current state parameters of the vehicle control object into a control recommendation model, and determining the control recommendation information for the vehicle control object based on the driving scenario category and the current state parameters of the vehicle control object through the control recommendation model; wherein the control recommendation model is obtained by fine-tuning a large model based on state parameter samples of the vehicle control object under each driving scenario category and the corresponding annotated control recommendation information.

[0011] In some implementations, the annotated control recommendation information is obtained by the following steps: for each driving scenario category in the various driving scenario categories, original control recommendation information corresponding to a state parameter sample of a vehicle control object under the driving scenario category is obtained; corpus generalization processing is performed on the original control recommendation information under the driving scenario category to obtain generalized control recommendation information under the driving scenario category, wherein the generalized control recommendation information is a different representation of the original control recommendation information; and based on the original control recommendation information and the generalized control recommendation information under the driving scenario category, annotated control recommendation information corresponding to a state parameter sample of a vehicle control object under the driving scenario category is obtained.

[0012] In some implementations, determining the control recommendation information for the vehicle control object based on the driving scenario category and the current state parameters of the vehicle control object includes: obtaining the user's preference control data for the vehicle control object; and determining the control recommendation information for the vehicle control object based on the driving scenario category, the current state parameters of the vehicle control object and the preference control data.

[0013] According to a second aspect of an embodiment of the present application, a vehicle control processing device is provided, comprising: a classification unit, configured to perform driving scene classification on a received user request to obtain a driving scene category, wherein the driving scene category represents a current driving environment in which the user is located; an acquisition unit, configured to acquire current state parameters of a vehicle control object related to the driving scene category; and a determination unit, configured to determine control recommendation information for the vehicle control object based on the driving scene category and the current state parameters of the vehicle control object.

[0014] According to a third aspect of an embodiment of the present application, an electronic device is provided, comprising a memory and a processor; the memory is connected to the processor and is used to store programs; the processor is used to implement the vehicle control processing method as described in the first aspect by running the program in the memory.

[0015] According to a fourth aspect of an embodiment of the present application, a storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the vehicle control processing method as described in the first aspect is implemented.

[0016] According to a fifth aspect of an embodiment of the present application, a computer program product is provided, comprising computer program instructions, which, when executed by a processor, cause the processor to execute: a vehicle control processing method as described in the first aspect.

[0017] The vehicle control processing method, device, equipment, medium and product provided in the embodiments of the present application can realize accurate classification of driving scenes by classifying driving scenes for received user requests, accurately identify the user's current driving environment, and thus predict the user's vehicle control intention. Afterwards, by obtaining the current state parameters of the vehicle control object related to the driving scene category, and combining the driving scene category and the current state parameters of the vehicle control object, control recommendation information for the vehicle control object can be generated in a targeted manner, thereby improving the intelligence and personalization of vehicle control. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0019] Figure 1 A flowchart of a vehicle control processing method provided in an embodiment of the present application;

[0020] Figure 2 A schematic diagram of classifying driving scenes according to user requests using a scene classification model provided in an embodiment of the present application;

[0021] Figure 3 A schematic diagram of the correspondence between driving scenario categories and vehicle control objects provided in an embodiment of the present application;

[0022] Figure 4 A flowchart of providing control recommendation information to a user in combination with preference control data provided in an embodiment of the present application;

[0023] Figure 5 A schematic diagram of the structure of a vehicle control processing device provided in an embodiment of the present application;

[0024] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The technical solution provided in the embodiments of the present application can be exemplarily applied to hardware devices such as processors, electronic devices, servers (including cloud servers), or packaged into software programs to be run. When the hardware device executes the processing of the technical solution of the embodiments of the present application, or the above-mentioned software program is run, the automatic splitting of the target task and the automatic calling of the application program interface required for the task can be achieved to complete the purpose of the target task. The embodiments of the present application only exemplarily introduce the specific processing of the technical solution of the present application, and do not limit the specific implementation form of the technical solution of the present application. Any technical implementation form that can execute the processing of the technical solution of the present application can be adopted by the embodiments of the present application.

[0026] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0027] Before introducing this application solution, the relevant technologies are first introduced:

[0028] In traditional car systems, vehicle status control mainly relies on the coordination of sensors and basic control logic. For example, when the rain sensor detects rainfall, it automatically triggers the vehicle control system to close the sunroof and windows to prevent rain from entering the car. Similarly, when the temperature sensor detects that the temperature inside the car exceeds the set threshold, the air-conditioning system will automatically adjust the temperature to ensure the comfort of the interior environment. In addition, the light sensor can adjust the interior lighting and ambient lighting to adapt to different lighting conditions during the day and at night; and the on-board noise recognition device can intelligently recommend adjusting the media volume by identifying the decibel level in the current environment. However, although these vehicle status control functions are automated, they rely too much on direct feedback from sensors, lack consideration of user personalized needs and sensitivity to complex environmental changes.

[0029] This type of automated control system can usually only perform simple, preset actions and cannot make complex decisions or provide dynamic suggestions, which limits the flexibility and adaptability of the system. With the advancement of technology, especially the breakthroughs of large language models represented by ChatGPT in natural language processing, computer vision and other fields, new possibilities have been provided for improving the intelligence of vehicle systems. By learning large amounts of data, these models can deeply understand and generate text, images and other content, enhancing the application potential of artificial intelligence in multiple industries. In the fields of autonomous driving, intelligent customer service, medical diagnosis, etc., the application of large models has demonstrated its advantages in handling complex tasks, such as environmental analysis and safety recommendations in autonomous driving systems, problem solving in intelligent customer service, and auxiliary decision-making in medical diagnosis.

[0030] In the field of smart cockpits, the integration of advanced technologies such as speech recognition, emotion recognition, and environmental perception can significantly improve driving experience and driving safety. For example, after integrating large language models, the in-vehicle system can not only better understand the needs of users, but also provide more personalized services and suggestions, such as adjusting the music playlist according to the driver's emotional state, or planning the itinerary in advance based on the weather forecast. With the continuous development of large language model technology, future in-vehicle systems will be more intelligent and able to provide a richer and more personalized user experience.

[0031] In view of this, the embodiments of the present application are committed to providing a vehicle control processing method, device, equipment, medium and product to improve the intelligence and personalization of vehicle status control and enhance user experience, which are described in detail one by one in the following embodiments.

[0032] Exemplary Methods

[0033] Figure 1 A flowchart of a vehicle control processing method provided in an embodiment of the present application. Figure 1As shown, the vehicle control processing method provided in this embodiment includes steps S101-S103:

[0034] S101: Classify the driving scene of the received user request to obtain a driving scene category.

[0035] The driving scene category represents the user's current driving environment or the cockpit environment required by the user.

[0036] In this embodiment, the user request refers to a query statement (query) input by the user through various channels (such as the vehicle system interface, voice assistant or mobile phone application), which contains the user's explicit or implicit intention for vehicle control.

[0037] User requests can help the vehicle system understand the user's vehicle control needs. In different driving scenarios, the user's control intentions may be different. Since vehicle control scenarios are highly diverse, in order to accurately predict the user's vehicle control intentions, it is necessary to classify the user's requests into driving scenarios to determine the user's current driving environment.

[0038] In some embodiments, the user request may be classified into driving scenarios through a scene classification model to determine the current driving environment of the user. Specifically, step S101 includes: inputting the received user request into the scene classification model, and classifying the received user request into driving scenarios through the scene classification model to obtain driving scenario categories; wherein the scene classification model is obtained by training a bidirectional encoder representation model according to user request samples under each driving scenario category and the corresponding annotated driving scenario categories.

[0039] The training process of the scene classification model aims to enable the model to accurately identify the user's current driving environment or required cockpit environment by learning from a large number of user request samples. The training process specifically includes the following steps a1 to a3:

[0040] Step a1: Obtain user request samples under each driving scene category and the corresponding annotated driving scene category.

[0041] In this embodiment, the user requests samples as training samples. The labeled driving scene categories may be manually labeled driving scene categories, which are used as labels for training samples.

[0042] For example, if the user request sample is "it's raining", the corresponding labeled driving scene category can be "rainy weather"; if the user request sample is "heavy haze", the corresponding labeled driving scene category is "polluted weather"; if the user request sample is "the weather is really good today", the corresponding labeled driving scene category is "clear weather".

[0043] The process of obtaining user request samples under each driving scenario category includes the following steps a11 to a13:

[0044] Step a11: For each driving scenario category in each driving scenario category, obtain an original user request sample under the driving scenario category.

[0045] The original user request samples can be obtained by manually constructing user query sentence samples. In order to ensure comprehensive coverage of each driving scene category, user query sentence samples can be constructed for each driving scene category.

[0046] Step a12: perform corpus generalization processing on the original user request samples under the driving scene category to obtain generalized user request samples under the driving scene category, where the generalized user request samples are different expressions of the original user request samples.

[0047] In order to improve the diversity of training samples and comprehensive coverage of driving scenarios, the GPT-4 API provided by OPenAI can be used to generalize the corpus of these original user request samples, thereby generating a wider range of variants of the original user request samples.

[0048] Corpus generalization processing includes synonym replacement, part-of-speech analysis, and grammatical analysis. Synonym replacement refers to replacing some nouns in the original user request sample with their synonyms, thereby expanding and enriching the training sample. When replacing synonyms, it is necessary to maintain the consistency of the part of speech so that the replaced text is logical and consistent with the original context. In addition, the replaced text also needs to be analyzed grammatically to ensure that its grammatical structure conforms to the language expression habits, and to ensure fluency and accuracy.

[0049] For example, the original user request sample is "turn on the air conditioner". Through corpus generalization processing, the generalized user request samples can be obtained as "turn on the air conditioner", "turn on the air conditioner" and other different expressions; the original user request sample is "I want to be quiet for a while". Through corpus generalization processing, the generalized user request samples can be obtained as "be quiet", "I need a quiet environment", "don't be so noisy", "be quiet", "too noisy" and other different expressions; the original user request sample is "I want to take a break". Through corpus generalization processing, the generalized user request samples can be obtained as "I want to sleep for a while", "a little sleepy", "take a nap" and other different expressions; the original user request sample is "I want to relax". Through corpus generalization processing, the generalized user request samples can be obtained as "a little annoying", "let's create a happy atmosphere", "I want to create a happy environment", "I want to create a relaxed and pleasant environment" and other different expressions; the original user request sample is "I feel so cold". Through corpus generalization processing, the generalized user request samples can be obtained as "the cold wind makes me uncomfortable", "why is the air conditioner so cold" and other different expressions.

[0050] Step a13: obtaining user request samples under the driving scenario category according to the original user request samples and the generalized user request samples under the driving scenario category.

[0051] By merging the original user request samples and the generalized user request samples under each driving scenario category, the user request samples under each driving scenario category can be obtained.

[0052] After step a1, the training process of the scene classification model may further include step a2 and step a3. The implementation of step a2 and step a3 is as follows:

[0053] Step a2: Obtain a predicted driving scene category based on user request samples under each driving scene category through the scene classification model to be trained.

[0054] The scene classification model to be trained can be a bidirectional encoder representation model (Bidirectional Encoder Representations from Transformers, BERT), and its structure can be a 3*384 BERT. During the training process, a tokenizer is used to preprocess the text corresponding to the user request sample, such as converting the text into a format that the BERT model can understand, including adding special symbols (such as [CLS] and [SEP]) and converting it into a Token ID. After that, the pre-trained BERT model is loaded, and a top layer for classification is added and training parameters are set. Next, the model is trained using user request samples. The input of the model is the user request sample, and the output is the predicted driving scene category. Specifically, the user request samples under each driving scene category are input into the scene classification model to be trained, so that it can predict the corresponding driving scene category based on the user request samples under each driving scene category, thereby obtaining the predicted driving scene category.

[0055] Step a3: Determine the driving scene prediction loss based on the difference between the predicted driving scene category and the labeled driving scene category, and converge the scene classification model to be trained based on the driving scene prediction loss to obtain the scene classification model.

[0056] By calculating the difference between the predicted driving scene category and the labeled driving scene category, the prediction error of the model can be quantified. This quantified error is the "driving scene prediction loss". This loss value not only reflects the prediction ability of the model in the current state, but also serves as an important basis for guiding the subsequent learning direction of the model.

[0057] In order to minimize this prediction loss, optimization algorithms such as gradient descent can be used to gradually adjust the parameters in the model so that the model can better fit the training samples. As the training process continues, the driving scene prediction loss will gradually decrease, indicating that the model's accuracy in classifying driving scenes is improving. When the model's classification accuracy reaches the preset accuracy or the change in the loss value tends to be stable, the model is considered to have converged, and the model obtained at this time is an optimized trained scene classification model.

[0058] After the scene classification model is trained, the scene classification model can be applied to classify the driving scenes of user requests so as to classify them into different driving scene categories.

[0059] Figure 2 A schematic diagram of a driving scene classification model according to a user request provided in an embodiment of the present application. Figure 2As shown, when the user request is "it's raining", it can be classified into the driving scene category of "rainy weather"; the user request is "there is a lot of haze", it can be classified into the driving scene category of "polluted weather"; the user request is "the weather is really good today", it can be classified into the driving scene category of "clear weather".

[0060] The user request in this embodiment refers to human-computer dialogue. In the cockpit environment, there may be two types of dialogues: human-human dialogue and human-computer dialogue. Therefore, after receiving the original user request, it is necessary to identify the original user request to distinguish its type and determine the subsequent processing method.

[0061] The user request can be in the form of voice input or text input. For a user request input by voice, speech recognition technology (such as automatic speech recognition, ASR) can be used to transcribe the speech content into text, and its intention can be analyzed through natural language processing (NLP) technology.

[0062] For text requests input through a touch screen or other input devices, text analysis can be performed directly to determine whether it belongs to human-computer dialogue.

[0063] In some embodiments, a rejection model can be used to identify a human-computer dialogue or a human-human dialogue. The rejection model is a model specially designed to distinguish valid requests from invalid requests, and can effectively identify the type of user requests.

[0064] The rejection model can be trained with a large amount of labeled human-to-human and human-to-computer dialogue data. The training data includes: human-to-human dialogue samples, that is, conversations between passengers, such as "How are you today?"; and human-to-computer dialogue samples, that is, requests made by users to the in-vehicle system, such as "Turn on the air conditioner" or "Navigate home".

[0065] The received original user request is input into the rejection model, and the rejection model outputs a classification result indicating whether the request is a human-computer dialogue or a human-human dialogue.

[0066] If the rejection model identifies the original user request as a person-to-person conversation, the system will ignore the request and not process it further. This type of conversation is usually not related to vehicle control, so no further analysis or control recommendation information generation is required. If the rejection model identifies the original user request as a user request to the vehicle system, the system will continue with the subsequent processing steps, namely the steps of driving scenario classification, obtaining current state parameters, and generating control recommendation information.

[0067] Continue reading Figure 1 The vehicle control processing method of this embodiment may further include step S102.

[0068] S102: Acquire current state parameters of vehicle control objects related to the driving scenario category.

[0069] After the driving scenario category is determined, the vehicle control object corresponding to the driving scenario category can be determined based on the driving scenario category and the corresponding relationship between the driving scenario category and the vehicle control object, and the current state parameters of the vehicle control object corresponding to the driving scenario category can be obtained.

[0070] Figure 3 Schematic diagram of the corresponding relationship between driving scene categories and vehicle control objects provided in the embodiment of the present application. Figure 3 As shown, when the driving scene category is "rainy weather", the corresponding vehicle control objects include weather, sunroof, car windows and air conditioning; when the driving scene category is "polluted weather", the corresponding vehicle control objects include weather, sunroof, car windows and air conditioning; when the driving scene category is "snowy weather", the corresponding vehicle control objects include weather, sunroof, car windows, air conditioning and air conditioning heating, etc.; when the driving scene category is "clear weather", the corresponding vehicle control objects include weather, the user's geographical location, etc.

[0071] After the vehicle control object is determined, the current state parameters of the vehicle control object may be further determined so as to provide intelligent vehicle control suggestions to the user.

[0072] S103: Determine control recommendation information for the vehicle control object according to the driving scene category and the current state parameters of the vehicle control object.

[0073] In this embodiment, the control recommendation information for the vehicle control object refers to suggestions for controlling the vehicle control object, and is output by voice or displayed on a vehicle-mounted screen.

[0074] In some embodiments, step S103 can be implemented by a control recommendation model. Specifically, step S103 includes: inputting the driving scene category and the current state parameters of the vehicle control object into the control recommendation model, and determining the control recommendation information for the vehicle control object according to the driving scene category and the current state parameters of the vehicle control object through the control recommendation model; wherein the control recommendation model is obtained by fine-tuning the large model according to the state parameter samples of the vehicle control object under each driving scene category and the corresponding annotated control recommendation information.

[0075] The training process of the control recommendation model includes the following steps b1 to b3:

[0076] Step b1: Obtain state parameter samples of vehicle control objects under each driving scenario category and corresponding annotated control recommendation information.

[0077] In this embodiment, the state parameter samples of the vehicle control object are used as training samples. The labeled control recommendation information can be an artificially constructed response sentence used as a label of the training sample.

[0078] For example, when the driving scenario category is "snowy weather" and the state parameter sample of the vehicle control object is "air conditioning and heating are off", the labeling control recommendation information can be a query sentence of "do you need to turn on the air conditioning and heating?"; when the driving scenario category is "rainy weather" and the state parameter sample of the vehicle control object is "air conditioning and heating are off, the sunroof is closed and the windows are open", the labeling control recommendation information can be "do you need to turn on the air conditioning and heating and close the windows?"

[0079] The process of obtaining the labeling control recommendation information under each driving scene category includes the following steps b11 to b13:

[0080] Step b11: for each driving scenario category in each driving scenario category, original control recommendation information corresponding to the state parameter sample of the vehicle control object under the driving scenario category is obtained.

[0081] The original control recommendation information can be obtained by manually constructing a response statement. In order to ensure comprehensive coverage of each driving scenario category, the original control recommendation information can be constructed for each driving scenario category.

[0082] Step b12: performing corpus generalization processing on the original control recommendation information under the driving scenario category to obtain generalized control recommendation information under the driving scenario category, where the generalized control recommendation information is a different representation of the original control recommendation information.

[0083] In order to improve the diversity of training samples and comprehensive coverage of driving scenarios, the original control recommendation information can be generalized through the GPT-4 API provided by OPenAI to generate a wider range of variants of the original control recommendation information.

[0084] Among them, corpus generalization processing includes synonym replacement, part-of-speech analysis and grammatical analysis. Synonym replacement refers to replacing some nouns in the original control recommendation information with their synonyms, thereby expanding and enriching the training samples. When replacing synonyms, it is necessary to maintain the consistency of part-of-speech so that the replaced text is logical and consistent with the original context. In addition, the replaced text needs to be analyzed grammatically to ensure that its grammatical structure conforms to the language expression habits, and to ensure fluency and accuracy.

[0085] Step b13: Obtain the labeled control recommendation information corresponding to the state parameter sample of the vehicle control object under the driving scenario category according to the original control recommendation information and the generalized control recommendation information under the driving scenario category.

[0086] By merging the original control recommendation information and the generalized control recommendation information under each driving scenario category, the labeled control recommendation information under each driving scenario category can be obtained.

[0087] After step b1, the training process of controlling the recommendation model may further include steps b2 and b3. The implementation of steps b2 and b3 is as follows:

[0088] Step b2: obtaining predictive control recommendation information based on the state parameter samples of the vehicle control object under each driving scenario category through the control recommendation model to be trained.

[0089] The control recommendation model to be trained may be a large language model. The state parameter samples of the vehicle control objects under each driving scenario category are input into the control recommendation model to be trained so that the control recommendation information corresponding to the state parameter samples of the vehicle control objects under each driving scenario category can be predicted, and the predicted control recommendation information can be obtained.

[0090] During the training process, it is necessary to construct prompts based on the state parameter samples of the vehicle control objects under each driving scenario category. The state parameter samples of the vehicle control objects under each driving scenario category are used as training corpus, and the annotated control recommendation information is used as the answer during supervised fine-tuning (SFT), thereby obtaining several<prompt,answer> The data pairs are used as the instruction following dataset. Then, the instruction following dataset is used to perform instruction tuning on the Large Language Model (LLM) in a supervised manner to obtain a supervised fine-tuning model, i.e., a control recommendation model. The large language model can be Qwen0.5B, MobiLlama 0.5B, etc.

[0091] The instruction following data set can be as follows:

[0092] Example 1:

[0093] Prompt words: Assume that you are an intelligent in-vehicle assistant, and give the user control recommendation information based on the current state parameter sample of the vehicle. If you cannot determine the control recommendation information, please answer: I did not understand the user's intention; Current vehicle state parameters: Weather information: heavy rain; Air conditioning status: off; Sunroof status: open; Window status: closed; Please reply to the user's vehicle control operation suggestions;

[0094] Reply: Do you need me to turn on the air conditioner and close the skylight?

[0095] Example 2:

[0096] Prompt words: Assume that you are an intelligent in-vehicle assistant, and give the user control recommendation information based on the current state parameter sample of the vehicle. If the control recommendation information cannot be determined, please answer: I did not understand the user's intention; Current vehicle state parameters: Weather information: Heavy rain; Air conditioning status: Off; Sunroof status: Open; Window status: Unknown; Please reply to the user's vehicle control operation suggestions;

[0097] Reply: Do you need me to turn on the air conditioner and close the sunroof?

[0098] Example 3:

[0099] Prompt words: Assume that you are an intelligent in-vehicle assistant, and give the user control recommendation information based on the current state parameter sample of the vehicle. If the control recommendation information cannot be determined, please answer: I did not understand the user's intention; Current vehicle state parameters: Weather information: fog; Air conditioning state: off; Sunroof state: open; Window state: unknown; Please reply to the user's vehicle control operation suggestions;

[0100] Reply: Do you need me to turn on the air conditioner, close the skylight, and turn on the air conditioner's purification mode?

[0101] Step b3: determine the control recommendation prediction loss according to the difference between the predicted control recommendation information and the labeled control recommendation information, and converge the control recommendation model to be trained based on the control recommendation prediction loss to obtain the control recommendation model.

[0102] By calculating the difference between the predicted control recommendation information and the labeled control recommendation information, the prediction error of the model can be quantified. This quantified error is the "control recommendation prediction loss". This loss value not only reflects the prediction ability of the model in the current state, but also serves as an important basis for guiding the subsequent learning direction of the model.

[0103] In order to minimize this prediction loss, optimization algorithms such as gradient descent can be used to gradually adjust the parameters in the model so that the model can better fit the training samples. As the training process continues, the control recommendation prediction loss will gradually decrease, indicating that the accuracy of the control recommendation information generated by the model is improving. When the control recommendation accuracy of the model reaches the preset accuracy or the change in the loss value tends to be stable, the model is considered to have converged, and the model obtained at this time is the control recommendation model that has been optimized and trained.

[0104] After the control recommendation model is trained, control recommendation information can be generated for users through the control recommendation model.

[0105] When generating control recommendation information for users through the control recommendation model, in order to improve the personalization of vehicle control functions, the preferred control data of the vehicle can also be customized for the user, and the control recommendation information can be provided to the user in combination with the preferred control data. The details are as follows:

[0106] Figure 4 A flowchart of providing control recommendation information to a user in combination with preference control data provided in an embodiment of the present application. Figure 4 As shown, step S103 may also include step S401 and step S402:

[0107] S401. Obtain user preference control data for vehicle control objects.

[0108] Among them, the user's preference control data for the vehicle control object refers to a series of parameters or options set by the user according to personal preferences or habits when using the vehicle control system. These data reflect the user's personalized needs for vehicle control functions.

[0109] Preference control data usually includes various parameters set by users in the vehicle control system, such as the degree of refinement of seat adjustment, preference for air conditioning temperature, sound effect style of audio settings, and driving mode selection tendency, etc. These data reflect the user's personalized preferences.

[0110] S402: Determine control recommendation information for the vehicle control object according to the driving scene category, the current state parameters of the vehicle control object, and the preferred control data.

[0111] Taking the rainy driving scenario as an example, different users have different vehicle control requirements when driving in rainy days. Some users think that they only need to close the sunroof and windows on rainy days, and do not need to turn on the air conditioner. However, some users think that they need to close the sunroof and windows on rainy days, and turn on the air conditioner to keep the interior environment comfortable and dry. Therefore, when generating control recommendation information for different users, it is necessary to refer to the user's personal preference control data to ensure that the recommended information provided can accurately meet the user's personalized needs and achieve personalized responses.

[0112] In summary, this embodiment intelligently recommends corresponding vehicle control operation suggestions by understanding the user's natural language request (query) and combining it with the current vehicle device status parameters. The large model can understand the user's needs by combining the driving scene category and the current status parameters of the vehicle control object, and predict the user's next vehicle control behavior, thereby automatically generating vehicle control operation suggestions for the user to provide a more intelligent vehicle control function.

[0113] Partial examples of user requests, driving scenario categories, vehicle control objects, and control recommendation information may be shown in Table 1 below:

[0114] Table 1 Examples of user requests, driving scenario categories, vehicle control objects, and control recommendation information

[0115]

[0116]

[0117] It should be noted that the user request, driving scenario category, vehicle control object and control recommendation information in Table 1 above are all exemplary descriptions and do not limit this embodiment. The user can add driving scenario categories according to actual needs.

[0118] In addition, by analyzing the user's current driving environment data, it can provide in-car environment setting suggestions in different driving scenario categories, such as automatically adjusting the in-car temperature according to weather changes, or adjusting the brightness and color of the in-car lights when driving at night to improve driving safety and comfort. This kind of intelligent vehicle control recommendation can not only enhance the interactive experience of the smart cockpit, but also improve driving safety and convenience.

[0119] In addition, by learning the user's preferred controls and automatically adjusting vehicle control operation suggestions in subsequent interactions, it can deeply understand user needs and environmental changes, provide more intelligent and personalized vehicle control functions, and further improve driving convenience.

[0120] Exemplary Devices

[0121] Corresponding to the above-mentioned vehicle control processing method, an embodiment of the present application also provides a vehicle control processing device. Figure 5 Schematic diagram of a vehicle control processing device provided in an embodiment of the present application. Figure 5 As shown, the vehicle control processing device provided in the embodiment of the present application includes: a classification unit 501, an acquisition unit 502 and a determination unit 503; wherein the classification unit 501 is used to classify the driving scene for the received user request to obtain a driving scene category, and the driving scene category represents the user's current driving environment; the acquisition unit 502 is used to obtain the current state parameters of the vehicle control object related to the driving scene category; the determination unit 503 is used to determine the control recommendation information for the vehicle control object according to the driving scene category and the current state parameters of the vehicle control object.

[0122] In some embodiments, the classification unit 501 performs driving scene classification on the received user request to obtain a driving scene category, including: inputting the received user request into a scene classification model, and performing driving scene classification on the received user request through the scene classification model to obtain a driving scene category; wherein the scene classification model is obtained by training a bidirectional encoder representation model based on user request samples under each driving scene category and the corresponding labeled driving scene category.

[0123] In some embodiments, the user request samples under each driving scenario category are obtained by the following steps: for each driving scenario category in the various driving scenario categories, an original user request sample under the driving scenario category is obtained; the original user request samples under the driving scenario category are subjected to corpus generalization processing to obtain generalized user request samples under the driving scenario category, wherein the generalized user request samples are different expressions of the original user request samples; and the user request samples under the driving scenario category are obtained according to the original user request samples and the generalized user request samples under the driving scenario category.

[0124] In some embodiments, the acquisition unit 502 acquires current state parameters of the vehicle control object related to the driving scenario category, including: determining the vehicle control object corresponding to the driving scenario category based on the driving scenario category and the correspondence between the driving scenario category and the vehicle control object; acquiring the current state parameters of the vehicle control object corresponding to the driving scenario category.

[0125] In some embodiments, the determination unit 503 determines the control recommendation information for the vehicle control object based on the driving scenario category and the current state parameters of the vehicle control object, including: inputting the driving scenario category and the current state parameters of the vehicle control object into a control recommendation model, and determining the control recommendation information for the vehicle control object based on the driving scenario category and the current state parameters of the vehicle control object through the control recommendation model; wherein the control recommendation model is obtained by fine-tuning a large model based on state parameter samples of the vehicle control object under each driving scenario category and the corresponding annotated control recommendation information.

[0126] In some embodiments, the annotated control recommendation information is obtained by the following steps: for each driving scenario category in the various driving scenario categories, obtaining the original control recommendation information corresponding to the state parameter sample of the vehicle control object under the driving scenario category; performing corpus generalization processing on the original control recommendation information under the driving scenario category to obtain generalized control recommendation information under the driving scenario category, wherein the generalized control recommendation information is a different representation of the original control recommendation information; and obtaining the annotated control recommendation information corresponding to the state parameter sample of the vehicle control object under the driving scenario category based on the original control recommendation information and the generalized control recommendation information under the driving scenario category.

[0127] In some embodiments, the determination unit 503 determines the control recommendation information for the vehicle control object based on the driving scenario category and the current state parameters of the vehicle control object, including: obtaining the user's preference control data for the vehicle control object; determining the control recommendation information for the vehicle control object based on the driving scenario category, the current state parameters of the vehicle control object and the preference control data.

[0128] The vehicle control processing device provided in this embodiment belongs to the same application concept as the vehicle control processing method provided in the above embodiments of this application, and can execute the vehicle control processing method provided in any of the above embodiments of this application, and has the corresponding functional modules and beneficial effects of executing the vehicle control processing method. For technical details not fully described in this embodiment, please refer to the specific processing content of the vehicle control processing method provided in the above embodiments of this application, and will not be repeated here.

[0129] The functions implemented by the above classification unit 501, acquisition unit 502 and determination unit 503 may be implemented by the same or different processors respectively, which is not limited in the embodiment of the present application.

[0130] It should be understood that the units in the above devices can be implemented in the form of a processor calling software. For example, the device includes a processor, the processor is connected to a memory, and instructions are stored in the memory. The processor calls the instructions stored in the memory to implement any of the above methods or realize the functions of each unit of the device, wherein the processor can be a general-purpose processor, such as a CPU or a microprocessor, etc., and the memory can be a memory in the device or a memory outside the device. Alternatively, the units in the device can be implemented in the form of hardware circuits, and the functions of some or all units can be realized by designing the hardware circuits. The hardware circuit can be understood as one or more processors; for example, in one implementation, the hardware circuit is an ASIC, and the functions of some or all of the above units are realized by designing the logical relationship of the components in the circuit; for another example, in another implementation, the hardware circuit can be implemented by PLD, taking FPGA as an example, which can include a large number of logic gate circuits, and the connection relationship between the logic gate circuits is configured by the configuration file, so as to realize the functions of some or all of the above units. All units of the above devices can be implemented in the form of a processor calling software, or in the form of hardware circuits, or in part by a processor calling software, and the remaining part is implemented in the form of hardware circuits.

[0131] In an embodiment of the present application, a processor is a circuit with the ability to process signals. In one implementation, the processor may be a circuit with the ability to read and run instructions, such as a CPU, a microprocessor, a GPU, or a DSP; in another implementation, the processor may implement certain functions through the logical relationship of a hardware circuit, and the logical relationship of the hardware circuit is fixed or reconfigurable, such as a hardware circuit implemented by an ASIC or PLD, such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document to implement the hardware circuit configuration can be understood as the process of the processor loading instructions to implement the functions of some or all of the above units. In addition, it can also be a hardware circuit designed for artificial intelligence, which can be understood as an ASIC, such as an NPU, TPU, DPU, etc.

[0132] It can be seen that each unit in the above device can be one or more processors (or processing circuits) configured to implement the above method, such as: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms.

[0133] In addition, all or part of the units in the above device can be integrated together, or can be implemented independently. In one implementation, these units are integrated together and implemented in the form of a SOC. The SOC may include at least one processor for implementing any of the above methods or implementing the functions of each unit of the device. The type of the at least one processor may be different, for example, including a CPU and an FPGA, a CPU and an artificial intelligence processor, a CPU and a GPU, etc.

[0134] Exemplary Electronic Devices

[0135] The present application embodiment provides an electronic device, see Figure 6 As shown, the electronic device includes:

[0136] Memory 200 and processor 210;

[0137] The memory 200 is connected to the processor 210 and is used to store programs;

[0138] The processor 210 is used to implement the vehicle control processing method disclosed in any of the above embodiments by running the program stored in the memory 200.

[0139] Specifically, the electronic device may further include: a bus, a communication interface 220 , an input device 230 and an output device 240 .

[0140] The processor 210, the memory 200, the communication interface 220, the input device 230 and the output device 240 are connected to each other via a bus.

[0141] A bus may include a pathway that transfers information between components of a computer system.

[0142] The processor 210 may be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the scheme of the present invention. It may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0143] The processor 210 may include a main processor, and may also include a baseband chip, a modem, and the like.

[0144] The memory 200 stores a program for executing the technical solution of the present invention, and may also store an operating system and other key services. Specifically, the program may include a program code, and the program code includes a computer operation instruction. More specifically, the memory 200 may include a read-only memory (ROM), other types of static storage devices that can store static information and instructions, a random access memory (RAM), other types of dynamic storage devices that can store information and instructions, a disk storage, a flash, and the like.

[0145] The input device 230 may include a device for receiving data and information input by a user, such as a keyboard, a mouse, a camera, a scanner, a light pen, a voice input device, a touch screen, a pedometer, or a gravity sensor.

[0146] Output device 240 may include devices that allow information to be output to a user, such as a display screen, printer, speaker, etc.

[0147] The communication interface 220 may include any transceiver or the like to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0148] The processor 210 executes the program stored in the memory 200 and calls other devices, which can be used to implement each step of any vehicle control processing method provided in the above embodiments of the present application.

[0149] An embodiment of the present application also proposes a chip, which includes a processor and a data interface. The processor reads and runs a program stored in a memory through the data interface to execute the vehicle control processing method introduced in any of the above embodiments. The specific processing process and its beneficial effects can be found in the embodiment introduction of the above-mentioned vehicle control processing method.

[0150] Exemplary computer program products and storage media

[0151] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps in the vehicle control processing method according to various embodiments of the present application described in any of the above embodiments of this specification.

[0152] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0153] In addition, the embodiment of the present application may also be a storage medium on which a computer program is stored. The computer program is executed by a processor to execute the steps of the vehicle control processing method according to various embodiments of the present application described in any of the above embodiments of this specification, and specifically the following steps may be implemented:

[0154] Step S101: classify the driving scene of the received user request to obtain a driving scene category, where the driving scene category represents the current driving environment of the user;

[0155] Step S102, obtaining current state parameters of vehicle control objects related to the driving scene category;

[0156] Step S103: Determine control recommendation information for the vehicle control object according to the driving scene category and the current state parameter of the vehicle control object.

[0157] For the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0158] It should be noted that each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the embodiments can be referred to each other. For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0159] The steps in the methods of each embodiment of the present application can be adjusted in order, combined and deleted according to actual needs, and the technical features recorded in each embodiment can be replaced or combined.

[0160] The modules and sub-modules in the devices and terminals of the various embodiments of the present application can be combined, divided and deleted according to actual needs.

[0161] In the several embodiments provided in the present application, it should be understood that the disclosed terminals, devices and methods can be implemented in other ways. For example, the terminal embodiments described above are only schematic, for example, the division of modules or submodules is only a logical function division, and there may be other division methods in actual implementation, for example, multiple submodules or modules can be combined or integrated into another module, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.

[0162] The modules or submodules described as separate components may or may not be physically separated, and the components of the modules or submodules may or may not be physical modules or submodules, that is, they may be located in one place, or they may be distributed on multiple network modules or submodules. Some or all of the modules or submodules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0163] In addition, each functional module or submodule in each embodiment of the present application may be integrated into one processing module, or each module or submodule may exist physically separately, or two or more modules or submodules may be integrated into one module. The above-mentioned integrated modules or submodules may be implemented in the form of hardware or in the form of software functional modules or submodules.

[0164] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0165] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly by hardware, software units executed by a processor, or a combination of the two. The software units may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0166] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0167] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A vehicle control processing method, characterized in that: include: Classifying the driving scene of the received user request to obtain a driving scene category, where the driving scene category represents the current driving environment of the user; Acquire current state parameters of vehicle control objects related to the driving scenario category; According to the driving scene category and the current state parameter of the vehicle control object, control recommendation information for the vehicle control object is determined.

2. The method according to claim 1, characterized in that The driving scene classification of the received user request to obtain the driving scene category includes: Inputting the received user request into a scene classification model, and classifying the received user request into a driving scene by using the scene classification model to obtain a driving scene category; The scene classification model is obtained by training a bidirectional encoder representation model according to user request samples under each driving scene category and the corresponding annotated driving scene category.

3. The method according to claim 2, characterized in that The user request samples under each driving scenario category are obtained by the following steps: For each driving scenario category in the driving scenario categories, obtaining an original user request sample under the driving scenario category; Performing corpus generalization processing on the original user request sample under the driving scene category to obtain a generalized user request sample under the driving scene category, wherein the generalized user request sample is a different expression of the original user request sample; According to the original user request samples and the generalized user request samples under the driving scenario category, the user request samples under the driving scenario category are obtained.

4. The method according to any one of claims 1 to 3, characterized in that The obtaining of current state parameters of the vehicle control object related to the driving scene category includes: Determining the vehicle control object corresponding to the driving scene category according to the driving scene category and the corresponding relationship between the driving scene category and the vehicle control object; Obtain current state parameters of the vehicle control object corresponding to the driving scenario category.

5. The method according to any one of claims 1 to 3, characterized in that: The determining, according to the driving scene category and the current state parameter of the vehicle control object, control recommendation information for the vehicle control object comprises: Inputting the driving scenario category and the current state parameters of the vehicle control object into a control recommendation model, and determining control recommendation information for the vehicle control object according to the driving scenario category and the current state parameters of the vehicle control object through the control recommendation model; The control recommendation model is obtained by fine-tuning and training a large model according to state parameter samples of vehicle control objects under various driving scenario categories and corresponding annotated control recommendation information.

6. The method according to claim 5, characterized in that The labeling control recommendation information is obtained by the following steps: For each driving scenario category in the driving scenario categories, obtaining original control recommendation information corresponding to the state parameter sample of the vehicle control object under the driving scenario category; Performing corpus generalization processing on the original control recommendation information under the driving scenario category to obtain generalized control recommendation information under the driving scenario category, wherein the generalized control recommendation information is a different representation of the original control recommendation information; According to the original control recommendation information and the generalized control recommendation information under the driving scene category, the labeled control recommendation information corresponding to the state parameter sample of the vehicle control object under the driving scene category is obtained.

7. The method according to any one of claims 1 to 3, characterized in that The determining, according to the driving scene category and the current state parameter of the vehicle control object, control recommendation information for the vehicle control object comprises: Acquiring user preference control data for the vehicle control object; According to the driving scene category, the current state parameter of the vehicle control object and the preference control data, control recommendation information for the vehicle control object is determined.

8. A vehicle control processing device, characterized in that: include: A classification unit, configured to classify a driving scene according to a received user request to obtain a driving scene category, wherein the driving scene category represents a current driving environment of the user; an acquisition unit, used to acquire current state parameters of a vehicle control object related to the driving scene category; A determination unit is used to determine control recommendation information for the vehicle control object according to the driving scene category and the current state parameter of the vehicle control object.

9. An electronic device, characterized in that: including memory and processor; The memory is connected to the processor and is used to store programs; The processor is configured to implement the method according to any one of claims 1 to 7 by running the program in the memory.

10. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, the method according to any one of claims 1 to 7 is implemented.

11. A computer program product, characterized in that The method comprises computer program instructions, which, when executed by a processor, cause the processor to implement the method as claimed in any one of claims 1 to 7.

Images