Vehicle scene service orchestration method, device, equipment, medium and program product

Abstract

The invention relates to a vehicle scene service orchestration method and device, equipment, a medium and a program product. The vehicle scene service orchestration method comprises the following steps: determining the definition of an intention of voice information input by a user; based on the definition of the intention, determining a vehicle atomization service matched with the voice information; and configuring the vehicle atomization service based on the behavior habits of the user to obtain the currently required vehicle scene service. Therefore, not only is the vehicle scene service formulated by voice supported, but also personalized customization of the vehicle scene service can be realized, and a practical and comfortable vehicle scene service is created.

Description

Technical Field

[0001] The present application relates to the field of automobile intelligent control technology, and specifically to a vehicle scenario service orchestration method, device, equipment, medium, and program product. Background Art

[0002] Vehicle scenario services refer to specific user actions performed when using a vehicle in a specific context. For example, activating voice navigation, playing music, or turning on air conditioning while driving in the summer. Currently, the vehicle's service orchestration functionality requires users to manually drag, pull, or select the required services. However, because the visual interface cannot display all of the vehicle's atomic services at once, users lack clarity about which atomic services are available, leading to high barriers to entry and time-consuming operations. Summary of the Invention

[0003] In view of this, the present application provides at least one vehicle scenario service orchestration method, apparatus, device, medium, and program product.

[0004] The technical solution of the embodiment of the present application is implemented as follows:

[0005] On the one hand, an embodiment of the present application provides a vehicle scenario service orchestration method, which includes: determining the clarity of the intention of the voice information input by the user; based on the clarity of the intention, determining the vehicle atomic service that matches the voice information; configuring the vehicle atomic service based on the user's behavioral habits to obtain the currently required vehicle scenario service.

[0006] In some embodiments, the above-mentioned determination of the clarity of the intention of the voice information input by the user includes: obtaining the voice information input by the user; converting the voice information based on the user's voice characteristics to obtain text information corresponding to the voice information; the voice characteristics include: accent, speaking speed, and vocal habits; based on the text information, determining the clarity of the intention of the voice information.

[0007] In some embodiments, the above-mentioned determination of the clarity of the intention of voice information based on text information includes: inputting the text information into multiple feature extraction modules in the intention classification model in sequence, and outputting the text features corresponding to each feature extraction module; performing feature fusion on the text features corresponding to the multiple feature extraction modules to obtain fused text features; inputting the fused text features into the classification module in the intention classification model, and outputting the classification results; the classification results represent the clarity of the intention of the text information.

[0008] In some embodiments, the above-mentioned determination of the vehicle atomic service that matches the voice information based on the clarity of the intention includes: when the intention is clear, the text information corresponding to the voice information is input into the first service orchestration model, and the vehicle atomic service that matches the text information is output; when the intention is unclear, the text information corresponding to the voice information is input into the second service orchestration model, and the vehicle atomic service that matches the text information is output.

[0009] In some embodiments, the above-mentioned first service orchestration model is trained through the following steps: screening out a first vehicle scenario service with clear intention from the historical vehicle scenario services orchestrated before the current moment; preprocessing the first vehicle scenario service to obtain a processed first vehicle scenario service; training a first fine-tuning model in the attention layer of the base model based on the processed first vehicle scenario service to obtain a trained first fine-tuning model; and determining the first service orchestration model based on the trained first fine-tuning model and the base model.

[0010] In some embodiments, the above-mentioned preprocessing of the first vehicle scenario service includes at least one of the following: data cleaning of the first vehicle scenario service; anomaly removal of the first vehicle scenario service; and format conversion of the first vehicle scenario service.

[0011] In some embodiments, the above-mentioned training of the first fine-tuning model in the attention layer of the base model based on the processed first vehicle scenario service to obtain a trained first fine-tuning model includes: inputting the processed first vehicle scenario service into the first fine-tuning model, and outputting the processing result of the first fine-tuning model; based on the output result of the first fine-tuning model, determining the overall weight matrix of the first fine-tuning model and the base model, and the incremental matrix of the first fine-tuning model; based on the overall weight matrix and the incremental matrix, adjusting the parameter matrix of the first fine-tuning model until the iteration stop condition is reached, thereby obtaining a trained first fine-tuning model.

[0012] In some embodiments, the above-mentioned second service orchestration model is trained through the following steps: screening out second vehicle scenario services with unclear intentions from historical vehicle scenario services orchestrated before the current moment; preprocessing the second vehicle scenario service to obtain a processed second vehicle scenario service; training a second fine-tuning model in the attention layer of the base model based on the processed second vehicle scenario service to obtain a trained second fine-tuning model; and determining the second service orchestration model based on the trained second fine-tuning model and the base model.

[0013] In some embodiments, the above-mentioned configuration of the vehicle atomic service based on the user's behavioral habits to obtain the currently required vehicle scenario service includes: obtaining the user's common parameters for the vehicle atomic service from the service parameter management system based on the user's identification and the functional category of the vehicle atomic service; the service parameter management system is updated in real time based on changes in the user's behavior and changes in the vehicle's environment; common parameters represent the user's behavioral habits; and the vehicle atomic service is configured based on the common parameters to obtain the currently required vehicle scenario service.

[0014] In some embodiments, the vehicle scenario service orchestration method also includes: performing safety detection on the vehicle scenario service; performing conflict detection on the vehicle scenario service; using safety detection and conflict detection to determine the executability of the vehicle scenario service; and when both safety detection and conflict detection indicate that there are no abnormalities in the vehicle scenario service, calling the application programming interface of the vehicle scenario service to execute the vehicle scenario service.

[0015] On the other hand, an embodiment of the present application provides a vehicle scenario service orchestration device, which includes: a determination module for determining the clarity of the intention of the voice information input by the user; a processing module for determining the vehicle atomic service that matches the voice information based on the clarity of the intention; the processing module is also used to configure the vehicle atomic service based on the user's behavioral habits to obtain the currently required vehicle scenario service.

[0016] On the other hand, an embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0017] On the other hand, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which implements some or all of the steps in the above method when executed by a processor.

[0018] On the other hand, an embodiment of the present application provides a computer program, including computer-readable code. When the computer-readable code is executed in a computer device, a processor in the computer device executes some or all of the steps for implementing the above method.

[0019] On the other hand, an embodiment of the present application provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program. When the computer program is read and executed by a computer, some or all of the steps in the above method are implemented.

[0020] In an embodiment of the present application, the clarity of the intention of the voice information input by the user is determined; based on the clarity of the intention, the vehicle atomic service that matches the voice information is determined; based on the user's behavioral habits, the vehicle atomic service is configured to obtain the currently required vehicle scene service. In this way, based on the clarity of the intention of the voice information input by the user, the vehicle atomic service that matches the voice information is determined, and the conversion of the user's voice to the vehicle atomic service can be intelligently realized; based on the user's behavioral habits, the vehicle atomic service is configured to realize personalized customization of the vehicle scene service. Thus, it not only supports the use of voice to formulate vehicle scene services, but also realizes personalized customization of vehicle scene services, creating practical and comfortable vehicle scene services for users. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Schematic diagram of the implementation process of a vehicle scenario service arrangement method provided in an embodiment of the present application Figure 1 ;

[0022] Figure 2 Schematic diagram of the implementation process of a vehicle scenario service arrangement method provided in an embodiment of the present application Figure 2 ;

[0023] Figure 3 Schematic diagram of the implementation process of a vehicle scenario service arrangement method provided in an embodiment of the present application Figure 3 ;

[0024] Figure 4 Schematic diagram of the implementation process of a vehicle scenario service arrangement method provided in an embodiment of the present application Figure 4 ;

[0025] Figure 5 A schematic diagram of the implementation process of the orchestration model in a vehicle scenario service orchestration method provided in an embodiment of the present application;

[0026] Figure 6 A schematic diagram of the structure of a vehicle scenario service arrangement device provided in an embodiment of the present application;

[0027] Figure 7 A hardware entity diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The technical solution of the present application is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0029] In order to enable people in this technical field to better understand the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly described below in combination with the drawings 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.

[0030] The terms "first," "second," and "third," etc. in the specification, embodiments, claims, and figures of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. In addition, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusions, for example, including a series of steps or units. A method, system, product, or apparatus is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to the process, method, product, or apparatus.

[0031] Currently, the vehicle's service orchestration function has at least the following problems: First, the manual orchestration method is unfriendly; second, users' descriptions of services vary from person to person, with a high threshold for use, making it difficult for users to fluently express the required scenario services and dependencies; third, users are relatively unfamiliar with the service parameters they frequently use, making it difficult for them to use the services accurately.

[0032] Therefore, embodiments of the present application provide a vehicle scenario service orchestration method that can be applied to computer devices, including but not limited to fixed devices and / or mobile devices. For example, the fixed devices include but are not limited to: personal computers (PCs) or servers, and the servers can be cloud servers or ordinary servers. The computer device can be a device in the vehicle or other devices outside the vehicle.

[0033] Figure 1 Schematic diagram of the implementation process of a vehicle scenario service arrangement method provided in an embodiment of the present application Figure 1 .like Figure 1 As shown, the method may include the following steps 101 to 103:

[0034] Step 101: Determine the clarity of the intention of the voice information input by the user.

[0035] Voice information refers to the voice commands conveyed by the user to the vehicle, or it can refer to the voice signals collected by the vehicle. The intent of the voice information refers to the user's current service needs. The clarity of the intent refers to whether the intent is clear.

[0036] In some embodiments, step 101 may be implemented by recognizing the user's input voice information based on the user's voice characteristics to determine the clarity of the voice information's intent. Voice characteristics include at least accent, speaking speed, and vocalization habits.

[0037] In some implementations, step 101 may be specifically implemented by converting the voice information input by the user based on the user's voice characteristics to obtain corresponding text information; and recognizing the text information to determine the clarity of the intention of the voice information.

[0038] Step 102: Based on the clarity of the intention, determine the vehicle atomic service that matches the voice information.

[0039] Vehicle atomic services break down complex vehicle functions into small, independent basic units. These units implement independent and reusable functional modules through the Application Programming Interface (API). Vehicle atomic services that match voice information refer to one or more vehicle atomic services that match user needs.

[0040] In some embodiments, step 102 may be implemented by using different matching methods based on the clarity of the intent to determine the vehicle atomic service that matches the voice information. Matching methods include at least a matching method based on a service orchestration model, a manual matching method, and a database-based matching method.

[0041] In one feasible implementation, if the matching method is based on a service orchestration model, then when the intent is clear, the first service orchestration model is used to determine the vehicle atomic service that matches the voice information; when the intent is unclear, the second service orchestration model is used to determine the vehicle atomic service that matches the voice information. Both the first service orchestration model and the second service orchestration model can be large language models (LLMs).

[0042] In one feasible implementation, if the matching method is a database-based matching method, then when the intent is clear, the first database matching method is used to determine the vehicle atomic service that matches the voice information; when the intent is unclear, the second database matching method is used to determine the vehicle atomic service that matches the voice information. The first database matching method is used to clearly filter out the vehicle atomic service that matches the voice information from the database based on the intent. The second database matching method is used to filter out the vehicle atomic service that matches the voice information from the database using a fuzzy method; specifically, first determine the text information similar to the voice information, and then filter the vehicle atomic service that matches the similar text information from the database.

[0043] Step 103: Configure the vehicle atomic service based on the user's behavioral habits to obtain the currently required vehicle scenario service.

[0044] In some embodiments, the specific implementation method of step 103 can be: based on the user's behavioral habits, determine the user's common parameters for vehicle atomic services; configure the vehicle atomic services based on the common parameters to obtain the currently required vehicle scenario services.

[0045] In an embodiment of the present application, the clarity of the intention of the voice information input by the user is determined; based on the clarity of the intention, the vehicle atomic service that matches the voice information is determined; based on the user's behavioral habits, the vehicle atomic service is configured to obtain the currently required vehicle scene service. In this way, based on the clarity of the intention of the voice information input by the user, the vehicle atomic service that matches the voice information is determined, and the conversion of the user's voice to the vehicle atomic service can be intelligently realized; based on the user's behavioral habits, the vehicle atomic service is configured to realize personalized customization of the vehicle scene service. Thus, it not only supports users to formulate vehicle scene services with voice, but also realizes personalized customization of vehicle scene services, creating practical and comfortable vehicle scene services for users.

[0046] Figure 2 Schematic diagram of the implementation process of a vehicle scenario service arrangement method provided in an embodiment of the present application Figure 2 .like Figure 2 As shown, the method may include the following steps 201 to 206:

[0047] Step 201: Obtain the voice information input by the user.

[0048] In some embodiments, step 201 may be implemented by receiving a voice message from a user to the vehicle via an application on a mobile phone, or by receiving a voice message from a user to the vehicle via a visual interface of the vehicle.

[0049] When implemented, when the user issues a voice command, the vehicle system quickly starts the voice acquisition module to collect voice information.

[0050] Step 202: Convert the voice information based on the voice characteristics of the user to obtain text information corresponding to the voice information; the voice characteristics include accent, speaking speed, and vocalization habits.

[0051] In some embodiments, step 202 may be implemented by using a noise reduction algorithm to reduce noise on the voice information, and using speech recognition technology to analyze the processed voice signal based on the user's voice characteristics to obtain text information corresponding to the voice information. The speech recognition technology may include, but is not limited to, a speech recognition model and a speech recognition engine.

[0052] The speech recognition technology may be an automatic speech recognition technology (ASR). Noise reduction processing is performed on the speech information to eliminate the interference of environmental noise and ensure that the collected speech information is clean and accurate.

[0053] In some embodiments, the specific implementation method of step 202 can be: slicing the voice information based on the user's voice characteristics to obtain multiple voice slices; converting the multiple voice slices based on the user's voice characteristics to obtain text information of the multiple voice slices; integrating the multiple voice slices to obtain text information corresponding to the voice information.

[0054] For example, when a user says "Turn the screen towards me when starting navigation", the vehicle system can accurately recognize the text information in the voice command and will not mistakenly recognize unnecessary noise text.

[0055] It should be noted that the voice processing part adopts the industry's advanced voice recognition technology to provide users with efficient, accurate and convenient voice recognition, allowing the vehicle system to hear clearly and understand, greatly improving the practicality of the vehicle scene service arrangement method provided in the embodiment of this application.

[0056] Step 203: Determine the clarity of the intention of the voice information based on the text information.

[0057] Here, the above steps 201 to 203 correspond to the above step 101, and the specific implementation of the above step 101 may be referred to during implementation.

[0058] In some implementations, step 203 may be specifically implemented by processing the text information using a binary classification model to determine whether the intention of the voice information is clear.

[0059] In some implementations, step 203 may be implemented by steps 2031 to 2033 as follows:

[0060] Step 2031: Input the text information into multiple feature extraction modules in the intention classification model in sequence, and output the text features corresponding to each feature extraction module.

[0061] The intent classification model is a trained binary classification model. The intent classification model is used to determine whether the intent of the voice information is clear. In a feasible implementation, the intent classification model may include: multiple feature extraction modules, a feature fusion layer, and a classification layer. Among them, the feature extraction module may be a Transformer layer, which is used to extract text features in text information; each Transformer layer may include a multi-head attention mechanism and a feedforward neural network; the multi-head attention mechanism is used to capture different feature relationships in text information, and the feedforward neural network is used to further extract and transform the output of the multi-head attention mechanism. The feature fusion layer is used to fuse the text features corresponding to multiple feature extraction modules. The classification layer is used to determine whether the intent of the text information is clear based on the fused text features.

[0062] Step 2032: perform feature fusion on the feature representations corresponding to the multiple feature extraction modules to obtain fused text features.

[0063] In some implementations, a feature fusion layer may be used to fuse features of outputs of multiple Transformer layers to obtain fused text features.

[0064] Step 2033: input the fused text features into the classification module in the intention classification model, and output a classification result; the classification result represents the clarity of the intention of the text information.

[0065] The classification module refers to the classification layer. For example, the classification layer can be a softmax layer.

[0066] Step 204: Based on the clarity of the intention, determine the vehicle atomic service that matches the voice information.

[0067] In some implementations, step 204 may be implemented by steps 2041 to 2042 as follows:

[0068] Step 2041: When the intention is clear, the text information corresponding to the voice information is input into the first service orchestration model, and the vehicle atomic service matching the text information is output.

[0069] The first service orchestration model is used to process the explicit intent. The first service orchestration model can be trained through the following steps a to d.

[0070] Step 2042: When the intention is unclear, input the text information corresponding to the voice information into the second service orchestration model, and output the vehicle atomic service that matches the text information.

[0071] The intent is unclear, which can also be called fuzzy intent. The first service orchestration model is used to process the unclear intent. The second service orchestration model can be trained through the following steps e to h.

[0072] It should be noted that the first service orchestration model and the second service orchestration model have two capabilities: one is to understand the relationship between the user's vernacular language (text information for recognizing the user's voice) and the vehicle-side atomic services; the other is to be able to understand and infer the service combination that the user may want based on the user's vernacular language (vehicle atomic services matching the text information).

[0073] Step 205: Based on the user's identification and the functional category of the vehicle atomic service, obtain the user's common parameters regarding the vehicle atomic service from the service parameter management system; the service parameter management system is updated in real time based on changes in the user's behavior and changes in the vehicle's environment; the common parameters represent the user's behavioral habits.

[0074] The user's identifier can be an ID code, used to identify the user. The service parameter management system collects user vehicle usage habits and, when automatically generating and creating vehicle scenario services, adaptively modifies the parameters of the generated services to personalize vehicle scenario services and truly tailor vehicle scenario orchestration services to each individual.

[0075] In some embodiments, the specific implementation method of building a service parameter management system can be: by using the method of embedding points, collecting information on users' use of vehicle-side scene services and uploading it to the cloud; using data mining technology and machine learning technology to analyze and extract the collected information, and generate user tags with semantic information; classifying, storing and indexing user tags to obtain a service parameter management system. It should be noted that when collecting information, this application only collects information recognized by the law and does not involve user privacy. Among them, user tags are classified, stored and indexed according to the user's identification and the functional category of the vehicle atomic service.

[0076] Update user tags in real time or periodically. As user behavior and preferences change, timely adjust and add new tags, and delete tags that are no longer applicable. For example, consider the air conditioning air volume and temperature settings frequently used over the past seven days. As the seasons change, the user's air conditioning temperature and air volume settings will also change accordingly.

[0077] In some embodiments, the specific implementation method of step 205 can be: obtaining a user tag based on the user's identification, and obtaining the user's commonly used service parameters for the vehicle atomic service in recent time based on the user tag and the functional category of the vehicle atomic service.

[0078] Step 206: Configure the vehicle atomic service based on the common parameters to obtain the currently required vehicle scenario service.

[0079] Here, the above steps 205 to 206 correspond to the above step 103, and the specific implementation of the above step 103 may be referred to during implementation.

[0080] In some embodiments, after step 206, the following steps may also be performed: performing a safety check on the vehicle scenario service; performing a conflict check on the vehicle scenario service; the safety check and the conflict check are used to determine the executability of the vehicle scenario service; when both the safety check and the conflict check indicate that there is no abnormality in the vehicle scenario service, calling the application programming interface of the vehicle scenario service to execute the vehicle scenario service.

[0081] Safety and conflict detection for vehicle scenario services ensures that services can be executed safely in the current environment. Both safety and conflict detection are built on expert systems to avoid service execution security issues. For example, a heavy rain outside might conflict with opening the sunroof. Service execution conflicts can also be avoided, such as opening a window conflicting with activating the external air circulation system.

[0082] In some embodiments, the first service orchestration model in step 2041 can be trained through the following steps a to d:

[0083] Step a: Filter out a first vehicle scenario service with a clear intention from historical vehicle scenario services arranged before the current moment.

[0084] The historical vehicle scenario service refers to the vehicle scenario service arranged before the current moment. The first vehicle scenario service refers to the vehicle scenario service with a clear intention.

[0085] In some implementations, step a may be specifically implemented by manually screening out a first vehicle scenario service with a clear intention from historical vehicle scenario services.

[0086] In some embodiments, the specific implementation method of step a can also be: using the above-mentioned intention classification model to determine the clarity of the intentions of multiple historical vehicle scenario services; and using the historical vehicle scenario service with clear intentions as the first vehicle scenario service.

[0087] Step b: pre-processing the first vehicle scenario service to obtain the processed first vehicle scenario service.

[0088] In some embodiments, the "preprocessing of the first vehicle scenario service" in step b includes at least one of the following: data cleaning of the first vehicle scenario service; anomaly removal of the first vehicle scenario service; and format conversion of the first vehicle scenario service.

[0089] Data cleansing for the first vehicle scenario service removes duplicate, meaningless, and incorrectly formatted records. Anomaly removal removes records that don't match the actual context. Format conversion organizes the vehicle scenario service into a question-and-answer format for easier model processing. For example: Question (user voice) - Answer (matched vehicle atomic service).

[0090] Step c: Based on the processed first vehicle scenario service, train the first fine-tuning model in the attention layer of the base model to obtain the trained first fine-tuning model.

[0091] The first fine-tuning model may refer to a LoRA fine-tuning model. The LoRA fine-tuning model includes two trainable low-rank matrices (parameter matrices), and the LoRA matrix can be inserted into the attention layer.

[0092] In some embodiments, the specific implementation method of step c can be: based on the processed first vehicle scene service, adjust the parameter matrix of the first fine-tuning model in the attention layer of the base model until the iteration stop condition is met to obtain the trained first fine-tuning model.

[0093] In some embodiments, step c can be implemented by the following steps c1 to c3:

[0094] Step c1: input the processed first vehicle scenario service into the first fine-tuning model, and output the processing result of the first fine-tuning model.

[0095] Step c2: Based on the output result of the first fine-tuning model, determine the overall weight matrix of the first fine-tuning model and the base model, and the incremental matrix of the first fine-tuning model.

[0096] Step c3: Based on the overall weight matrix and the incremental matrix, adjust the parameter matrix of the first fine-tuning model until an iteration stop condition is reached to obtain the trained first fine-tuning model.

[0097] In a standard neural network model, the calculation of the linear layer can be expressed as: y = Wx; where W is the weight matrix, x is the input, and y is the output. Traditional fine-tuning requires updating the large matrix W, but LoRA's approach keeps W unchanged and adds a trainable incremental matrix ΔW, so that the fine-tuned output is: y = (W + ΔW)x.

[0098] In order to reduce the number of parameters of ΔW, LoRA decomposes ΔW into two low-rank matrices A and B, namely: ΔW = AB; among them, the dimensions of A and B are much smaller than W, which helps to reduce the parameters that need to be updated during fine-tuning.

[0099] During the training process, the gradient descent algorithm is used to update the parameters of the A matrix and B matrix introduced by LoRA. The W of the original model will not be updated. The forward propagation of the model will still calculate the complete Wx, and will also calculate the increment ΔWx = (AB)x, and then add them together. In this way, LoRA fine-tuning can adapt to new task data.

[0100] Step d: Determine the first service orchestration model based on the trained first fine-tuning model and the base model.

[0101] In some embodiments, the second service orchestration model in step 2042 can be trained through the following steps e to h:

[0102] Step e: Filter out a second vehicle scenario service with unclear intention from the historical vehicle scenario services arranged before the current moment.

[0103] It should be noted that the training method of the second service orchestration model is similar to that of the first service orchestration model, and when implementing it, reference may be made to the specific implementation method of the aforementioned first service orchestration model.

[0104] The second vehicle scenario service refers to a vehicle scenario service with unclear intentions.

[0105] In some implementations, step e may be specifically implemented by manually screening out the first vehicle scenario service with unclear intent from historical vehicle scenario services.

[0106] In some embodiments, the specific implementation method of step e can also be: using the above-mentioned intention classification model to determine the clarity of the intentions of multiple historical vehicle scenario services; and using the historical vehicle scenario service with unclear intentions as the first vehicle scenario service.

[0107] Step f: pre-process the second vehicle scenario service to obtain the processed second vehicle scenario service.

[0108] In some embodiments, the "preprocessing of the second vehicle scenario service" in step f includes at least one of the following: data cleaning of the second vehicle scenario service; anomaly removal of the second vehicle scenario service; and format conversion of the second vehicle scenario service.

[0109] Step g: Based on the processed second vehicle scenario service, train the second fine-tuning model in the attention layer of the base model to obtain the trained second fine-tuning model.

[0110] The second fine-tuning model may also be a LoRA fine-tuning model.

[0111] In some embodiments, the specific implementation method of step g can be: based on the processed second vehicle scene service, adjust the parameter matrix of the second fine-tuning model in the attention layer of the base model until the iteration stop condition is met to obtain a trained second fine-tuning model.

[0112] In some embodiments, the specific implementation method of step g can also be: input the processed second vehicle scenario service into the second fine-tuning model, and output the processing result of the second fine-tuning model; based on the output result of the second fine-tuning model, determine the overall weight matrix of the second fine-tuning model and the base model, as well as the incremental matrix of the second fine-tuning model; based on the overall weight matrix and the incremental matrix, adjust the parameter matrix of the second fine-tuning model until the iteration stop condition is reached, and obtain the trained second fine-tuning model.

[0113] Step h: Determine the second service orchestration model based on the trained second fine-tuning model and the base model.

[0114] It should be noted that this application is based on the same base model and performs dynamic scheduling on different LoRA fine-tuning models to achieve vehicle scenario service generation under different needs.

[0115] In the embodiment of the present application, in the process of converting the user's voice into text information, the influence of different accents, speaking speeds and pronunciation habits on voice recognition is fully considered, and high-accuracy voice recognition is achieved. Based on the clarity of the intention of the voice information input by the user, different service orchestration models are used to determine the vehicle atomic service that matches the voice information, which can intelligently realize the understanding and creation relationship between the user's voice and the vehicle atomic service; based on the user's behavioral habits, the parameters of the vehicle atomic service are personalized, which can realize fast, accurate and personalized creation of vehicle scene services. Therefore, it not only supports users to formulate vehicle scene services with voice, but also realizes personalized customization of vehicle scene services, creating practical and comfortable vehicle scene services for users.

[0116] The following describes the application of the vehicle scenario service orchestration method provided in an embodiment of the present application in an actual scenario.

[0117] Background research shows that existing technologies have the following three main deficiencies: 1. Existing solutions only provide pre-placement, triggering, and execution-related definitions and cloud-based orchestration functions for orchestrated business functions, and do not support the ability to formulate scenario services based on voice; 2. Existing solutions cannot support service creation capabilities; 3. Existing solutions cannot support service personalization capabilities.

[0118] Based on the above problems, this application proposes a creative and personalized voice scene arrangement scheme (corresponding to the above-mentioned vehicle scene service arrangement method), which is a technical solution based on a generative big model. This application uses a big model for scene modeling through the input of user voice, and integrates the big model thinking chain capability (Chain of Thought, CoT) of service creation to automatically generate vehicle scene services with multiple service combinations for the user's required car needs, and then combines the user's car usage habits to provide personalized scene services for the user. In this way, the understanding and creation relationship between user voice and vehicle scene service is realized, and then the service parameters are personalized configured according to the user's historical car usage habits, so that the user can quickly, accurately, creatively and personalizedly generate vehicle scene services.

[0119] This application solves the following three major pain points: First, it solves the unfriendly nature of manual arrangement by users through direct voice speaking, thereby improving convenience; second, it supports scene arrangement in vernacular to solve the problem that users have a high threshold for using standard descriptions of services; third, it supports personalized service generation, so that the generated services can fit the user's car usage habits, which is a breakthrough technology for intelligent interaction of smart cars.

[0120] This application includes four modules: speech understanding, service generation and creation, personalized service adaptation, and function call.

[0121] 1. Speech comprehension;

[0122] In this application, the voice processing function plays a vital role. Through the existing high-accuracy voice recognition technology, the user's input voice can be accurately converted into text information. When the user issues a voice command, the system quickly starts the voice acquisition module, collects the sound signal, and uses an efficient algorithm to perform noise reduction processing to eliminate the interference of environmental noise and ensure that the collected voice information is clear and accurate. Then, the voice recognition engine analyzes the processed voice information and converts it into understandable text information. In this process, the system fully takes into account different accents, speaking speeds and pronunciation habits, and achieves high-accuracy voice recognition through a large amount of training data and continuously optimized models.

[0123] For example, when a user says "Turn the screen towards me when starting navigation", the system can accurately recognize the instruction and provide the text to the next module without misidentifying redundant noise text. The system will accurately recognize and convey the instruction to the service generation and creation module.

[0124] The voice processing function adopts the industry's advanced voice recognition technology to provide users with efficient, accurate and convenient voice recognition and understanding, allowing the system to hear clearly and understand, greatly improving the practicality of the product.

[0125] 2. Service generation and creation;

[0126] The main task of service generation and creation is to enable the big prediction model to have two capabilities: one is to understand the relationship between the user's vernacular language and the vehicle's atomic services; the other is to be able to understand and infer the service combination that the user may want based on the user's vernacular language.

[0127] Based on the scenario requirements, a binary classification model with a multi-layer Transformer architecture is designed to distinguish whether the user's statement has a clear intention. For example, a clear intention (when I start navigation, the screen turns to me) or an ambiguous intention (I want to sleep).

[0128] The specific implementation is as follows:

[0129] 1. Data input: Receive the data to be classified, the data type is text.

[0130] 2. Multi-layer Transformer processing: Input data is processed sequentially through multiple Transformer layers. Each Transformer layer includes a multi-head attention mechanism and a feedforward neural network. The multi-head attention mechanism is used to capture the relationships between different features in the data, and the feedforward neural network further extracts and transforms the output of the attention mechanism.

[0131] 3. Feature fusion: The text features output by multiple Transformer layers are fused to obtain fused text features.

[0132] 4. Classification output: Input the fused text features into the classification layer to obtain the binary classification results.

[0133] Based on whether the user's voice intent is clear, different model links are used, as follows:

[0134] Route 1: Clear Intent: The recognized user voice input is input into the LoRA_1 secondary fine-tuned orchestration model (corresponding to the first service orchestration model mentioned above). After understanding the user's explicit verbal description, the corresponding vehicle-side atomic service is determined and output. For example: input = "When I start navigation, the screen turns to me"; output = {"trigger":"navigation service, navigation parameters", "action":"screen service, screen parameters"}.

[0135] Route 2: Fuzzy Intent: The recognized user voice content is input into the LoRA_2 secondary fine-tuned orchestration model (corresponding to the second service orchestration model mentioned above). After understanding the user's ambiguous vernacular description, the CoT capability is used to creatively output the corresponding vehicle-side atomic service. For example: input = "I want to sleep"; output = {"action":"Ambient light service, ambient light parameters", "Music service, music parameters", "Seat service, seat parameters", "Window service, window parameters"}.

[0136] Technical implementation plan:

[0137] 1. Sample data collection and preprocessing. Scenario orchestration-related text data is collected from multiple channels, including historical records of vehicle orchestration functions, product documentation, and manual annotations. Both explicit and ambiguous intent samples must contain at least 20,000 entries. After collecting the sample data, it is cleaned to remove duplicate, meaningless, and incorrectly formatted records. The samples are then formatted according to the question-and-answer format used for fine-tuning the large model to facilitate model processing.

[0138] 2. LoRA fine-tuning model. Select an open source base model of a certain size and design a LoRA fine-tuning architecture, including two trainable low-rank matrices. The LoRA matrix can be inserted into the attention layer. The detailed process is as follows:

[0139] In a standard neural network, the linear layer calculation can be expressed as: y = Wx; where W is the weight matrix, x is the input, and y is the output. Traditional fine-tuning requires updating this large matrix, W. However, LoRA's approach maintains W constant and adds a trainable increment, ΔW, so that the fine-tuned output is: y = (W + ΔW)x.

[0140] In order to reduce the number of parameters of ΔW, LoRA decomposes ΔW into two low-rank matrices A and B, namely: ΔW = AB; among them, the dimensions of A and B are much smaller than W, which helps to reduce the parameters that need to be updated during fine-tuning.

[0141] During the training process, the gradient descent algorithm is used to update the parameters of the A and B matrices introduced by LoRA. The W of the original model will not be updated. The forward propagation of the model will still calculate the complete Wx, and at the same time, the increment ΔWx = (AB)x will be calculated and then added. In this way, LoRA fine-tuning can adapt to new task data.

[0142] 3. After the two fine-tuned LoRAs are trained, they can be used in conjunction with the base model. When the pre-order classification model divides the user input into clear intents, the system will automatically call LoRA_1 and combine the base model to infer the input and output vehicle scene services; when the pre-order classification model divides the user input into fuzzy intents, the system will automatically call LoRA_2 and combine the base model to infer the input and create services, and output vehicle scene services. So far, this application can dynamically schedule on different fine-tuned LoRAs based on the same base model to achieve vehicle scene service generation under different requirements.

[0143] 3. Personalized service adaptation;

[0144] By building a labeling system (corresponding to the above-mentioned service parameter management system) to collect user car usage habits (corresponding to the above-mentioned user behavior habits), when automatically generating and creating vehicle scene services, the labeling system is called and the parameters of the generated services are adaptively modified to achieve personalized vehicle scene services, and truly ensure that vehicle scene orchestration services are tailored to each person and are different for each person.

[0145] The detailed technical solution is as follows:

[0146] 1. User tag collection. Signals indicating vehicle-side user service usage are collected and uploaded to the cloud through tracking (this does not involve confidential user information, and only legally permitted data is tracked). Data mining and machine learning techniques are used to analyze and extract the collected data, generating user tags with semantic information. For example, commonly used air conditioning air volume and temperature settings over the past seven days.

[0147] 2. Tag management and updating. Establish a tag system to categorize, store, and index user tags. Update user tags in real time or periodically. As user behavior and preferences change, adjust and add new tags promptly, and delete tags that are no longer applicable. For example, as the seasons change, the user's air conditioning temperature and air volume settings will change accordingly.

[0148] 3. Personalized service generation module. Figure 4 As shown, after the model generates the corresponding service, the system will query the user's common service parameters corresponding to the vehicle scenario service in recent time based on the user's identification, and assign the parameters to the service parameters of the vehicle scenario service, thereby realizing personalized service generation.

[0149] The purpose of personalized service adaptation is to accurately grasp user needs through in-depth mining and analysis of user tags, and provide users with highly personalized services, thereby improving user satisfaction and service quality.

[0150] 4. Function call;

[0151] After generating corresponding vehicle scenario services based on user descriptions, safety and conflict detection must be performed on the service combination to ensure that the services can be executed safely in the current environment. Both safety and conflict detection modules are built on expert systems to avoid service execution safety issues, such as opening the sunroof during a heavy rainstorm. They also avoid service execution conflicts, such as opening a window and turning on the external air circulation system.

[0152] In summary, this application provides users with a creative and personalized voice scene arrangement function. It not only supports users to define scenes by voice, but also supports users to automatically create service combinations through vernacular interaction. It further automatically customizes service settings based on user usage habits. It allows users to become the product that defines the scene, creating practical and comfortable car scene services for themselves.

[0153] like Figure 3 As shown, the personalized voice scene service arrangement solution of the embodiment of the present application includes the following steps 301 to 304:

[0154] Step 301: Obtain user voice input.

[0155] Step 302: The orchestration big model automatically generates corresponding services based on user voice, and uses the big model's general knowledge capabilities to produce creative services based on prompt engineering.

[0156] Step 303: Obtain tag data of the user's car-using habits, switch and generate service parameters in the service, and implement personalized service.

[0157] Step 304: Call and execute the service based on the generated service.

[0158] like Figure 4As shown, the personalized voice scene service orchestration solution of the embodiment of the present application specifically includes four parts: voice input, orchestration big model, labeling system and service execution. Among them, voice input includes: voice input, voice recognition (ASR recognition). The orchestration big model includes: text input distribution, low-rank matrix, orchestration big model, service combination; specifically, the text input is distributed to the orchestration big model to obtain the service combination; in this process, the low-rank matrix can be updated based on the text input. The labeling system includes: user identification, user service label, new service combination; specifically, the user service label is obtained based on the user identification to obtain a new service combination. Service execution includes: service API library, function call, security and conflict detection, vehicle scene service generation; specifically, function call is performed through the service API library, security and conflict detection is performed on the service combination output by the big model, and after detection, the vehicle scene service is generated.

[0159] like Figure 5 As shown, the processing flow of the orchestration large model includes: inputting the text input into the binary classification model; if the classification result is that the intention is clear, the text input is processed based on LoRA_1 and the language large model (corresponding to the above-mentioned base model) to obtain a service combination; if the classification result is that the intention is unclear, the text input is processed based on LoRA_2 and the language large model to obtain a service combination.

[0160] The points to be protected in this application include at least: 1. Automatic generation of vehicle atomization services based on vernacular voice based on a large model; 2. Vehicle scenario services based on a combination of hundreds of vernacular voice automatic creation services based on a large model; 3. Model architecture that dynamically solves two types of problems with a single large model; 4. Personalized scenario service solutions.

[0161] Based on the foregoing embodiments, an embodiment of the present application provides a vehicle scenario service orchestration device, which includes the various units included and the various modules included in each unit, and can be implemented by a processor in a computer device; of course, it can also be implemented by a specific logic circuit; in the implementation process, the processor can be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP) or a field programmable gate array (FPGA), etc.

[0162] Figure 6 A schematic diagram of the structure of a vehicle scene service arrangement device provided in an embodiment of the present application is shown as follows: Figure 6 As shown, the vehicle scenario service arrangement device 600 includes: a determination module 610 and a processing module 620, wherein:

[0163] Determination module 610, for determining the clarity of the intention of the voice information input by the user;

[0164] A processing module 620 is configured to determine a vehicle atomic service that matches the voice information based on the clarity of the intention;

[0165] The processing module 620 is further used to configure the vehicle atomic service based on the user's behavioral habits to obtain the currently required vehicle scenario service.

[0166] In some embodiments, the determination module 610 is further used to obtain voice information input by the user; convert the voice information based on the voice characteristics of the user to obtain text information corresponding to the voice information; the voice characteristics include: accent, speaking speed, and vocal habits; based on the text information, determine the clarity of the intention of the voice information.

[0167] In some embodiments, the determination module 610 is also used to input the text information into multiple feature extraction modules in the intention classification model in sequence, and output the text features corresponding to each feature extraction module; perform feature fusion on the text features corresponding to the multiple feature extraction modules to obtain fused text features; input the fused text features into the classification module in the intention classification model, and output the classification results; the classification results represent the clarity of the intention of the text information.

[0168] In some embodiments, the processing module 620 is further used to, when the intention is clear, input the text information corresponding to the voice information into the first service orchestration model, and output the vehicle atomic service that matches the text information; when the intention is unclear, input the text information corresponding to the voice information into the second service orchestration model, and output the vehicle atomic service that matches the text information.

[0169] In some embodiments, the processing module 620 is also used to screen out a first vehicle scenario service with clear intention from the historical vehicle scenario services arranged before the current moment; preprocess the first vehicle scenario service to obtain a processed first vehicle scenario service; train the first fine-tuning model in the attention layer of the base model based on the processed first vehicle scenario service to obtain the trained first fine-tuning model; and determine the first service orchestration model based on the trained first fine-tuning model and the base model.

[0170] In some embodiments, the processing module 620 is further used to perform data cleaning on the first vehicle scenario service; remove anomalies on the first vehicle scenario service; and convert the format of the first vehicle scenario service.

[0171] In some embodiments, the processing module 620 is also used to input the processed first vehicle scenario service into the first fine-tuning model and output the processing result of the first fine-tuning model; based on the output result of the first fine-tuning model, determine the overall weight matrix of the first fine-tuning model and the base model, as well as the incremental matrix of the first fine-tuning model; based on the overall weight matrix and the incremental matrix, adjust the parameter matrix of the first fine-tuning model until the iteration stop condition is reached, thereby obtaining the trained first fine-tuning model.

[0172] In some embodiments, the processing module 620 is also used to screen out a second vehicle scenario service with unclear intentions from the historical vehicle scenario services arranged before the current moment; preprocess the second vehicle scenario service to obtain a processed second vehicle scenario service; train a second fine-tuning model in the attention layer of the base model based on the processed second vehicle scenario service to obtain a trained second fine-tuning model; and determine the second service orchestration model based on the trained second fine-tuning model and the base model.

[0173] In some embodiments, the processing module 620 is also used to obtain the user's common parameters regarding the vehicle atomic service from the service parameter management system based on the user's identification and the functional category of the vehicle atomic service; the service parameter management system is updated in real time based on changes in user behavior and changes in the vehicle's environment; the common parameters represent the user's behavioral habits; the vehicle atomic service is configured based on the common parameters to obtain the currently required vehicle scenario service.

[0174] In some embodiments, the processing module 620 is also used to perform safety detection on the vehicle scenario service; perform conflict detection on the vehicle scenario service; the safety detection and the conflict detection are used to determine the executability of the vehicle scenario service; when the safety detection and the conflict detection both indicate that there is no abnormality in the vehicle scenario service, call the application programming interface of the vehicle scenario service to execute the vehicle scenario service.

[0175] The description of the above device embodiment is similar to the description of the above method embodiment and has similar beneficial effects as the method embodiment. In some embodiments, the functions or modules included in the device provided in the embodiments of the present application can be used to perform the methods described in the above method embodiments. For technical details not disclosed in the device embodiments of the present application, please refer to the description of the method embodiments of the present application for understanding.

[0176] It should be noted that, in the embodiment of the present application, if the above-mentioned vehicle scenario service arrangement method is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application can be embodied in the form of a software product in essence or in other words, the part that contributes to the relevant technology can be stored in the form of a software product. The software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk. In this way, the embodiment of the present application is not limited to any specific hardware, software or firmware, or any combination of hardware, software and firmware.

[0177] An embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0178] The embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements some or all of the steps in the above method. The computer-readable storage medium may be transient or non-transient.

[0179] An embodiment of the present application provides a computer program, including computer-readable code. When the computer-readable code is run in a computer device, a processor in the computer device executes some or all of the steps for implementing the above method.

[0180] An embodiment of the present application provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and when the computer program is read and executed by a computer, implements some or all of the steps in the above method. The computer program product can be implemented specifically by hardware, software, or a combination thereof. In some embodiments, the computer program product is embodied as a computer storage medium. In other embodiments, the computer program product is embodied as a software product, such as a software development kit (SDK), etc.

[0181] It should be noted that the descriptions of the various embodiments above tend to emphasize the differences between the various embodiments, and their similarities or similarities can be referenced to each other. The descriptions of the above device, storage medium, computer program, and computer program product embodiments are similar to the descriptions of the above method embodiments and have similar beneficial effects as the method embodiments. For technical details not disclosed in the embodiments of the device, storage medium, computer program, and computer program product of this application, please refer to the description of the method embodiments of this application for understanding.

[0182] It should be noted that Figure 7 A schematic diagram of a hardware entity of a computer device in an embodiment of the present application is shown in FIG. Figure 7 As shown, the hardware entity of the computer device 700 includes: a processor 701, a communication interface 702 and a memory 703, wherein:

[0183] Processor 701 generally controls the overall operation of computer device 700 .

[0184] The communication interface 702 enables the computer device to communicate with other terminals or servers through a network.

[0185] The memory 703 is configured to store instructions and applications executable by the processor 701. It can also cache data to be processed or processed by the processor 701 and various modules in the computer device 700 (for example, image data, audio data, voice communication data, and video communication data). This can be implemented using flash memory (FLASH) or random access memory (RAM). Data can be transmitted between the processor 701, the communication interface 702, and the memory 703 via a bus 704.

[0186] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned steps / processes does not mean the order of execution, and the execution order of each step / process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The above-mentioned serial numbers of the embodiments of the present application are for description only and do not represent the advantages and disadvantages of the embodiments.

[0187] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0188] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0189] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed across multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0190] In addition, all functional units in the embodiments of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the above-mentioned integrated units can be implemented in the form of hardware or in the form of hardware plus software functional units.

[0191] Those skilled in the art will understand that all or part of the steps of implementing the above-mentioned method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above-mentioned method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM), magnetic disks or optical disks, and other media that can store program codes.

[0192] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the relevant technology, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0193] The above is only an implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A vehicle scenario service arrangement method, characterized in that: The vehicle scenario service arrangement method comprises: Determine the clarity of the user's input voice message's intent; Based on the clarity of the intent, determining a vehicle atomic service that matches the voice information; The vehicle atomic service is configured based on the user's behavioral habits to obtain the currently required vehicle scenario service.

2. The vehicle scenario service arrangement method according to claim 1, characterized in that: The determining of the clarity of the intention of the voice information input by the user includes: Obtaining voice information input by the user; Converting the voice information based on the voice characteristics of the user to obtain text information corresponding to the voice information; the voice characteristics include: accent, speaking speed, and vocalization habits; Based on the text information, the clarity of the intent of the voice information is determined.

3. The vehicle scenario service arrangement method according to claim 2, characterized in that: The determining, based on the text information, the clarity of the intention of the voice information includes: Inputting the text information into multiple feature extraction modules in the intention classification model in sequence, and outputting the text features corresponding to each feature extraction module; Performing feature fusion on the text features corresponding to the plurality of feature extraction modules to obtain fused text features; The fused text features are input into a classification module in the intention classification model, and a classification result is output; the classification result represents the clarity of the intention of the text information.

4. The vehicle scenario service arrangement method according to any one of claims 1 to 3, characterized in that: The determining of the vehicle atomic service matching the voice information based on the clarity of the intention includes: In the case where the intention is clear, inputting text information corresponding to the voice information into a first service orchestration model, and outputting a vehicle atomic service matching the text information; When the intention is unclear, the text information corresponding to the voice information is input into the second service orchestration model, and the vehicle atomic service matching the text information is output.

5. The vehicle scenario service arrangement method according to claim 4, characterized in that: The first service orchestration model is trained by the following steps: Filter out the first vehicle scenario service with clear intention from the historical vehicle scenario services arranged before the current moment; Preprocessing the first vehicle scenario service to obtain a processed first vehicle scenario service; Training a first fine-tuning model at the attention layer of the base model based on the processed first vehicle scenario service to obtain the trained first fine-tuning model; The first service orchestration model is determined based on the trained first fine-tuning model and the base model.

6. The vehicle scenario service arrangement method according to claim 5, characterized in that: The preprocessing of the first vehicle scenario service includes at least one of the following: Performing data cleaning on the first vehicle scenario service; Remove abnormalities from the first vehicle scenario service; The first vehicle scenario service is format converted.

7. The vehicle scenario service arrangement method according to claim 5, characterized in that: The first fine-tuning model in the attention layer of the base model is trained based on the processed first vehicle scenario service to obtain the trained first fine-tuning model, including: Inputting the processed first vehicle scenario service into the first fine-tuning model, and outputting a processing result of the first fine-tuning model; Determine, based on the output result of the first fine-tuning model, an overall weight matrix of the first fine-tuning model and the base model, and an increment matrix of the first fine-tuning model; Based on the overall weight matrix and the incremental matrix, the parameter matrix of the first fine-tuning model is adjusted until an iteration stop condition is reached to obtain the trained first fine-tuning model.

8. The vehicle scenario service arrangement method according to claim 4, characterized in that: The second service orchestration model is trained by the following steps: Filter out a second vehicle scenario service with unclear intention from historical vehicle scenario services arranged before the current moment; Preprocessing the second vehicle scenario service to obtain a processed second vehicle scenario service; Training a second fine-tuning model in the attention layer of the base model based on the processed second vehicle scenario service to obtain a trained second fine-tuning model; The second service orchestration model is determined based on the trained second fine-tuning model and the base model.

9. The vehicle scenario service arrangement method according to claim 1, characterized in that: The configuring the vehicle atomic service based on the user's behavior habits to obtain the currently required vehicle scenario service includes: Based on the user's identification and the functional category of the vehicle atomic service, the user's common parameters for the vehicle atomic service are obtained from a service parameter management system; the service parameter management system is updated in real time based on changes in the user's behavior and changes in the vehicle's environment; the common parameters represent the user's behavior habits; The vehicle atomic service is configured based on the common parameters to obtain the currently required vehicle scenario service.

10. The vehicle scenario service arrangement method according to claim 1, characterized in that: The vehicle scenario service arrangement method further includes: Performing safety testing on the vehicle scenario service; Performing conflict detection on the vehicle scenario service; the safety detection and the conflict detection are used to determine the executability of the vehicle scenario service; When both the safety detection and the conflict detection indicate that there is no abnormality in the vehicle scenario service, an application programming interface of the vehicle scenario service is called to execute the vehicle scenario service.

11. A vehicle scene service arrangement device, characterized in that: The vehicle scenario service arrangement device comprises: A determination module, used to determine the clarity of the intention of the voice information input by the user; A processing module, configured to determine a vehicle atomic service matching the voice information based on the clarity of the intention; The processing module is also used to configure the vehicle atomic service based on the user's behavioral habits to obtain the currently required vehicle scenario service.

12. A computer device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, wherein: When the processor executes the program, the steps of the method according to any one of claims 1 to 10 are implemented.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

14. A computer program product, comprising a non-transitory computer-readable storage medium storing a computer program, wherein when the computer program is read and executed by a computer, the steps of the method according to any one of claims 1 to 10 are implemented.

Images