Vehicle control method, server and computer readable storage medium

Abstract

The invention discloses a vehicle control method, a server and a computer readable storage medium. The method comprises the steps of obtaining a trigger event and current sensing information; and based on a pre-configured large language model and a pre-constructed function point knowledge base, according to the trigger event and the current perception information, determining a target function point associated with the trigger event, current state information of the target function point and target state information. And determining a vehicle control instruction and an operation suggestion of the vehicle control instruction according to the current state information and the target state information. And issuing a vehicle control instruction and an operation suggestion to the vehicle to complete vehicle control. Therefore, the server actively provides services for the user according to the trigger event and the current perception information. Moreover, through the pre-constructed function point knowledge base, the user intention can be accurately understood according to the trigger event, and the target function point and the vehicle control instruction capable of realizing the user intention can be determined, so that the user experience is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of vehicle control technology, and in particular to a vehicle control method, a server and a computer-readable storage medium. Background Art

[0002] In the related art, the in-car assistant interacts with the user to facilitate the user's operation. However, the in-car assistant has defects in the vertical field proprietary knowledge, and the reasoning ability of business specific terms is insufficient, and it cannot accurately understand the user's intention, which affects the user experience. Summary of the invention

[0003] The present application provides a vehicle control method, a voice interaction method, a server and a computer-readable storage medium.

[0004] The present application provides a vehicle control method, the method comprising:

[0005] Obtain trigger events and current perception information;

[0006] Based on a pre-configured large language model and a pre-built function point knowledge base, according to the trigger event and the current perception information, determine a target function point associated with the trigger event, current state information of the target function point, and target state information;

[0007] Determining a vehicle control instruction and an operation suggestion for the vehicle control instruction according to the current state information and the target state information;

[0008] The vehicle control instruction and the operation suggestion are sent to the vehicle to complete the vehicle control.

[0009] In this way, the server obtains the trigger event and the current perception information. Then, based on the pre-configured large language model and the pre-built function point knowledge base, the server determines the target function point associated with the trigger event, the current state information of the target function point, and the target state information according to the trigger event and the current perception information. Then, the server determines the vehicle control instructions and the operation suggestions for the vehicle control instructions according to the current state information and the target state information. Finally, the server sends the vehicle control instructions and the operation suggestions to the vehicle to complete the vehicle control. In this way, the server actively provides services to the user based on the trigger event and the current perception information. Moreover, through the pre-built function point knowledge base, it is possible to accurately understand the user's intention based on the trigger event, and determine the target function points and vehicle control instructions that can achieve the user's intention, thereby enhancing the user experience.

[0010] In some implementations, the function point knowledge base is constructed by the following steps:

[0011] Determining, according to the vehicle function points, associated attributes and function point operations of the vehicle function points;

[0012] Determining functional description information of the function point operation according to the associated attribute and the function point operation;

[0013] The function point knowledge base is constructed according to the associated attributes, the function point operations and the function description information.

[0014] In this way, the function point knowledge base is constructed through the following steps: First, the server determines the associated attributes and function point operations of the vehicle function points based on the vehicle function points. Next, the server determines the functional description information of the function point operations based on the associated attributes and function point operations. Finally, the server constructs the function point knowledge base based on the associated attributes, function point operations and functional description information. In this way, the function point knowledge base can provide detailed information for each function point, including associated attributes, function point operations and functional description information, thereby helping the large language model to accurately understand the function and purpose of each function point, thereby enhancing the user experience. In addition, the function point knowledge base can help the large language model cope with a variety of different scenarios and situations, thereby improving the robustness of the large language model.

[0015] In some embodiments, the large language model includes a preset prompt word, and the method based on the pre-configured large language model and the pre-built function point knowledge base, according to the trigger event and the current perception information, determines the target function point associated with the trigger event, the current state information of the target function point, and the target state information, including:

[0016] Based on the preset prompt word and according to the trigger event, acquiring the target function point, the target function point operation corresponding to the target function point and the associated background knowledge from the function point knowledge base;

[0017] The current state information is determined according to the target function point and the current perception information.

[0018] In this way, based on the preset prompt words, the server obtains the target function point, the target function point operation corresponding to the target function point, and the associated background knowledge from the function point knowledge base according to the trigger event. Then, the server determines the current state information according to the target function point and the current perception information. In this way, by pre-setting the preset prompt words, the large language model can be guided to use its own powerful semantic understanding and reasoning capabilities to accurately obtain the required data information from the function point knowledge base according to the trigger event, thereby accurately generating vehicle control instructions that can improve the user experience.

[0019] In some implementations, based on the preset prompt word and according to the trigger event, acquiring the target function point, the target function point operation corresponding to the target function point, and the associated background knowledge from the function point knowledge base includes:

[0020] Determining user intent based on the trigger event;

[0021] Determining target-related attributes according to the user intention;

[0022] Acquire, from the function point knowledge base according to the target association attribute, a target function point associated with the trigger event, a target function point operation corresponding to the target function point, and target function description information corresponding to the target function point operation;

[0023] The associated background knowledge is determined according to the target function description information.

[0024] In this way, the server determines the user's intention based on the triggering event. Next, the server determines the target-related attributes based on the user's intention. Then, the server obtains the target function point associated with the triggering event, the target function point operation corresponding to the target function point, and the target function description information corresponding to the target function point operation from the function point knowledge base based on the target-related attributes. Finally, the server determines the associated background knowledge based on the target function description information. In this way, by accurately understanding the user's intention and determining the target-related attributes, the target function point associated with the triggering event can be accurately determined from the function point knowledge base, reducing the number of function points that need to be inferred by the large language model, speeding up the inference speed of the large language model, and thus enhancing the user experience.

[0025] In some embodiments, the determining, based on the preconfigured large language model and the pre-built function point knowledge base, according to the trigger event and the current perception information, the target function point associated with the trigger event, the current state information of the target function point, and the target state information includes:

[0026] Based on the preset prompt word, according to the target function point operation and the associated background knowledge, the target state information is determined.

[0027] In this way, based on the preset prompt words, the server determines the target state information according to the target function point operation and the associated background knowledge. In this way, according to the preset prompt words, the target function point operation and the associated background knowledge, the target state information of the target function point is accurately determined, the operation efficiency is improved, and the user experience is enhanced.

[0028] In certain embodiments, the method further comprises:

[0029] Based on a preset template, a reasoning process description is determined according to the trigger event, the current perception information, the current state information and the target state information, and the reasoning process description is used to describe the process by which the large language model determines the vehicle control instructions and the operation suggestions according to the trigger event.

[0030] In this way, based on the preset template, the server determines the reasoning process description according to the trigger event, current perception information, current state information and target state information. The reasoning process description is used to describe the process of the large language model determining the vehicle control instructions and operation suggestions according to the trigger event. In this way, through the determined reasoning process description, the correctness of the reasoning ideas of the large language model can be judged, potential improvement points can be identified, and subsequent optimization of the large language model can be facilitated.

[0031] In some embodiments, the operation suggestion includes a first operation suggestion, a second operation suggestion, and a third operation suggestion, wherein the first operation suggestion is used to indicate the vehicle operation instruction that needs to be executed immediately, the second operation suggestion is used to indicate the vehicle operation instruction that requires user confirmation whether to execute, and the third operation suggestion is used to indicate the vehicle operation instruction that does not need to be executed.

[0032] In this way, the operation suggestion includes a first operation suggestion, a second operation suggestion, and a third operation suggestion, wherein the first operation suggestion is used to indicate a vehicle operation instruction that needs to be executed immediately, the second operation suggestion is used to indicate a vehicle operation instruction that requires the user to confirm whether to execute, and the third operation suggestion is used to indicate a vehicle operation instruction that does not need to be executed. In this way, by distinguishing different types of operation suggestions, it is possible to ensure that emergency operations are executed immediately, thereby improving the safety of the vehicle. In addition, by allowing users to participate in the decision-making process, the system can better meet the personalized needs of users and improve user satisfaction.

[0033] In some embodiments, the sending the vehicle control instruction and the operation suggestion to the vehicle to complete the vehicle control includes:

[0034] In a case where the operation suggestion is the first operation suggestion or the second operation suggestion, the vehicle control instruction and the operation suggestion are sent to the vehicle to complete the vehicle control.

[0035] In this way, when the operation suggestion is the first operation suggestion or the second operation suggestion, the server sends the vehicle control instruction and the operation suggestion to the vehicle to complete the vehicle control. In this way, by sending the vehicle control instruction and the operation suggestion, the vehicle control can be completed quickly and the vehicle control efficiency can be improved. In addition, by sending the operation suggestion, it can help the user understand which operations need to be performed and confirm whether to perform them, thereby improving the user experience.

[0036] An embodiment of the present application provides a server, which includes a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the above method is implemented.

[0037] The embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0038] Additional aspects and advantages of the embodiments of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the embodiments of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0040] Figure 1 It is one of the flowcharts of the vehicle control method of certain embodiments of the present application;

[0041] Figure 2 This is a second flow chart of a vehicle control method according to certain embodiments of the present application;

[0042] Figure 3 This is a third flow chart of a vehicle control method according to certain embodiments of the present application;

[0043] Figure 4 This is a fourth flow chart of a vehicle control method according to certain embodiments of the present application;

[0044] Figure 5 This is a fifth flow chart of a vehicle control method according to certain embodiments of the present application;

[0045] Figure 6 This is the sixth flow chart of the vehicle control method of certain embodiments of the present application;

[0046] Figure 7 This is the seventh flow chart of the vehicle control method of certain embodiments of the present application. DETAILED DESCRIPTION

[0047] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals represent the same or similar elements or elements having the same or similar functions from beginning to end. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the embodiments of the present application, and cannot be understood as limiting the embodiments of the present application.

[0048] Currently, users interact with in-car assistants to achieve convenient vehicle control functions. Users can use in-car assistants to control vehicle equipment, such as adjusting air conditioning temperature, playing music, navigation, etc., thereby improving driving convenience and safety. However, in actual applications, in-car assistants have the defect of insufficient reasoning ability, especially when it comes to vertical field proprietary knowledge.

[0049] First, the in-car assistant has defects in understanding vertical field proprietary knowledge. Vehicle control involves knowledge in many professional fields, such as automotive engineering, electronic technology, communication protocols, etc. The special terms and concepts in these fields may be unfamiliar to the in-car assistant, which makes it difficult to understand and reason about this knowledge. For example, when a user asks about certain technical parameters or functions of the vehicle, the in-car assistant may not accurately understand the user's intention and cannot provide the correct answer or operation suggestions.

[0050] Secondly, the in-car assistant also has the problem of insufficient reasoning ability when processing business-specific terms. Vehicle control involves many specific business scenarios and operations, such as air conditioning temperature adjustment, seat position adjustment, navigation destination setting, etc. These business scenarios and operations usually involve some specific terms and terms, and the in-car assistant may have difficulty understanding these terms and terms. For example, when the user uses some professional vehicle control instructions or queries, the in-car assistant may not be able to accurately understand the user's intentions, generate correct vehicle control instructions, or provide accurate operation suggestions.

[0051] Based on the above questions, please refer to Figure 1 , the embodiment of the present application provides a vehicle control method, the method comprising:

[0052] 011: Get trigger events and current perception information;

[0053] 012: Based on the pre-configured large language model and the pre-built function point knowledge base, according to the trigger event and the current perception information, determine the target function point associated with the trigger event, the current state information of the target function point, and the target state information;

[0054] 013: Determine a vehicle control command and an operation suggestion of the vehicle control command according to the current state information and the target state information;

[0055] 014: Send vehicle control instructions and operation suggestions to the vehicle to complete vehicle control.

[0056] The embodiment of the present application also provides a server, including a memory and a processor. The vehicle control method of the embodiment of the present application can be implemented by the server of the embodiment of the present application. Specifically, a computer program is stored in the memory, and the processor is used to obtain trigger events and current perception information. And based on the pre-configured large language model and the pre-built function point knowledge base, according to the trigger event and the current perception information, the target function point associated with the trigger event, the current state information and the target state information of the target function point are determined. The processor is also used to determine the vehicle control instructions and the operation suggestions of the vehicle control instructions based on the current state information and the target state information. And the vehicle control instructions and operation suggestions are sent to the vehicle to complete the vehicle control.

[0057] The embodiment of the present application also provides a vehicle control device. The vehicle control method of the embodiment of the present application can be implemented by the vehicle control device of the embodiment of the present application. Specifically, the vehicle control device includes an acquisition module, a determination module and a sending module. The acquisition module is used to acquire trigger events and current perception information. The determination module is used to determine the target function point associated with the trigger event, the current state information of the target function point, and the target state information based on the trigger event and the current perception information based on a pre-configured large language model and a pre-built function point knowledge base. The determination module is also used to determine the vehicle control instructions and the operation suggestions of the vehicle control instructions based on the current state information and the target state information. The sending module is used to send the vehicle control instructions and the operation suggestions to the vehicle to complete the vehicle control.

[0058] Specifically, a trigger event can be understood as a user's operation or request on the vehicle, which can be explicit, such as user voice commands, touch screen operations, etc.; it can also be implicit, such as the vehicle detecting specific environmental changes or user behavior.

[0059] Current perception information refers to data collected by various sensors and systems of the vehicle, which is used to understand and respond to user needs and the environment around the vehicle, including user perception information and environmental perception information.

[0060] User perception information refers to data and information about vehicle users (such as drivers and passengers) collected by the vehicle through its sensors and systems, including user behavior. User behavior refers to the user's operation and behavior patterns, such as adjusting seats, switching driving modes, using touch screens, etc. In some embodiments, user perception information also includes user physiological information, such as heart rate, facial expressions, etc., which can help the system assess the user's emotions and comfort.

[0061] Environmental perception information refers to the data and information about the vehicle and the surrounding environment collected by the vehicle through its sensors, cameras and other detection equipment, including the state perception information of the functions of the on-board equipment and the current environmental status information of the vehicle. State perception information involves the functional status of various devices inside the vehicle, such as whether information systems such as the navigation system, entertainment system, and communication system are running, and their current settings and status. Or whether comfort systems such as air conditioning, seat heating / ventilation, and window control are activated, and their current temperature, wind speed and other settings. It can also be whether driving assistance systems such as adaptive cruise control, lane keeping assist, and automatic parking systems are enabled, as well as their current parameters and status. Environmental status information involves traffic conditions, weather conditions, and road conditions.

[0062] The pre-configured large language model refers to a pre-trained large language model that can analyze the characteristics of the current scene and user needs based on triggering events and current perception information, and determine the target functional points and subsequent processing to achieve user needs.

[0063] The function point knowledge base refers to a pre-built database containing the function points and related information of each vehicle device. It can provide the large language model with an understanding of the functions and characteristics of each vehicle device, and is an important basis for the model to make reasoning and decisions.

[0064] The target function point refers to the function point that best meets the user's intention or needs, determined by the trigger event and the current perception information, and can meet the user's current needs and optimize the vehicle's performance and the passenger experience. The large language model will generate corresponding vehicle control instructions based on the target function point, current state information, and target state information to achieve user needs or optimize the user's driving experience.

[0065] The current state information of the target function point refers to the description of the state of the target function point at the current moment, such as the current air conditioning temperature, the current air conditioning air volume, the current audio volume, the current audio playing track, etc. The current state information can help the large language model understand the current state of the vehicle, such as whether the air conditioning is turned on, the audio volume, etc., so as to avoid repeated operations. For example, if the air conditioning is already turned on, there is no need to turn it on again.

[0066] The target state information of the target function point refers to the state that the model expects the target function point to achieve, such as target air conditioning temperature, target air conditioning air volume, target audio volume, target audio playback tracks, etc. The target state information is an important basis for the large language model to generate vehicle control commands. For example, to set the air conditioning temperature to 25°C, the corresponding control command needs to be sent to the air conditioning system.

[0067] Vehicle control instructions refer to the control signals sent by the large language model to the vehicle equipment based on the inference results, which are used to change the state of the target functional point to achieve the target state.

[0068] The operation suggestion of the vehicle control instruction refers to the confidence level of the large language model in the generated vehicle control instruction, which reflects the trust level of the large language model in the current reasoning result. The operation confidence level is divided into three levels: immediate execution, recommended execution, and keeping the current state unchanged. Among them, immediate execution means that the large language model is very confident in the vehicle control instruction and believes that executing the instruction immediately is the best choice. Recommended execution means that the large language model has a certain confidence in the vehicle control instruction, but still needs user confirmation before execution. Keeping the current state means that the large language model believes that the current state has met the user's needs and no operation is required. The operation suggestion of the vehicle control instruction can reduce the risk of misoperation of the large language model. For example, if the confidence level of the vehicle control instruction is low, the large language model will not directly execute the vehicle control instruction, but will make operation suggestions to the user, and the user will ultimately decide whether to execute it.

[0069] The system collects user operations and real-time information about the vehicle and environment, namely trigger events and current perception information. For example, the trigger event is obtained: the user connects to a Bluetooth phone. The current perception information is obtained: the temperature inside the car is 28 degrees Celsius and the temperature outside the car is 30 degrees Celsius.

[0070] Next, the server determines the target function point and state information. Based on the trigger event and current perception information, the server combines the pre-configured large language model and function point knowledge base to determine the target function point related to the trigger event, as well as the current state information and target state information of the target function point. Continuing with the above example, the target function point determined by the large language model is "air conditioning switch, window switch and multimedia volume control", and the corresponding current state information is obtained, and the corresponding target state information is generated.

[0071] Then, the server will determine the vehicle control command and the corresponding operation suggestion. The server will generate the vehicle control command and give the operation suggestion based on the current state and target state of the function point. Continuing with the above example, based on the determined target function point "window switch", the current state information "window open" and the target state information "window closed", the vehicle control command "close window" is determined, and the corresponding operation suggestion is "recommend execution".

[0072] Finally, the server will send the vehicle control instructions and corresponding operation suggestions. The server will send the vehicle control instructions and corresponding operation suggestions to the vehicle, and the vehicle will perform the corresponding operations to complete the vehicle control.

[0073] In summary, the server obtains the trigger event and the current perception information. Then, based on the pre-configured large language model and the pre-built function point knowledge base, the server determines the target function point associated with the trigger event, the current state information of the target function point, and the target state information according to the trigger event and the current perception information. Then, the server determines the vehicle control instructions and the operation suggestions for the vehicle control instructions based on the current state information and the target state information. Finally, the server sends the vehicle control instructions and the operation suggestions to the vehicle to complete the vehicle control. In this way, the server actively provides services to the user based on the trigger event and the current perception information. Moreover, through the pre-built function point knowledge base, it is possible to accurately determine the target function points related to the trigger event and the vehicle control instructions that improve the user experience, thereby enhancing the user experience.

[0074] See also Figure 2 In some implementations, the function point knowledge base is constructed by the following steps:

[0075] 021: According to the vehicle function points, determine the associated attributes and function point operations of the vehicle function points;

[0076] 022: Determine the functional description information of the function point operation according to the associated attributes and the function point operation;

[0077] 023: Build a function point knowledge base based on associated attributes, function point operations and function description information.

[0078] In some embodiments, the determination module is further used to determine the associated attributes and function point operations of the vehicle function points according to the vehicle function points, and to determine the functional description information of the function point operations according to the associated attributes and the function point operations, and to construct a function point knowledge base according to the associated attributes, the function point operations and the functional description information.

[0079] In some embodiments, the processor is further configured to determine, based on the vehicle function points, associated attributes and function point operations of the vehicle function points, determine functional description information of the function point operations based on the associated attributes and function point operations, and construct a function point knowledge base based on the associated attributes, function point operations, and functional description information.

[0080] Specifically, the associated attributes of a vehicle function point refer to the characteristics or parameters of the vehicle function point, that is, which aspects of the user or physical environment the function point can affect. For example, the associated attributes of the vehicle function point "steering wheel heating" include temperature and comfort.

[0081] The function point operation of a vehicle function point refers to an action or command to change the vehicle function point, which can be used to change the state or behavior of the function point. For example, the function point operation of the vehicle function point "steering wheel heating" is "turn off the steering wheel heating, turn on the steering wheel heating, adjust the steering wheel heating to level 1, adjust the steering wheel heating to level 2, and adjust the steering wheel heating to level 3".

[0082] The functional description information of the function point operation refers to the detailed description of the function point operation, which explains the specific meaning and applicable scenarios of the function point operation, and helps the large language model understand the function point operation and make correct decisions. For example, the functional description information of the function point operation "turn off the steering wheel heating" is "Definition: turn off the steering wheel heating function and stop heating the steering wheel. Applicable scenarios: [In warm seasons or when the ambient temperature is high, there is no need to heat the steering wheel to save energy and avoid overheating.]". The functional description information of the function point operation "turn on the steering wheel heating" is "Definition: start the steering wheel heating function and start heating the steering wheel. Applicable scenarios: [In cold winter or low temperature environments, heating the steering wheel can improve the driver's comfort and grip, keep fingers flexible, and improve driving safety.]". The functional description information of the function point operation "steering wheel heating adjusted to 1st gear" is "Definition: set the steering wheel heating to a low gear to provide a gentle heating effect. Applicable scenarios: [When the temperature is not extremely low, the lowest heating level can keep the steering wheel warm without overheating; when driving for a long time, avoid discomfort caused by long-term high-heat heating.]". The functional description information of the function point operation "Adjust the steering wheel heating to level 2" is "Definition: Set the steering wheel heating to a medium level to provide a moderate heating effect. Applicable scenarios: [In colder weather, provide enough heat to keep the steering wheel comfortable.]". The functional description information of the function point operation "Adjust the steering wheel heating to level 3" is "Definition: Set the steering wheel heating to a high level to provide a stronger heating effect. Applicable scenarios: [In extremely cold conditions, provide maximum heating to ensure that the steering wheel quickly reaches a comfortable temperature; especially when the driver needs to feel the warmth of the steering wheel quickly.]".

[0083] Firstly, according to the functional points of the vehicle, the associated attributes and functional point operations of each vehicle functional point are determined.

[0084] Next, the functional description information of each function point operation needs to be determined based on the associated attributes and function point operations.

[0085] Finally, a function point knowledge base is constructed based on the associated attributes, function point operations and function description information. In some implementations, the function point knowledge base can be stored in a format such as JSON to facilitate system query and call.

[0086] In some implementations, the data information in the constructed function point knowledge base may include:

[0087] "Steering wheel heating":{

[0088] "Relation Attributes":[

[0089] "Temperature", "Comfort"

[0090] ],

[0091] "selection":[

[0092] "Turn off steering wheel heating",

[0093] "Turn on steering wheel heating",

[0094] "Steering wheel heating set to level 1",

[0095] "Steering wheel heating set to level 2",

[0096] "Steering wheel heating set to level 3"

[0097] ],

[0098] "Turn off steering wheel heating":{

[0099] "Definition": "Turn off the steering wheel heating function and stop heating the steering wheel.",

[0100] "Applicable scenarios":[

[0101] "During warm seasons or when ambient temperatures are high, the heated steering wheel is not required to save energy and avoid overheating." ]

[0103] },

[0104] "Turn on steering wheel heating":{

[0105] "Definition":"Start the steering wheel heating function and start heating the steering wheel.",

[0106] "Applicable scenarios":[

[0107] "In cold winter or low-temperature environments, a heated steering wheel can improve the driver's comfort and grip, keep fingers flexible, and improve driving safety." ]

[0109] },

[0110] "Steering wheel heating set to level 1":{

[0111] "Definition":"Set the steering wheel heating to low setting, providing gentle heating.",

[0112] "Applicable scenarios":[

[0113] "When the temperature is not extremely low, the lowest heating level can keep the steering wheel warm but not overheat.",

[0114] "When driving for long periods of time, avoid prolonged high heat heating that may cause discomfort." ]

[0116] },

[0117] "Steering wheel heating set to level 2":{

[0118] "Definition":"Set the steering wheel heating to medium level, providing moderate heating effect.",

[0119] "Applicable scenarios":[

[0120] "Provides enough heat to keep the steering wheel comfortable in colder weather." ]

[0122] },

[0123] "Steering wheel heating set to level 3":{

[0124] "Definition":"Set the steering wheel heating to a high position to provide a stronger heating effect.",

[0125] "Applicable scenarios":[

[0126] "Provides maximum heating in extremely cold conditions, ensuring the steering wheel quickly reaches a comfortable temperature.",

[0127] "Especially in situations where the driver needs to feel the warmth of the steering wheel quickly." ]

[0129] }

[0130] In this way, the function point knowledge base is constructed through the following steps: First, the server determines the associated attributes and function point operations of the vehicle function points based on the vehicle function points. Next, the server determines the functional description information of the function point operations based on the associated attributes and function point operations. Finally, the server constructs the function point knowledge base based on the associated attributes, function point operations and functional description information. In this way, the function point knowledge base can provide detailed information for each function point, including associated attributes, function point operations and functional description information, thereby helping the large language model to accurately understand the function and purpose of each function point, thereby enhancing the user experience. In addition, the function point knowledge base can help the large language model cope with a variety of different scenarios and situations, thereby improving the robustness of the large language model.

[0131] See also Figure 3In some embodiments, the large language model includes a preset prompt word. Step 012 (determining the target function point associated with the trigger event, the current state information of the target function point, and the target state information according to the trigger event and the current perception information based on the pre-configured large language model and the pre-built function point knowledge base) includes:

[0132] 0121: Based on the preset prompt word and according to the triggering event, the target function point, the target function point operation corresponding to the target function point and the associated background knowledge are obtained from the function point knowledge base;

[0133] 0122: Determine the current state information based on the target function point and current perception information.

[0134] In some embodiments, the acquisition module is further used to acquire the target function point, the target function point operation corresponding to the target function point, and the associated background knowledge from the function point knowledge base based on the preset prompt word and the trigger event. The determination module is further used to determine the current state information based on the target function point and the current perception information.

[0135] In some embodiments, the processor is further configured to obtain a target function point, a target function point operation corresponding to the target function point, and associated background knowledge from a function point knowledge base based on a preset prompt word and a trigger event, and determine current state information based on the target function point and current perception information.

[0136] Specifically, the preset prompt words refer to the text that guides the large language model to perform reasoning. It tells the large language model what role to play, how to process trigger events and perception information, and how to generate reasoning results. Specifically, the preset prompt words include role settings, task objectives, reasoning requirements, and operation requirements. In some embodiments, the preset prompt words may be:

[0137] You will play the role of a personal assistant in a car scene. Based on the trigger event and current perception information, you will determine whether the vehicle equipment needs to be adjusted and complete the equipment status table. Please note that all equipment changes need to revolve around <trigger event> as the main goal.

[0138] ##Triggering Events

[0139] <Trigger Event>

[0140] ###Current Perception Information

[0141] <Current Perception Information>

[0142] ##Device Status Table

[0143] <target function point>, <target function point operation>, <current status information>, <target status information>, <operation suggestion>, <reasoning process description>

[0144] Background

[0145] <Related background knowledge>

[0146] ##Require

[0147] <Requirements>

[0148] ###Operation Confidence

[0149] 1. You can only select one of the three values ​​["Execute immediately", "Suggest execution", "Keep current status"] as the operation confidence

[0150] 2. "Immediate execution" means to perform the operation on the user immediately

[0151] 3. "Suggested execution" means giving the user an operation suggestion during the <thinking process>, and the user can choose whether to perform the operation

[0152] 4. "Keep current state" means no operation

[0153] Thought Process

[0154] 1. In the <trigger event>, based on the <environmental information>, <vehicle information>, <user information>, etc., determine the current state, target state, and operation confidence of each device in the <device state table>, and give the <thinking process>

[0155] Target Status

[0156] 1. If the current state is considered perfect, the target state can be consistent with the current state.

[0157] The target function point operation refers to the operation options that can be performed on the target function point, such as "open", "close", "adjust temperature", etc.

[0158] Related background knowledge refers to the background information related to the target function point. For example, if the trigger event is "the user connects to a Bluetooth call", the related background knowledge may be "1. Media volume 8 and 9 are comfortable, and the volume is noisy after 14. 2. Navigation, voice, and call volume 15 is comfortable, and the volume above 20 is noisy. 3. Air conditioning volume, 1st gear: suitable for driving at night when you need to slightly adjust the temperature in the car and want to keep the car quiet to avoid strong winds disturbing sleep or rest; 2nd to 5th gear: suitable for daily driving, when you need to adjust the temperature in the car moderately without generating too much noise; 6th to 8th gear: suitable for hot or cold weather, when you need to adjust the temperature in the car quickly; 9th to 10th gear: suitable for extreme weather conditions, such as extreme heat or cold, when you need to quickly improve the temperature conditions in the car, such as when the car suddenly overheats or overcools."

[0159] The large language model will obtain the target function points, corresponding function point operations and related background knowledge associated with the trigger event from the function point knowledge base based on the preset prompt word template and the trigger event. Continuing with the above example, for the trigger event "the user connects to a Bluetooth call", the large language model will obtain the function points related to the associated attribute "noise and comfort" "air conditioning switch, window switch, air conditioning volume, playback control, headrest mode, Bluetooth call volume, multimedia volume and navigation broadcast volume", as well as the operations corresponding to these function points "air conditioning switch: 'turn on air conditioning', 'turn off air conditioning'", "window switch: 'open windows', 'close windows', 'open windows 10%', 'open windows 30%' , 'Open the windows 50%', 'Open the windows 70%', 'Open the windows 90%'" and "Air conditioning air volume: 'Adjust the air conditioning air volume to level 1', 'Adjust the air conditioning air volume to level 2', 'Adjust the air conditioning air volume to level 3', 'Adjust the air conditioning air volume to level 4', 'Adjust the air conditioning air volume to level 5', 'Adjust the air conditioning air volume to level 6', 'Adjust the air conditioning air volume to level 7', 'Adjust the air conditioning air volume to level 8', 'Adjust the air conditioning air volume to level 9', 'Adjust the air conditioning air volume to level 10'", etc. The function point operations of other target function points will not be repeated here.

[0160] Next, the large language model determines the current state information of these function points based on the target function points and the current perception information. Continuing with the above example, the current state information of the target function point "air conditioning switch" is "air conditioning on". The current state information of the target function point "window switch" is "["driver's window closed", "passenger's window closed", "left rear window closed", "right rear window closed"]". The current state information of the target function point "air conditioning air volume" is "air conditioning air volume adjusted to level 6".

[0161] In this way, based on the preset prompt words, the server obtains the target function point, the target function point operation corresponding to the target function point, and the associated background knowledge from the function point knowledge base according to the trigger event. Then, the server determines the current state information according to the target function point and the current perception information. In this way, by pre-setting the preset prompt words, the large language model can be guided to use its own powerful semantic understanding and reasoning capabilities to accurately obtain the required data information from the function point knowledge base according to the trigger event, thereby accurately generating vehicle control instructions that can improve the user experience.

[0162] See also Figure 4 In some implementations, step 0121 (obtaining a target function point, a target function point operation corresponding to the target function point, and associated background knowledge from a function point knowledge base based on a preset prompt word and a triggering event) includes:

[0163] 01211: Determine user intent based on triggering events;

[0164] 01212: Determine target-related attributes based on user intent;

[0165] 01213: acquiring, from a function point knowledge base according to a target association attribute, a target function point associated with the triggering event, a target function point operation corresponding to the target function point, and target function description information corresponding to the target function point operation;

[0166] 01214: Determine relevant background knowledge based on target function description information.

[0167] In some embodiments, the determination module is further used to determine the user intention according to the triggering event. And determine the target associated attribute according to the user intention. The determination module is also used to obtain the target function point associated with the triggering event, the target function point operation corresponding to the target function point, and the target function description information corresponding to the target function point operation from the function point knowledge base according to the target associated attribute. And determine the associated background knowledge according to the target function description information.

[0168] In some embodiments, the processor is further configured to determine the user intention according to the triggering event, and determine the target associated attribute according to the user intention. The processor is further configured to obtain the target function point associated with the triggering event, the target function point operation corresponding to the target function point, and the target function description information corresponding to the target function point operation from the function point knowledge base according to the target associated attribute, and determine the associated background knowledge according to the target function description information.

[0169] Specifically, user intent refers to the goal or need that the user wants to achieve inferred from the triggering event and current perceived information.

[0170] Goal-related attributes refer to device attributes that are related to user intent and triggering events, and are usually related to user goals or expectations, such as comfort, temperature, noise, etc.

[0171] The server analyzes the user's intention based on the trigger event. For example, for the trigger event "user connects to a Bluetooth phone call", the user's intention may be to hear the call content clearly.

[0172] Next, the server determines the associated attributes related to the target function point according to the user's intention. For example, for call clarity, the associated attributes may include volume, noise, etc.

[0173] Then, the server will obtain the target function point associated with the triggering event, the corresponding function point operation and function description information from the function point knowledge base according to the target association attribute.

[0174] Finally, the system will determine the associated background knowledge related to the target function point operation based on the function description information. For example, for volume adjustment, the associated background knowledge may include the volume comfort range, the appropriate volume value in different scenarios, etc.

[0175] In this way, the server determines the user's intention based on the triggering event. Next, the server determines the target-related attributes based on the user's intention. Then, the server obtains the target function point associated with the triggering event, the target function point operation corresponding to the target function point, and the target function description information corresponding to the target function point operation from the function point knowledge base based on the target-related attributes. Finally, the server determines the associated background knowledge based on the target function description information. In this way, by accurately understanding the user's intention and determining the target-related attributes, the target function point associated with the triggering event can be accurately determined from the function point knowledge base, reducing the number of function points that need to be inferred by the large language model, speeding up the inference speed of the large language model, and thus enhancing the user experience.

[0176] See also Figure 5 In some embodiments, step 012 (determining the target function point associated with the triggering event, the current state information of the target function point, and the target state information according to the triggering event and the current perception information based on the pre-configured large language model and the pre-built function point knowledge base) includes:

[0177] 0123: Based on the preset prompt words, the target function point operation and related background knowledge are used to determine the target status information.

[0178] In some implementations, the determination module is further configured to determine target state information based on preset prompt words, target function point operations, and associated background knowledge.

[0179] In some embodiments, the processor is further configured to determine target state information based on preset prompt words, target function point operations and associated background knowledge.

[0180] Specifically, based on the preset prompt word, the server determines the target state information according to the target function point operation and the associated background knowledge. For example, if the target function point operation is multimedia volume adjustment, the server determines whether the current multimedia volume is appropriate based on the background knowledge and determines the multimedia volume value to be adjusted.

[0181] The following is a new example to illustrate the vehicle control method of the implementation mode of the present application. The scenario is "the user answers a Bluetooth call, the temperature inside the car is 28 degrees Celsius, the temperature outside the car is 30 degrees Celsius, the multimedia volume is 10, and the Bluetooth call volume is 20".

[0182] The server obtains the trigger event "user connects to the Bluetooth call" and the current perception information "the temperature inside the car is 28 degrees Celsius, the temperature outside the car is 30 degrees Celsius, the multimedia volume is 10, and the Bluetooth call volume is 20".

[0183] Next, based on the pre-configured large language model and the pre-built function point knowledge base, the target function points "multimedia volume, Bluetooth call volume" and related background knowledge "media volume 8, 9 is comfortable, and the volume is too noisy after 14; navigation, voice, and call volume 15 is comfortable, and the volume greater than 20 is too noisy" are obtained.

[0184] Then, the current state information of the target function point "multimedia volume is 5, Bluetooth call volume is 20" is determined.

[0185] Subsequently, the large language model determines that the current multimedia volume is too low and needs to be adjusted to 8 or 9, and the Bluetooth call volume is too high and needs to be adjusted to 15. It also generates vehicle control instructions and corresponding operation suggestions: "The current multimedia volume is judged to be too high and needs to be adjusted to 8 or 9 immediately (immediate execution); the Bluetooth call volume is too high and it is recommended to adjust it to 15 (recommended execution)".

[0186] Finally, vehicle control instructions will be sent to the vehicle, which will automatically adjust the multimedia volume and ask the user for opinion on adjusting the Bluetooth call volume to improve the call experience.

[0187] In this way, based on the preset prompt words, the server determines the target state information according to the target function point operation and the associated background knowledge. In this way, according to the preset prompt words, the target function point operation and the associated background knowledge, the target state information of the target function point is accurately determined, the operation efficiency is improved, and the user experience is enhanced.

[0188] See also Figure 6 In some embodiments, the method further comprises:

[0189] 015: Based on the preset template, determine the reasoning process description according to the trigger event, current perception information, current state information and target state information.

[0190] In some embodiments, the determination module is further used to determine the reasoning process description based on a preset template, according to the triggering event, current perception information, current state information and target state information.

[0191] In some embodiments, the processor is further configured to determine a reasoning process description based on a preset template, according to a triggering event, current perception information, current state information, and target state information.

[0192] Specifically, the reasoning process description refers to how, after a triggering event occurs, the large language model determines whether the vehicle function points need to be adjusted based on the current perception information and function point database, and gives adjustment suggestions or performs operations.

[0193] The server will determine the current state information and target state information of the target function point based on the preset prompt word template, combined with the trigger event and the current perception information. For example, for the trigger event of "user connects to a Bluetooth call", the system will obtain the current multimedia volume and Bluetooth call volume, and determine the volume value that needs to be adjusted.

[0194] Next, the server generates a reasoning process description based on the preset prompt word template, combined with the trigger event, current perception information, current state information and target state information. In some embodiments, the reasoning process description is output to the user so that the user can understand how the system makes decisions.

[0195] Continuing with the above example, the generated reasoning process description may be "The user makes a Bluetooth call, the current multimedia volume is 10, and the Bluetooth call volume is 20. According to the associated background knowledge, the media volume of 8 and 9 is comfortable, and the volume is too loud after 14; the navigation, voice, and call volume of 15 is comfortable, and the volume greater than 20 is too loud. Therefore, the system recommends adjusting the multimedia volume to 8 or 9 and the Bluetooth call volume to 15 to improve the call experience."

[0196] In this way, based on the preset template, the server determines the reasoning process description according to the trigger event, current perception information, current state information and target state information. The reasoning process description is used to describe the process of the large language model determining the vehicle control instructions and operation suggestions according to the trigger event. In this way, through the determined reasoning process description, the correctness of the reasoning ideas of the large language model can be judged, potential improvement points can be identified, and subsequent optimization of the large language model can be facilitated.

[0197] In some embodiments, the operation suggestion includes a first operation suggestion, a second operation suggestion, and a third operation suggestion, wherein the first operation suggestion is used to indicate a vehicle operation instruction that needs to be executed immediately, the second operation suggestion is used to indicate a vehicle operation instruction that requires user confirmation whether to execute, and the third operation suggestion is used to indicate a vehicle operation instruction that does not need to be executed.

[0198] Specifically, the system will determine the type of each action suggestion based on the target state information and the current state information. If the target state information is different from the current state information, an action needs to be performed, which can be divided into three types according to the urgency of the action:

[0199] First action suggestion (execute immediately): When the action can improve the user experience and will not affect the user's safety, the system will generate a first action suggestion, which needs to be executed immediately. For example, when the temperature outside the car is too high and the air conditioner needs to be turned on immediately, the system will generate a first action suggestion.

[0200] Second operation suggestion (recommended execution): When the operation can improve the user experience but will affect the user's safety, the system will generate a second operation suggestion and require the user to confirm whether to execute it. For example, when the user drives for a long time and the system recommends turning on the seat ventilation, the system will generate a second operation suggestion.

[0201] Third operation suggestion (no need to execute): When the operation does not need to be executed, the system will generate a third operation suggestion. For example, when the air conditioner is already turned on, the system will generate a third operation suggestion.

[0202] In this way, the operation suggestion includes a first operation suggestion, a second operation suggestion, and a third operation suggestion, wherein the first operation suggestion is used to indicate a vehicle operation instruction that needs to be executed immediately, the second operation suggestion is used to indicate a vehicle operation instruction that requires the user to confirm whether to execute, and the third operation suggestion is used to indicate a vehicle operation instruction that does not need to be executed. In this way, by distinguishing different types of operation suggestions, it is possible to ensure that emergency operations are executed immediately, thereby improving the safety of the vehicle. In addition, by allowing users to participate in the decision-making process, the system can better meet the personalized needs of users and improve user satisfaction.

[0203] See also Figure 7 In some embodiments, step 014 (issuing vehicle control instructions and operation suggestions to the vehicle to complete vehicle control) includes:

[0204] 0141: When the operation suggestion is the first operation suggestion or the second operation suggestion, a vehicle control instruction and the operation suggestion are sent to the vehicle to complete vehicle control.

[0205] In certain embodiments, the sending module is used to send the vehicle control instruction and the operation suggestion to the vehicle to complete the vehicle control when the operation suggestion is the first operation suggestion or the second operation suggestion.

[0206] In some embodiments, the processor is further configured to, when the operation suggestion is the first operation suggestion or the second operation suggestion, send a vehicle control instruction and the operation suggestion to the vehicle to complete vehicle control.

[0207] Specifically, the server will decide whether to send the vehicle control instruction and the operation suggestion according to the type of the operation suggestion. If the operation suggestion is the first operation suggestion or the second operation suggestion, the system will send the vehicle control instruction and the operation suggestion to the vehicle. After the vehicle receives the vehicle control instruction, it will perform the corresponding operation to complete the vehicle control.

[0208] In some embodiments, based on the number of times the user adopts the vehicle control instructions, vehicle control instructions in the recommended execution category with a large number of adoptions are changed to immediate execution efficiency control instructions for immediate execution, and vehicle control instructions in the immediate execution category with a large number of rejections are changed to recommended execution.

[0209] In this way, when the operation suggestion is the first operation suggestion or the second operation suggestion, the server sends the vehicle control instruction and the operation suggestion to the vehicle to complete the vehicle control. In this way, by sending the vehicle control instruction and the operation suggestion, the vehicle control can be completed quickly and the vehicle control efficiency can be improved. In addition, by sending the operation suggestion, it can help the user understand which operations need to be performed and confirm whether to perform them, thereby improving the user experience.

[0210] The present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the vehicle control method described above are implemented.

[0211] It is understood that a computer program includes computer program code. The computer program code may be in source code form, object code form, executable file or some intermediate form. Computer readable storage media may include: any entity or device capable of carrying computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), and software distribution media.

[0212] In the description of this specification, the descriptions with reference to the terms "specifically", "further", "particularly", "understandably", etc. are intended to mean that the specific features, structures, materials or characteristics described in conjunction with the embodiments or examples are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms are not intended to refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0213] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code that includes one or more executable requests for implementing specific logical functions or steps of a process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0214] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A vehicle control method, characterized in that: The method comprises: Obtain trigger events and current perception information; Based on a pre-configured large language model and a pre-built function point knowledge base, according to the trigger event and the current perception information, determine a target function point associated with the trigger event, current state information of the target function point, and target state information; Determining a vehicle control instruction and an operation suggestion for the vehicle control instruction according to the current state information and the target state information; The vehicle control instruction and the operation suggestion are sent to the vehicle to complete the vehicle control.

2. The method according to claim 1, characterized in that The function point knowledge base is constructed by the following steps: Determining, according to the vehicle function points, associated attributes and function point operations of the vehicle function points; Determining functional description information of the function point operation according to the associated attribute and the function point operation; The function point knowledge base is constructed according to the associated attributes, the function point operations and the function description information.

3. The method according to claim 2, characterized in that The large language model includes a preset prompt word, and the method based on the pre-configured large language model and the pre-built function point knowledge base determines the target function point associated with the trigger event, the current state information of the target function point, and the target state information according to the trigger event and the current perception information, including: Based on the preset prompt word and according to the trigger event, acquiring the target function point, the target function point operation corresponding to the target function point and the associated background knowledge from the function point knowledge base; The current state information is determined according to the target function point and the current perception information.

4. The method according to claim 3, characterized in that The acquiring the target function point, the target function point operation corresponding to the target function point and the associated background knowledge from the function point knowledge base based on the preset prompt word and according to the trigger event includes: Determining user intent based on the trigger event; Determining target-related attributes according to the user intention; Acquire, from the function point knowledge base according to the target association attribute, a target function point associated with the trigger event, a target function point operation corresponding to the target function point, and target function description information corresponding to the target function point operation; The associated background knowledge is determined according to the target function description information.

5. The method according to claim 3, characterized in that: The method of determining a target function point associated with the trigger event, current state information of the target function point, and target state information based on the pre-configured large language model and the pre-built function point knowledge base according to the trigger event and the current perception information includes: Based on the preset prompt word, according to the target function point operation and the associated background knowledge, the target state information is determined.

6. The method according to claim 1, characterized in that The method further comprises: Based on a preset template, a reasoning process description is determined according to the trigger event, the current perception information, the current state information and the target state information, and the reasoning process description is used to describe the process by which the large language model determines the vehicle control instructions and the operation suggestions according to the trigger event.

7. The method according to claim 1, characterized in that The operation suggestions include a first operation suggestion, a second operation suggestion and a third operation suggestion, wherein the first operation suggestion is used to indicate the vehicle operation instruction that needs to be executed immediately, the second operation suggestion is used to indicate the vehicle operation instruction that requires user confirmation whether to execute, and the third operation suggestion is used to indicate the vehicle operation instruction that does not need to be executed.

8. The method according to claim 7, characterized in that The sending of the vehicle control instruction and the operation suggestion to the vehicle to complete the vehicle control includes: In a case where the operation suggestion is the first operation suggestion or the second operation suggestion, the vehicle control instruction and the operation suggestion are issued to the vehicle to complete the vehicle control.

9. A server, characterized in that: The server includes a processor and a memory, wherein a computer program is stored in the memory. When the computer program is executed by the processor, the method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images