Driving task searching method and device and intelligent vehicle

Abstract

The invention relates to a driving task searching method and device and an intelligent vehicle. The method comprises the following steps: acquiring a to-be-searched driving task by identifying a driving task input instruction of a vehicle passenger; analyzing the to-be-searched driving task by adopting a large language model to obtain a plurality of branch tasks associated with the to-be-searched driving task and a plurality of sub-tasks obtained by splitting each branch task; acquiring real-time search data returned by the search agent component for each branch task, and correcting the real-time search data of each branch task by calling a data error correction function of a large language model to obtain corrected search data of each branch task; outputting a search content feedback result according to the corrected search data of each branch task; and presenting a search content feedback result according to a preset vehicle-mounted information interaction mode. By adopting the method, complex task requirements can be deeply understood, task associated information fusion and adjustment are realized, and the task processing efficiency and accuracy in the driving process are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent vehicles, in particular to a driving task search method and device, an intelligent vehicle, a computer readable storage medium and a computer program product. BACKGROUND

[0002] With the development of intelligent vehicle technology, the demand for intelligent search and task processing of vehicle information by drivers is increasing.

[0003] In traditional technology, vehicle navigation or information search processing mainly realizes information query through keyword matching or simple voice interaction, but when facing complex and dynamically changing driving tasks, there are still many deficiencies, which cannot meet the real-time complex search needs in the driving process. For example, for complex or multi-step tasks proposed by the driver, the task understanding ability is poor; the search and decision-making process is too single, affecting the accuracy and timeliness of the search results; lack of task processing management ability, easy to cause information overload or information deficiency, affecting driving safety and user experience. SUMMARY

[0004] Therefore, it is necessary to provide a driving task search method, device, intelligent vehicle, computer readable storage medium and computer program product capable of optimizing vehicle intelligent search processing to solve the above technical problems.

[0005] In a first aspect, the present application provides a driving task search method, which comprises:

[0006] obtaining a driving task to be searched by recognizing a driving task input instruction of a vehicle occupant;

[0007] analyzing the driving task to be searched by using a large language model to obtain a plurality of branch tasks associated with the driving task to be searched, and a plurality of subtasks obtained by disassembling each branch task;

[0008] obtaining real-time search data returned by a search agent component for each branch task, correcting the real-time search data of each branch task by calling a data correction function of the large language model, and obtaining corrected search data of each branch task;

[0009] outputting a search content feedback result according to the corrected search data of each branch task;

[0010] presenting the search content feedback result in a preset vehicle information interaction mode.

[0011] In one embodiment, the large language model is used to parse the driving task to be searched, thereby obtaining multiple branch tasks associated with the driving task to be searched, and multiple subtasks obtained by decomposing each branch task, including:

[0012] triggering the large language model to decompose the driving task to be searched into the plurality of branch tasks, and the plurality of subtasks associated with each branch task;

[0013] Constructing a structured task tree based on the task relationships between the subtasks in each branch task;

[0014] The structured task tree adopts a tree structure, wherein the root node of the structured task tree is used to represent the driving task to be searched, and the multiple branches of the structured task tree are used to represent one of the branch tasks respectively; the child nodes in each branch are used to represent a subtask obtained by decomposing the task represented by the parent node.

[0015] In one embodiment, outputting search content feedback results based on the corrected search data of each branch task includes:

[0016] Optimizing a target node in the structured task tree according to the relationship between the branches in the structured task tree to obtain an optimized task tree; the task represented by the target node at least includes a task that affects the logical coherence between the subtasks;

[0017] Selecting search data associated with each node in the optimized task tree from the corrected search data of each branch task to obtain target search data;

[0018] The target search data is fused and processed, and the search content feedback result is output.

[0019] In one embodiment, after the step of constructing a structured task tree based on the task relationships between the subtasks, the method further includes:

[0020] Acquiring vehicle position information and driving environment information of the vehicle;

[0021] Determining a tree structure parameter of the structured task tree based on the vehicle position information and the driving environment information; the tree structure parameter includes at least one of the number of branches or the tree depth; the tree depth is used to represent the maximum level of nodes in the structured task tree;

[0022] The original tree structure of the structured task tree is updated according to the tree structure parameters to obtain the updated structured task tree.

[0023] In one embodiment, determining the tree structure parameters of the structured task tree according to the vehicle position information and the driving environment information includes:

[0024] estimating a remaining driving time for the vehicle to reach a destination indicated by the to-be-searched driving task based on the vehicle position information, and determining a driving idle time for the vehicle occupants based on the driving environment information;

[0025] When the remaining driving time is less than a preset first time threshold and the idle driving time is less than a preset second time threshold, the number of branches of the structured task tree is set from a preset first branch number to a second branch number; wherein the second branch number is less than the first branch number, and the second branch number is less than or equal to the preset number threshold.

[0026] In one embodiment, presenting the search content feedback result according to the vehicle information interaction mode preset by the vehicle includes:

[0027] Playing the search content feedback results through the voice broadcast function of the vehicle;

[0028] and / or,

[0029] The search content feedback result is displayed through an on-board display device of the vehicle.

[0030] In a second aspect, the present application further provides a driving task search device, the device comprising:

[0031] A driving task acquisition module is used to obtain the driving task to be searched by identifying the driving task input instructions of the vehicle occupants;

[0032] a task decomposition module, configured to parse the driving task to be searched using a large language model to obtain a plurality of branch tasks associated with the driving task to be searched, and a plurality of subtasks obtained by decomposing each branch task;

[0033] A search data correction module is configured to obtain the real-time search data returned by the search proxy component for each branch task, and correct the real-time search data of each branch task by calling the data error correction function of the large language model to obtain corrected search data for each branch task;

[0034] A feedback result output module, configured to output search content feedback results based on the corrected search data of each branch task;

[0035] The search content presentation module is used to present the search content feedback results according to a preset vehicle information interaction method.

[0036] In a third aspect, the present application further provides an intelligent vehicle, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0037] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor.

[0038] In a fifth aspect, the present application also provides a computer program product, comprising a computer program, which implements the steps of the above method when executed by a processor.

[0039] The aforementioned driving task search method, device, intelligent vehicle, computer-readable storage medium, and computer program product identify a vehicle occupant's input driving task command to obtain a driving task to be searched. The method then uses a large language model to parse the driving task to obtain multiple branch tasks associated with the task, as well as multiple subtasks derived from each branch task. The method then obtains real-time search data returned by the search agent component for each branch task. The method then uses the data correction function of the large language model to correct the real-time search data for each branch task, obtaining corrected search data for each branch task. The method then outputs search content feedback results based on the corrected search data for each branch task, presenting the search content feedback results in a pre-set in-vehicle information interaction method. In this way, by optimizing in-vehicle intelligent search processing, based on structured task decomposition and real-time information correction mechanisms, the method achieves a deep understanding of complex task requirements, implements multi-dimensional information fusion and dynamic adjustment of task associations, and effectively improves the efficiency and accuracy of task processing during driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 A diagram illustrating an application environment of a driving task search method according to an embodiment;

[0042] Figure 2 1 is a flow chart of a driving task search method according to an embodiment;

[0043] Figure 3 A schematic diagram of a driving task search process flow in one embodiment;

[0044] Figure 4 A schematic diagram of a task decomposition process in one embodiment;

[0045] Figure 5 A schematic flow chart of a driving task search method according to another embodiment;

[0046] Figure 6 is a structural block diagram of a driving task search device in one embodiment;

[0047] Figure 7 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0049] The collection and processing of the relevant data in this application should be strictly in accordance with the requirements of relevant national laws and regulations when applied in practice, and the informed consent or separate consent of the personal information subject should be obtained. Subsequent data use and processing should be carried out within the scope of authorization of laws and regulations and the personal information subject.

[0050] In this application, when the facial (or other biometric) recognition technology involved is applied to specific products or technologies in the above embodiments of this application, the relevant data collection, use and processing processes should comply with national laws and regulations. Before collecting facial information, the information processing rules should be informed and the target object's separate consent should be obtained. Facial information should be processed strictly in accordance with the requirements of laws and regulations and personal information processing rules, and technical measures should be taken to ensure the security of relevant data.

[0051] The driving task search method provided in the embodiment of the present application can be applied to Figure 1In the application environment shown, the driving task search method can be applied to an in-vehicle intelligent search system, which includes an intelligent vehicle 102 and a server 104, wherein the intelligent vehicle 102 communicates with the server 104 via a network. A data storage system can store data that the server 104 needs to process. The data storage system can be integrated with the server 104, or placed on a cloud or other network server. The intelligent vehicle 102 can obtain a driving task to be searched by identifying the driving task input instructions of the vehicle occupant. The intelligent vehicle 102 can communicate with the server 104 to use a large language model to parse the driving task to be searched, obtaining multiple branch tasks associated with the driving task to be searched, as well as multiple subtasks derived from each branch task. The intelligent vehicle 102 then obtains real-time search data returned by the search agent component for each branch task. By invoking the data error correction function of the large language model, the real-time search data of each branch task is corrected to obtain corrected search data for each branch task. Then, based on the corrected search data for each branch task, the intelligent vehicle 102 outputs search content feedback results. The intelligent vehicle 102 can present the search content feedback results according to a preset in-vehicle information interaction method. The server 104 may be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0052] In an exemplary embodiment, Figure 2 As shown, a driving task search method is provided, which is applied to Figure 1 Taking the smart vehicle 102 in the example as an example, the method includes the following steps S201 to S205.

[0053] Step S201 : obtaining a driving task to be searched by identifying a driving task input instruction of a vehicle occupant.

[0054] The driving tasks to be searched may be driving-related tasks that have not yet been executed in the vehicle but need to be processed, such as tasks of navigating to a specific location, arranging charging en route, or planning a trip.

[0055] In practical applications, such as Figure 3 As shown, in the user input stage, the on-board voice recognition unit of the intelligent vehicle recognizes the search questions or task requests raised by the driver (i.e., the driving task input instructions of the vehicle occupants), and the on-board intelligent search system can obtain the driving task to be searched.

[0056] In step S202 , the driving task to be searched is parsed using a large language model to obtain a plurality of branch tasks associated with the driving task to be searched, and a plurality of subtasks obtained by decomposing each branch task.

[0057] Among them, the in-vehicle intelligent search system is equipped with a large language model. In the field of natural language processing, the large language model has powerful language understanding, generation and reasoning capabilities, which can be applied to scenarios such as in-vehicle intelligent navigation and driving task planning.

[0058] In the specific implementation, such as Figure 3 As shown, for the driving task to be searched, structured task decomposition can be performed through a large language model to automatically decompose the complex or multi-step driving task to be searched into multiple branch tasks, and multiple subtasks associated with each branch task; optionally, a multi-step, multi-branch structured task tree can be generated based on the task relationship between the subtasks in each decomposed branch task, so as to further search and process based on the structured task tree.

[0059] In one example, when a vehicle occupant asks "navigate to a specific location A", the traditional search method only searches based on the keyword "specific location A". The technical solution of this embodiment can use a large language model to comprehensively analyze multiple related factors based on the identified driving task to be searched, "navigate to a specific location A", to perform task parsing. For example, it can associate branch tasks of "weather in location A", "road congestion near location A", and "parking lot in location A". It can also combine the real-time data collected by various sensors of the smart vehicle to perform task parsing. For example, if it is detected that the vehicle needs to be charged, the subtask of "having charging piles" can be disassembled under the branch task of "parking lot in location A". This makes it possible to achieve effective structured disassembly of complex or multi-step tasks and has in-depth task understanding capabilities.

[0060] In another example, when a vehicle occupant asks "find a Chinese restaurant", the traditional search method only searches based on the keyword "Chinese restaurant"; the technical solution of this embodiment can use a large language model to comprehensively analyze multiple related factors based on the identified driving task to be searched "find a Chinese restaurant" to perform task analysis. For example, based on the family travel itinerary of the smart vehicle, it can associate the branch task of "suitable for family gatherings", the branch task of "parking space available", and the branch task of "highly rated restaurants", etc., to improve the accuracy of the search results.

[0061] Step S203: obtain the real-time search data returned by the search proxy component for each branch task, and correct the real-time search data of each branch task by calling the data error correction function of the large language model to obtain the corrected search data of each branch task.

[0062] Specifically, if Figure 3As shown, for the real-time search data acquisition stage, the vehicle-mounted intelligent search system can call the search agent component, and based on the search agent component, further call other agents or services to acquire real-time search data according to the branch tasks determined by the structured task tree; for example, the map application interface, the weather application interface, the traffic information service, the charging station information service, etc. can be further called, and the acquired real-time application data or service information can be fed back to the large language model as real-time search data.

[0063] Then, as shown in FIG. 2B, the vehicle-mounted intelligent search system can further include a large language model, which can be configured to receive the initial search data returned by the search agent component, and based on the initial search data, analyze, infer and dynamically correct errors or outdated information in each branch task to accurately and refine the content of each branch task. Figure 3 As shown, for the task content optimization and error correction stage, the large language model can re-analyze, infer and dynamically correct errors or outdated information in each branch task according to the real-time search data returned by the search agent component, to accurately and refine the content of each branch task. For example, for the weather query branch task, different weather application interfaces can be called to obtain searched weather data from different information sources, and the large language model can determine that some weather data has errors or is outdated, and then correct the data to obtain corrected search data for the branch task; the large language model can also correct the preliminary thinking through task optimization processing, for example, it can be determined that the driver's demand has changed during the interaction, such as changing "find a gas station" to "find the nearest charging station" to achieve dynamic task optimization. Thus, based on the cooperative working mode of the search agent component and the large language model, the vehicle-mounted intelligent search system can have the ability to correct task content in real time and correct information, achieve dynamic content correction and optimization of the task, and ensure the real-time, accuracy and effectiveness of the search information.

[0064] In an optional embodiment, for the initially returned real-time search data, each branch task can also be externally called for real-time data verification and automatic adjustment to ensure that the provided information is accurate and effective.

[0065] Step S204, output the search content feedback result according to the corrected search data of each branch task.

[0066] In an example, since each branch task has multiple subtasks, each subtask is associated with corresponding corrected search data; specifically, the data integration function of the large language model can be called to fuse the corrected search data associated with each subtask to generate a complete natural language, which includes the corrected search data associated with each subtask; the above natural language can be output as the search content feedback result. As shown in FIG. 2B, the vehicle-mounted intelligent search system can further include a data integration component, which can be configured to call the data integration function of the large language model to fuse the corrected search data associated with each subtask to generate a complete natural language, which includes the corrected search data associated with each subtask; the above natural language can be output as the search content feedback result. Figure 3As shown, for intelligent branch relationship management, a large language model can be used to intelligently determine the relationships between branch tasks and automatically handle branch logic relationships, such as conflict, complementation, replacement, and independence, thereby improving search accuracy and ensuring the coherence of task logic. For dynamic branch optimization, the number or depth of branches can be dynamically optimized by combining the vehicle's location information and driving environment information. Using dynamic branch optimization strategies can reduce the user's cognitive burden, reduce the risk of information overload, and effectively improve driving safety.

[0067] Step S205: presenting the search content feedback result according to the preset vehicle information interaction mode.

[0068] As an example, Figure 3 As shown, in the final interactive output stage, after completing the processing of all branch tasks, the in-vehicle intelligent search system can clearly display the search content feedback results to the vehicle occupants in the form of voice broadcast or graphical interface (i.e., the preset in-vehicle information interaction method), thereby ensuring the efficiency and safety of information interaction during driving.

[0069] Compared with traditional methods, the technical solution of this embodiment provides a more efficient, accurate and intelligent driving task search method, which can realize automated problem decomposition, real-time information fusion and proactive search decision-making in vehicle intelligent interaction scenarios. It can effectively improve the depth and accuracy of the vehicle-mounted intelligent search system's understanding of complex or multi-stage tasks, reduce information misjudgment and redundancy, and enable the system to have an in-depth understanding and real-time analysis capability of complex, ambiguous or dynamically changing user needs, so as to better meet the search needs during driving.

[0070] In the aforementioned driving task search method, the driving task to be searched is obtained by identifying the vehicle occupant's driving task input command. A large language model is then used to parse the driving task to obtain multiple branch tasks associated with the task, as well as multiple subtasks derived from each branch task. The search agent component then retrieves real-time search data returned for each branch task. The large language model's data correction function is then invoked to correct the real-time search data for each branch task, resulting in corrected search data for each branch task. Based on the corrected search data for each branch task, search content feedback results are then output and presented according to a pre-defined in-vehicle information interaction method. In this manner, by optimizing in-vehicle intelligent search processing, based on structured task decomposition and real-time information correction mechanisms, a deep understanding of complex task requirements can be achieved, enabling multi-dimensional information fusion and dynamic adjustment of task associations, effectively improving the efficiency and accuracy of task processing during driving.

[0071] In an exemplary embodiment, using a large language model to parse a driving task to be searched, obtaining multiple branch tasks associated with the driving task to be searched, and multiple subtasks obtained by decomposing each branch task, may include the following steps:

[0072] The large language model is triggered to decompose the driving task to be searched into multiple branch tasks, and each branch task is associated with multiple subtasks; based on the task relationships between the subtasks in each branch task, a structured task tree is constructed.

[0073] Among them, the structured task tree adopts a tree structure. The root node of the structured task tree can be used to represent the driving task to be searched, and the multiple branches of the structured task tree can be used to represent a branch task respectively; the child node in each branch can be used to represent a subtask obtained by decomposing the task represented by the parent node.

[0074] In practical applications, such as Figure 4 As shown in , for the search task proposed by the user (i.e., the driving task to be searched), in the structured decomposition stage, the large language model can automatically decompose the complex or multi-step task into multiple branch tasks and their associated subtasks, such as Figure 4 The middle branch task 1 and its subtask 1-1, and branch task 2 and its subtask 2-1, are used to construct a structured task tree for preliminary thinking. This intelligent decomposition and structural modeling of complex driving tasks using a large language model generate a clearly structured task tree, effectively improving task understanding accuracy and execution efficiency.

[0075] For example, if the driving task being searched is "travel around a certain place," the large language model can break it down into branch task 1, which recommends locations for individual travelers, and branch task 2, which recommends locations for family travelers. Branch task 1 might include a sub-node for analyzing attractions suitable for individual travelers, such as individual itineraries and photo check-in locations. Branch task 2 might include a sub-node for analyzing family theme parks, such as consumption patterns and recommended attractions.

[0076] In this embodiment, the large language model is triggered to decompose the driving task to be searched into multiple branch tasks, as well as multiple subtasks associated with each branch task. Then, a structured task tree is constructed based on the task relationship between the subtasks in each branch task. Based on the task decomposition and the establishment of the relationship between the subtasks, support can be provided for dynamic task optimization during the search process, which can effectively cope with the multi-constraint search and real-time change requirements in driving scenarios.

[0077] In an exemplary embodiment, outputting search content feedback results based on the corrected search data of each branch task may include the following steps:

[0078] According to the relationship between the branches in the structured task tree, the target node in the structured task tree is optimized to obtain the optimized task tree; the tasks represented by the target node at least include tasks that affect the logical coherence between the subtasks; from the corrected search data of each branch task, the search data associated with each node in the optimized task tree is selected to obtain the target search data; the target search data is fused and the search content feedback results are output.

[0079] For example, in the branch relationship intelligent management stage, the large language model can intelligently judge the relationship between each task branch to optimize the target node, such as automatically processing thought conflicts (task nodes with contradictory or conflicting information), thought supplementation (perfecting incomplete task nodes), thought replacement (replacing nodes with old information with updated information) or thought independence (independent processing of unrelated task nodes) to ensure the coherence of task logic.

[0080] In this embodiment, the target node in the structured task tree is optimized according to the relationship between the branches in the structured task tree to obtain an optimized task tree; then, the search data associated with each node in the optimized task tree can be selected from the revised search data of the above-mentioned branch tasks to obtain target search data; for example, the revised search data includes the revised search data of branch task A and the revised search data of branch task B; in the case that branch task B in the optimized task tree is deleted, the revised search data of branch task A can be used as the above-mentioned target search data. Then, the target search data is fused and the search content feedback result is output. Specifically, the data integration function of the large language model can be called to fuse the above-mentioned target search data to generate a complete natural language segment, which includes the information carried by the target search data, and then the above-mentioned natural language segment is output as the search content feedback result.

[0081] The technical solution of this embodiment optimizes the target nodes in the structured task tree according to the relationship between the branches in the structured task tree to obtain an optimized task tree, and selects target search data from the corrected search data based on the optimized task tree. This can realize dynamic adjustment of the search strategy based on the relationship between the nodes of the optimized task tree, which helps to improve the accuracy and efficiency of complex task processing.

[0082] In an exemplary embodiment, after the step of constructing a structured task tree according to the task relationships between the subtasks, the following steps may be further included:

[0083] Acquire vehicle position information and driving environment information of the vehicle; determine tree structure parameters of a structured task tree based on the vehicle position information and the driving environment information; the tree structure parameters include at least one of the number of branches or the tree depth; the tree depth is used to characterize the maximum level of nodes in the structured task tree; update the original tree structure of the structured task tree according to the tree structure parameters to obtain an updated structured task tree.

[0084] In the specific implementation, for the dynamic branch optimization stage, the number and complexity of task branches can be dynamically evaluated and controlled according to the real-time changes in the driving environment (i.e., driving environment information) and vehicle position information to adjust the structured task tree; for example, when it is detected that the vehicle is approaching the destination and the driver's operating time is reduced, the number of branches can be automatically reduced to simplify the task process, ensure driving safety and the timeliness of task completion.

[0085] In this embodiment, by obtaining the vehicle position information and driving environment information of the vehicle, the tree structure parameters of the structured task tree are determined according to the vehicle position information and driving environment information, and then the original tree structure of the structured task tree is updated according to the tree structure parameters to obtain an updated structured task tree, which can dynamically control the number and depth of task branches in real time, effectively avoiding information overload and cognitive burden problems.

[0086] In an exemplary embodiment, determining the tree structure parameters of the structured task tree according to the vehicle position information and the driving environment information may include the following steps:

[0087] Based on the vehicle position information, the remaining driving time for the vehicle to reach the destination indicated by the driving task to be searched is estimated, and based on the driving environment information, the driving idle time of the vehicle occupants is determined; when the remaining driving time is less than a preset first time threshold and the driving idle time is less than a preset second time threshold, the number of branches of the structured task tree is set from the preset first branch number to the second branch number.

[0088] The second number of branches is smaller than the first number of branches, and the second number of branches is smaller than or equal to a preset number threshold. The above number threshold may refer to the maximum number of branches in the structured task tree in the current driving scenario.

[0089] As an example, the driving idle time may refer to a period of time during driving when the vehicle does not need to be operated temporarily due to situations such as traffic congestion, waiting for a traffic light, or parking.

[0090] Specifically, the in-vehicle intelligent search system can limit the number of branches to 1 to 3 for short trips and emergency situations to ensure quick decision-making. It can also allow more branches to be processed in parallel for long trips or complex tasks. At the same time, it needs to be dynamically optimized based on the real-time remaining driving time and the estimated task completion time. For example, the number of branches can be adjusted based on the remaining driving time and the idle driving time to reduce the number of branches and meet actual driving needs.

[0091] In this embodiment, the remaining driving time for the vehicle to reach the destination indicated by the driving task to be searched is estimated based on the vehicle position information, and the driving idle time of the vehicle occupants is determined based on the driving environment information. Then, when the remaining driving time is less than a preset first time threshold and the driving idle time is less than a preset second time threshold, the number of branches of the structured task tree is set from the preset first branch number to the second branch number. Based on the dynamic adjustment of the number of task branches, the task structure can be automatically streamlined when time is tight, reducing the risk of information overload. At the same time, more task branches can be retained when time is ample, ensuring the comprehensiveness and accuracy of the search results, and achieving an adaptive balance between the efficiency and safety of driving task searches.

[0092] In an exemplary embodiment, presenting search content feedback results according to a preset in-vehicle information interaction mode of a vehicle may include the following steps:

[0093] Play the search content feedback results through the vehicle's voice broadcast function; and / or display the search content feedback results through the vehicle's onboard display device.

[0094] In actual applications, the search content feedback results can be clearly displayed to vehicle occupants through intuitive and efficient voice broadcast or graphical interface output, effectively improving the convenience of information interaction and the comfort of user experience during driving.

[0095] In an exemplary embodiment, Figure 5 As shown, a flow chart of another driving task search method is provided. In this embodiment, the method includes the following steps:

[0096] In step S501, the driving task to be searched is obtained by identifying the driving task input instructions of the vehicle occupant. In step S502, the large language model is triggered to decompose the driving task to be searched into multiple branch tasks and multiple subtasks associated with each branch task. In step S503, a structured task tree is constructed based on the task relationships between the subtasks in each branch task. In step S504, the real-time search data returned by the search agent component for each branch task is obtained. By invoking the data correction function of the large language model, the real-time search data of each branch task is corrected to obtain corrected search data for each branch task. In step S505, the target node in the structured task tree is optimized based on the relationships between the branches in the structured task tree to obtain an optimized task tree. In step S506, the search data associated with each node in the optimized task tree is selected from the corrected search data of each branch task to obtain target search data. In step S507, the target search data is fused and the search content feedback result is output. In step S508, the search content feedback result is played through the voice broadcast function of the vehicle, and / or the search content feedback result is displayed through the vehicle's onboard display device.

[0097] It should be noted that the specific definitions of the above steps can be found in the specific definitions of a driving task search method above, which will not be repeated here.

[0098] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0099] Based on the same inventive concept, embodiments of the present application also provide a driving task search device for implementing the aforementioned driving task search method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more of the following driving task search device embodiments can be found in the aforementioned limitations of the driving task search method and will not be further elaborated here.

[0100] In an exemplary embodiment, Figure 6As shown, a driving task search device is provided, comprising:

[0101] A driving task acquisition module 601 is configured to obtain a driving task to be searched by identifying a driving task input instruction of a vehicle occupant;

[0102] A task decomposition module 602 is configured to parse the driving task to be searched using a large language model to obtain a plurality of branch tasks associated with the driving task to be searched, and a plurality of subtasks obtained by decomposing each branch task;

[0103] A search data correction module 603 is configured to obtain the real-time search data returned by the search proxy component for each branch task, and correct the real-time search data of each branch task by invoking the data error correction function of the large language model to obtain corrected search data for each branch task;

[0104] Feedback result output module 604, configured to output search content feedback results based on the corrected search data of each branch task;

[0105] The search content presentation module 605 is configured to present the search content feedback result in accordance with a preset vehicle information interaction method.

[0106] In one embodiment, the task decomposition module 602 is specifically used to trigger the large language model to decompose the driving task to be searched into multiple branch tasks and multiple subtasks associated with each branch task; construct a structured task tree based on the task relationship between each subtask in each branch task; wherein the structured task tree adopts a tree structure, the root node of the structured task tree is used to represent the driving task to be searched, and the multiple branches of the structured task tree are used to represent one branch task respectively; the child node in each branch is used to represent a subtask decomposed from the task represented by the parent node.

[0107] In one embodiment, the feedback result output module 604 is specifically used to optimize the target node in the structured task tree according to the relationship between each branch in the structured task tree to obtain an optimized task tree; the task represented by the target node at least includes the task that affects the logical coherence between each subtask; from the corrected search data of each branch task, the search data associated with each node in the optimized task tree is selected to obtain the target search data; the target search data is fused and the search content feedback result is output.

[0108] In one embodiment, the apparatus further comprises:

[0109] A real-time information acquisition module, configured to acquire vehicle position information and driving environment information of the vehicle;

[0110] a tree structure parameter determination module, configured to determine a tree structure parameter of the structured task tree based on the vehicle position information and the driving environment information; the tree structure parameter comprising at least one of the number of branches or the tree depth; the tree depth being used to characterize the maximum level of nodes in the structured task tree;

[0111] The structured task tree updating module is used to update the original tree structure of the structured task tree according to the tree structure parameters to obtain the updated structured task tree.

[0112] In one embodiment, the tree structure parameter determination module is specifically used to estimate the remaining driving time of the vehicle to the destination indicated by the driving task to be searched based on the vehicle position information, and determine the driving idle time of the vehicle occupants based on the driving environment information; when the remaining driving time is less than a preset first time threshold and the driving idle time is less than a preset second time threshold, the number of branches of the structured task tree is set from a preset first branch number to a second branch number; wherein the second branch number is less than the first branch number, and the second branch number is less than or equal to a preset number threshold.

[0113] In one embodiment, the search content presentation module 605 is specifically configured to play the search content feedback result through the voice broadcast function of the vehicle; and / or display the search content feedback result through the vehicle-mounted display device of the vehicle.

[0114] Each module in the aforementioned driving task search device may be implemented in whole or in part through software, hardware, or a combination thereof. Each module may be embedded in or independent of a processor within a computer device in the form of hardware, or may be stored in a computer device memory in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0115] In an exemplary embodiment, a smart vehicle is provided. The smart vehicle includes an on-board processing unit. The internal structure of the on-board processing unit can be as follows: Figure 7As shown. The on-board processing unit includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the on-board processing unit is used to provide computing and control capabilities. The memory of the on-board processing unit includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the on-board processing unit is used to store data for implementing the above-mentioned driving task search method. The input / output interface of the on-board processing unit is used to exchange information between the processor and an external device. The communication interface of the on-board processing unit is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for controlling the in-vehicle environment is implemented.

[0116] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0117] In an exemplary embodiment, an intelligent vehicle is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0118] Obtaining a driving task to be searched by identifying a driving task input instruction of a vehicle occupant;

[0119] parsing the driving task to be searched using a large language model to obtain a plurality of branch tasks associated with the driving task to be searched, and a plurality of subtasks obtained by decomposing each branch task;

[0120] Obtaining real-time search data returned by the search proxy component for each branch task, and correcting the real-time search data of each branch task by calling the data error correction function of the large language model to obtain corrected search data for each branch task;

[0121] Outputting search content feedback results based on the revised search data of each branch task;

[0122] The search content feedback results are presented according to a preset in-vehicle information interaction method.

[0123] In one embodiment, when the processor executes the computer program, the processor further implements the steps of the driving task search method in the other embodiments described above.

[0124] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0125] Obtaining a driving task to be searched by identifying a driving task input instruction of a vehicle occupant;

[0126] parsing the driving task to be searched using a large language model to obtain a plurality of branch tasks associated with the driving task to be searched, and a plurality of subtasks obtained by decomposing each branch task;

[0127] Obtaining real-time search data returned by the search proxy component for each branch task, and correcting the real-time search data of each branch task by calling the data error correction function of the large language model to obtain corrected search data for each branch task;

[0128] Outputting search content feedback results based on the revised search data of each branch task;

[0129] The search content feedback results are presented according to a preset in-vehicle information interaction method.

[0130] In one embodiment, when the computer program is executed by a processor, the steps of the driving task search method in the other embodiments described above are also implemented.

[0131] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0132] Obtaining a driving task to be searched by identifying a driving task input instruction of a vehicle occupant;

[0133] parsing the driving task to be searched using a large language model to obtain a plurality of branch tasks associated with the driving task to be searched, and a plurality of subtasks obtained by decomposing each branch task;

[0134] Obtaining real-time search data returned by the search proxy component for each branch task, and correcting the real-time search data of each branch task by calling the data error correction function of the large language model to obtain corrected search data for each branch task;

[0135] Outputting search content feedback results based on the revised search data of each branch task;

[0136] The search content feedback results are presented according to a preset in-vehicle information interaction method.

[0137] In one embodiment, when the computer program is executed by a processor, the steps of the driving task search method in the other embodiments described above are also implemented.

[0138] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0139] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. The non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magneto resistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. The volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., without being limited thereto.

[0140] The technical features of the above embodiments can be combined arbitrarily. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present application.

[0141] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A driving task search method, characterized in that: The method comprises: Obtaining a driving task to be searched by identifying a driving task input instruction of a vehicle occupant; parsing the driving task to be searched using a large language model to obtain a plurality of branch tasks associated with the driving task to be searched, and a plurality of subtasks obtained by decomposing each branch task; Obtaining real-time search data returned by the search proxy component for each branch task, and correcting the real-time search data of each branch task by calling the data error correction function of the large language model to obtain corrected search data for each branch task; Outputting search content feedback results based on the revised search data of each branch task; The search content feedback results are presented according to a preset in-vehicle information interaction method.

2. The method according to claim 1, characterized in that The large language model is used to parse the driving task to be searched, thereby obtaining a plurality of branch tasks associated with the driving task to be searched, and a plurality of subtasks obtained by decomposing each branch task, including: triggering the large language model to decompose the driving task to be searched into the plurality of branch tasks, and the plurality of subtasks associated with each branch task; Constructing a structured task tree based on the task relationships between the subtasks in each branch task; The structured task tree adopts a tree structure, wherein the root node of the structured task tree is used to represent the driving task to be searched, and the multiple branches of the structured task tree are used to represent one of the branch tasks respectively; the child nodes in each branch are used to represent a subtask obtained by decomposing the task represented by the parent node.

3. The method according to claim 2, characterized in that Outputting search content feedback results according to the corrected search data of each branch task includes: Optimizing a target node in the structured task tree according to the relationship between the branches in the structured task tree to obtain an optimized task tree; the task represented by the target node at least includes a task that affects the logical coherence between the subtasks; Selecting search data associated with each node in the optimized task tree from the corrected search data of each branch task to obtain target search data; The target search data is fused and processed, and the search content feedback result is output.

4. The method according to claim 2, characterized in that After the step of constructing a structured task tree based on the task relationships between the subtasks, the method further includes: Acquiring vehicle position information and driving environment information of the vehicle; Determining a tree structure parameter of the structured task tree based on the vehicle position information and the driving environment information; the tree structure parameter includes at least one of the number of branches or the tree depth; the tree depth is used to represent the maximum level of nodes in the structured task tree; The original tree structure of the structured task tree is updated according to the tree structure parameters to obtain an updated structured task tree.

5. The method according to claim 4, characterized in that The determining of tree structure parameters of the structured task tree according to the vehicle position information and the driving environment information includes: estimating a remaining driving time for the vehicle to reach a destination indicated by the to-be-searched driving task based on the vehicle position information, and determining a driving idle time for the vehicle occupants based on the driving environment information; When the remaining driving time is less than a preset first time threshold and the idle driving time is less than a preset second time threshold, the number of branches of the structured task tree is set from a preset first branch number to a second branch number; wherein the second branch number is less than the first branch number, and the second branch number is less than or equal to the preset number threshold.

6. The method according to any one of claims 1 to 5, characterized in that Presenting the search content feedback result according to the vehicle information interaction mode preset by the vehicle includes: Playing the search content feedback results through the voice broadcast function of the vehicle; and / or, The search content feedback result is displayed through an on-board display device of the vehicle.

7. A driving task search device, characterized in that: The device comprises: A driving task acquisition module is used to obtain the driving task to be searched by identifying the driving task input instructions of the vehicle occupants; a task decomposition module, configured to parse the driving task to be searched using a large language model to obtain a plurality of branch tasks associated with the driving task to be searched, and a plurality of subtasks obtained by decomposing each branch task; A search data correction module is configured to obtain the real-time search data returned by the search proxy component for each branch task, and correct the real-time search data of each branch task by calling the data error correction function of the large language model to obtain corrected search data for each branch task; A feedback result output module, configured to output search content feedback results based on the corrected search data of each branch task; The search content presentation module is used to present the search content feedback results according to a preset vehicle information interaction method.

8. An intelligent vehicle comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. 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 6 are implemented.

10. A computer program product comprising a computer program, 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 6 are implemented.

Images