Vehicle-mounted task resource allocation method and device, electronic equipment, medium and product

Abstract

The invention discloses a vehicle-mounted task resource allocation method and device, electronic equipment, a medium and a product. The method comprises the steps of collecting multi-source data of a vehicle in real time; generating a dynamic task set according to the multi-source data; calculating an emergency degree score and a target coupling degree score of each task in the dynamic task set; determining a task grade of each task according to the emergency degree score and the target coupling degree score; and carrying out resource allocation on each task according to the task classification. According to the method, the problems of sudden task response delay and low resource utilization efficiency in a vehicle-mounted system can be solved.

Description

Technical Field

[0001] This application relates to the field of automotive electronic control technology, specifically to an onboard task resource allocation method, device, electronic device, readable storage medium, and computer program product. Background Technology

[0002] In intelligent vehicle systems, sudden tasks such as collision warning and emergency braking have extremely high real-time requirements. Traditional task scheduling methods, such as rate monotonic algorithms based on fixed priorities, heavily rely on the periodicity of tasks and lack the ability to dynamically respond to sudden events. This leads to delayed responses to highly urgent tasks due to queuing delays, seriously affecting driving safety. At the same time, existing methods generally ignore the resource correlation between tasks. The disorderly competition of multiple tasks for the same computing resources can easily lead to duplicate calculations and resource conflicts, significantly reducing the system's resource utilization efficiency and failing to meet the scheduling requirements of high real-time performance and high resource utilization in complex driving environments. Summary of the Invention

[0003] In view of the above problems, this application provides a method, apparatus, electronic device, readable storage medium and computer program product for allocating vehicle-mounted task resources, which can solve the problems of delayed response to sudden tasks and low resource utilization efficiency in vehicle-mounted systems.

[0004] Firstly, this application provides a method for allocating vehicle-mounted mission resources, including: Real-time collection of multi-source vehicle data; A dynamic task set is generated based on the multi-source data; Calculate the urgency score and target coupling score for each task in the dynamic task set; Based on the urgency score and the target coupling score, the task level of each task is determined; Resource allocation is performed for each task according to the task classification.

[0005] In the above technical solution, this method can collect multi-source vehicle data in real time and generate a dynamic task set, thereby achieving timely and comprehensive capture of vehicle operating status and potential tasks, providing accurate basis for subsequent scheduling. At the same time, by calculating the urgency score of each task, it can prioritize the identification of high-urgency tasks such as collision warnings, avoiding response delays due to queuing delays and ensuring driving safety. In addition, by calculating the target coupling score, it can effectively consider the resource correlation between tasks, reduce disorderly competition for the same computing resources by multiple tasks, and avoid duplicate calculations and resource conflicts. By combining the urgency score and the target coupling score to determine the task level and allocate resources accordingly, it can ultimately optimize resource allocation efficiency while dynamically responding to sudden tasks to improve the response speed of high-urgency tasks, meeting the dual requirements of high real-time performance and high resource utilization of the vehicle system in complex driving environments.

[0006] In some implementations, the real-time acquisition of multi-source vehicle data includes: The vehicle uses onboard sensors to collect sensor data about the surrounding environment in real time. Obtain operating status parameters from the vehicle's internal systems; Receive user interaction requests via voice, touch, or other means; By summarizing the sensor data, the operating status parameters, and the user interaction requests, multi-source data is obtained.

[0007] In the above technical solution, the method can collect data from multiple dimensions through vehicle sensors, vehicle internal systems and user interaction channels, comprehensively covering the vehicle's surrounding environment, its own operating status and user needs, ensuring that the acquired multi-source data is complete and real-time, providing comprehensive and timely basic data support for the accurate generation of subsequent dynamic task sets, and avoiding deviations in task generation due to missing or delayed data.

[0008] In some implementations, generating a dynamic task set based on the multi-source data includes: The multi-source data is standardized to obtain standardized data; A dynamic task set is generated based on preset rules and the standardized data.

[0009] In the above technical solution, the method can first standardize the multi-source data to eliminate the differences in format and dimension of data from different sources and avoid the interference of data inconsistency on task generation; then, it combines preset rules and standardized data to generate a dynamic task set to ensure that the generated tasks can accurately match the actual operating needs and potential scenarios of the vehicle.

[0010] In some implementations, calculating the urgency score and target coupling score for each task in the dynamic task set includes: Calculate the urgency score for each task in the dynamic task set; Calculate the resource requirement matrix for each of the tasks; Multiple task pairs are determined based on the dynamic task set; Calculate the intersection of resource requirements and task correlation factor for each task pair; Based on the task relevance factor and the intersection of resource requirements between tasks, calculate the original coupling score for each task pair; The original coupling score is normalized to obtain the normalized coupling score for each task pair. The target coupling score for each task is determined based on the normalized coupling score for each task pair.

[0011] In the above technical solution, the method can quickly identify high-urgency tasks by calculating the urgency of tasks and the resource coupling degree between tasks respectively. At the same time, it can effectively quantify the resource competition and dependency relationship between tasks by analyzing the intersection and correlation of resource requirements between task pairs, thereby providing a comprehensive quantitative basis for subsequent task classification and resource allocation, and improving the system's response speed and resource utilization efficiency for urgent tasks.

[0012] In some implementations, calculating the urgency score for each task in the dynamic task set includes: Extract the task features of each task in the dynamic task set; wherein, the task features include at least security impact features, deadline features, and user level features; An urgency score for each task is calculated based on the learnable coefficients obtained from pre-training and the task characteristics.

[0013] In the above technical solution, the method can extract multi-dimensional features such as the security impact of the task, the deadline and the user level, and perform weighted calculation by combining the pre-trained learnable coefficients, so as to achieve scientific quantification and dynamic adjustment of the urgency of the task.

[0014] In some implementations, calculating the resource requirement matrix for each of the tasks includes: Identify the task type for each of the tasks; Calculate the task cost for each task based on the task type; wherein the task cost includes computational cost and communication cost; Generate a resource requirement matrix for each task based on the task overhead.

[0015] In the above technical solution, the method can first identify the task type to clarify the direction of resource demand, then calculate the computational and communication overhead of the task in a targeted manner, and finally generate a resource demand matrix, so as to achieve accurate quantification and clear representation of the resources required for each task.

[0016] In some implementations, calculating the intersection of inter-task resource requirements and the task correlation factor for each task pair includes: The task-to-task resource requirement intersection for each task pair is determined based on the resource requirement matrix. Obtain the task feature matrix for each task; wherein the task feature matrix includes at least the task type code, normalized value of computation overhead, normalized value of communication overhead, historical co-occurrence count, resource overlap rate, and security correlation degree; The task correlation factor for each task pair is generated using a preset graph neural network and the task feature matrix.

[0017] In the above technical solution, the method can accurately calculate the resource intersection between task pairs through the resource demand matrix, and use the task feature matrix combined with graph neural network to deeply mine the semantic and statistical associations between tasks, thereby quantifying the coupling relationship between tasks from both the resource dimension and the feature dimension.

[0018] In some implementations, determining the task level for each task based on the urgency score and the target coupling score includes: Obtain the pre-set urgency threshold and coupling threshold; The task level of each task is determined based on the urgency threshold, the coupling threshold, the urgency score, and the target coupling score; The task is classified into one of the following levels: real-time critical level, elastic optimization level, and background batch processing level.

[0019] In the above technical solution, the method can clearly classify tasks into real-time critical level, elastic optimization level or background batch processing level based on preset urgency and coupling thresholds, combined with the urgency score and target coupling score of each task, so as to achieve fine-grained hierarchical management of tasks with different priorities.

[0020] In some implementations, allocating resources to each task according to the task hierarchy includes: The available computing resources of the vehicle-mounted computing platform are statistically analyzed. Based on the task classification, the first real-time critical task and the second non-real-time critical task in the dynamic task set are statistically analyzed. Allocate a preset proportion of the available computing resources to the first task, and pre-allocate a preset size of memory space to the first task; Determine the current remaining available resources; The remaining available resources are allocated to the second task according to a preset allocation ratio.

[0021] In the above technical solution, the method can ensure that real-time critical tasks have sufficient computing and memory resources to meet high real-time requirements by prioritizing the resource needs of real-time critical tasks; at the same time, the remaining resources are allocated to non-real-time critical tasks according to a preset ratio, so as to achieve efficient use of resources while ensuring system security and avoiding resource waste and task blocking.

[0022] Secondly, this application provides an on-board mission resource allocation device, comprising: The data acquisition unit is used to collect multi-source data from vehicles in real time. The generation unit is used to generate a dynamic task set based on the multi-source data; A calculation unit is used to calculate the urgency score and target coupling score of each task in the dynamic task set; A determining unit is configured to determine the task level of each task based on the urgency score and the target coupling score; A resource allocation unit is used to allocate resources to each task according to the task hierarchy.

[0023] In the above technical solution, the device can collect multi-source vehicle data in real time and generate a dynamic task set, thereby achieving timely and comprehensive capture of vehicle operating status and potential tasks, providing accurate basis for subsequent scheduling. At the same time, by calculating the urgency score of each task, it can prioritize the identification of high-urgency tasks such as collision warnings, avoiding response delays due to queuing delays and ensuring driving safety. In addition, by calculating the target coupling score, it can effectively consider the resource correlation between tasks, reduce disorderly competition for the same computing resources by multiple tasks, and avoid duplicate calculations and resource conflicts. By combining the urgency score and the target coupling score to determine the task level and allocate resources accordingly, it can ultimately optimize resource allocation efficiency while dynamically responding to sudden tasks to improve the response speed of high-urgency tasks, meeting the dual requirements of high real-time performance and high resource utilization of the vehicle system in complex driving environments.

[0024] Thirdly, this application provides an electronic device, the electronic device including a memory and a processor, the memory for storing a computer program, the processor running the computer program to cause the electronic device to perform the vehicle task resource allocation method described in any one of the first aspects.

[0025] Fourthly, this application provides a readable storage medium storing a computer program, which, when executed by a processor, performs the vehicle task resource allocation method described in any one of the first aspects.

[0026] Fifthly, this application provides a computer program product, which includes a computer program that, when executed by a processor, performs the vehicle task resource allocation method described in any one of the first aspects.

[0027] The beneficial effects of this application are: it can capture sudden tasks in real time through dynamic task sets, and combine multi-dimensional urgency assessment and dual threshold classification to achieve immediate identification and priority scheduling of sudden tasks.

[0028] At the same time, it can identify sharing opportunities through coupling degree assessment and reduce redundancy through differentiated resource allocation, thus solving the problem of resource mismatch.

[0029] Finally, it can also collect environmental data (traffic flow, road conditions, etc.) every 200 milliseconds and adjust the learnable coefficients (α, β, γ) using gradient descent to dynamically adapt the urgency assessment to the scenario and solve the problem of insufficient environmental adaptability. Attached Figure Description

[0030] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 This is a flowchart illustrating the vehicle-mounted task resource allocation method in some embodiments of this application; Figure 2 This is a flowchart illustrating the vehicle-mounted task resource allocation method in some embodiments of this application; Figure 3 This is a schematic diagram of the structure of the vehicle-mounted task resource allocation device in some embodiments of this application; Figure 4 This is a schematic diagram of the structure of an electronic device in some embodiments of this application. Detailed Implementation

[0032] The embodiments of the technical solution of this application will now be described in detail with reference to the accompanying drawings. These embodiments are only used to more clearly illustrate the technical solution of this application and are therefore merely examples, and should not be used to limit the scope of protection of this application.

[0033] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0034] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more (including two), similarly, "multiple sets" refers to two or more sets (including two sets), and "multiple pieces" refers to two or more pieces (including two pieces) unless otherwise explicitly defined.

[0035] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0036] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0037] Existing technologies rely on fixed priority lists (such as RM / DRM algorithms with a periodic focus), resulting in delayed responses to unexpected tasks such as collision warnings and emergency stop decisions. Furthermore, existing technologies do not assess task resource dependencies, leading to low utilization due to redundant calculations or resource contention. Additionally, the fixed scheduling parameters (such as priority weights) of existing technologies result in poor adaptability in scenarios such as high-speed congestion and poor network conditions.

[0038] To address the aforementioned technical problems, this application provides a method for allocating vehicle-mounted task resources. This method can dynamically capture sudden tasks in real time from a task set and, by combining multi-dimensional urgency assessment (prioritizing safety impact features) and dual-threshold grading (high urgency is directly classified as critical), achieve immediate identification and priority scheduling of sudden tasks. Simultaneously, this method can also identify sharing opportunities (such as shared feature maps for target detection and trajectory prediction) through coupling degree assessment (based on the intersection of resource requirements and graph neural network correlation factors), and reduce redundancy through differentiated resource allocation, thus solving the resource mismatch problem. Finally, this method can also collect environmental data (traffic flow, road conditions, etc.) every 200 milliseconds and adjust the learnable coefficients (α, β, γ) using gradient descent, enabling the urgency assessment to dynamically adapt to the scenario and solving the problem of insufficient environmental adaptability.

[0039] like Figure 1 As shown, some embodiments of this application provide a method for allocating vehicle-mounted task resources, which includes: S101. Real-time acquisition of multi-source vehicle data; S102. Generate a dynamic task set based on multi-source data; S103. Calculate the urgency score and target coupling score for each task in the dynamic task set; S104. Determine the task level for each task based on the urgency score and the target coupling score; S105. Allocate resources for each task according to the task level.

[0040] In some embodiments, multi-source data refers to a collection of vehicle surrounding environment perception data collected in real time by onboard sensors, operating status parameters obtained from the vehicle's internal systems, and interactive requests issued by the user through voice, touch, or other means.

[0041] In some embodiments, a dynamic task set refers to a set formed by standardizing multi-source data, generating corresponding tasks according to preset triggering rules (such as exceeding environmental risk parameters, abnormal vehicle status, etc.), and integrating them. This set is dynamically updated as the vehicle's operating status and the surrounding environment change.

[0042] In some embodiments, the urgency score refers to a quantitative indicator calculated by extracting the safety impact features (the degree of impact on driving safety), deadline features (the time node when the task needs to be completed), and user level features (the importance of the user), and combining them with pre-trained learnable coefficients for weighted calculation. It is used to accurately measure the priority level of a task, especially highlighting the urgency of safety-related tasks.

[0043] In some embodiments, the target coupling score refers to a quantitative index obtained by first determining the computational and communication overhead based on the task type to generate a resource requirement matrix, then determining task pairs and calculating the intersection of their resource requirements and the task correlation factor generated by the graph neural network, and then normalizing the result. This index is used to measure the degree of resource correlation between tasks and reflect the resource competition and dependency between tasks.

[0044] In some embodiments, task classification refers to the process of setting high / low urgency thresholds and high / low coupling thresholds, comparing the urgency score and target coupling score of a task with the corresponding thresholds, and classifying tasks into real-time critical level (high urgency or high coupling), elastic optimization level (medium urgency and medium coupling), and background batch processing level (low urgency and low coupling), thereby achieving fine-grained classification of tasks with different priorities.

[0045] In some embodiments, resource allocation refers to the process of first calculating the available resources of the vehicle computing platform, then prioritizing the allocation of sufficient computing resources and pre-allocated memory to real-time critical tasks, and distributing the remaining resources proportionally to elastically optimized tasks and background batch processing tasks.

[0046] In the above embodiments, this method can capture the vehicle's operating status and potential tasks in a timely and comprehensive manner by collecting multi-source vehicle data in real time and generating a dynamic task set, providing accurate basis for subsequent scheduling. At the same time, by calculating the urgency score of each task, it can prioritize the identification of high-urgency tasks such as collision warnings, avoiding response delays due to queuing delays and ensuring driving safety. In addition, by calculating the target coupling score, it can effectively consider the resource correlation between tasks, reduce the disorderly competition of multiple tasks for the same computing resources, and avoid duplicate calculations and resource conflicts. By combining the urgency score and the target coupling score to determine the task level and allocate resources accordingly, it can ultimately optimize resource allocation efficiency while dynamically responding to sudden tasks to improve the response speed of high-urgency tasks, meeting the dual requirements of high real-time performance and high resource utilization of the vehicle system in complex driving environments.

[0047] In some embodiments, multi-source data of the vehicle is collected in real time, including: The vehicle uses onboard sensors to collect sensor data about the surrounding environment in real time. Obtain operating status parameters from the vehicle's internal systems; Receive user interaction requests via voice, touch, or other means; By aggregating sensor data, operational status parameters, and user interaction requests, multi-source data is obtained.

[0048] In some embodiments, the method collects real-time perception data of the vehicle's surrounding environment through onboard sensors (such as lidar, cameras, etc.), obtains operating status parameters (such as vehicle speed, tire pressure, engine status, etc.) from the vehicle's internal systems, and simultaneously receives interactive requests from users via voice, touch, etc.

[0049] In some embodiments, the method acquires surrounding environmental data in real time through vehicle-mounted sensors (such as lidar and cameras) (the lidar outputs 10 frames of 3D point cloud per second with a resolution of 1024×64; the camera outputs 25 frames of 1920×1080 RGB images per second), obtains vehicle operating status parameters from the CAN bus (sampling frequency of 100Hz, including vehicle speed 0-120km / h, yaw rate -5° / s to 5° / s, tire pressure 200-300kPa, etc.), and receives user interaction requests through the voice module.

[0050] In the above embodiments, the method can collect data from multiple dimensions through vehicle sensors, vehicle internal systems and user interaction channels, comprehensively covering the vehicle's surrounding environment, its own operating status and user needs, ensuring that the acquired multi-source data is complete and real-time, providing comprehensive and timely basic data support for the accurate generation of subsequent dynamic task sets, and avoiding deviations in task generation due to missing or delayed data.

[0051] In some embodiments, generating a dynamic task set based on multi-source data includes: Standardize multi-source data to obtain standardized data; A dynamic task set is generated based on preset rules and standardized data.

[0052] In some embodiments, the method standardizes these multi-source data. For example, it converts sensing data in different formats into a unified data format to ensure that the data can be recognized and processed by subsequent modules. Then, according to preset triggering rules (such as when the collision risk parameter exceeds a set threshold), corresponding tasks are generated and integrated to form a dynamic task set.

[0053] In some embodiments, the method can standardize the collected data (point cloud is converted into a (x,y,z,intensity) quadruple matrix, image is compressed into JPEG format, CAN parameters are normalized to 0-1, and user requests are converted into structured instructions). Then, a dynamic task set is generated according to preset rules (e.g., when the number of point cloud clusters of obstacles detected by the LiDAR within 50 meters ahead is greater than 50, an "obstacle recognition" task is generated; when the tire pressure is less than 220 kPa, a "tire pressure warning" task is generated).

[0054] In the above embodiments, the method can first standardize multi-source data to eliminate differences in format and dimension of data from different sources and avoid interference from data inconsistency in task generation; then, it combines preset rules and standardized data to generate a dynamic task set to ensure that the generated tasks can accurately match the actual operating needs and potential scenarios of the vehicle.

[0055] In some embodiments, calculating the urgency score and target coupling score for each task in the dynamic task set includes: Calculate the urgency score for each task in the dynamic task set; Calculate the resource requirement matrix for each task; Multiple task pairs are determined based on a dynamic task set; Calculate the intersection of resource requirements and task correlation factor for each task pair; Calculate the original coupling score for each task pair based on the task relevance factor and the intersection of resource requirements between tasks; The original coupling score is normalized to obtain the normalized coupling score for each task pair. The target coupling score for each task is determined based on the normalized coupling score of each task pair.

[0056] In some embodiments, the method can calculate the coupling score based on the intersection of resource requirements and correlation factors between tasks through normalization. The key formula can be simplified to: Coupling Score = Normalized (Intersection of Resource Requirements × Correlation Factor).

[0057] In some embodiments, the original coupling value of task pair (i,j) is: V ij =|R i ∩R j | / (||R i ||2·||R j ||2)·I taskrel ; Among them, |R i ∩R j |The intersection of resource demands; ||R i ||2 is R i The L2 norm.

[0058] In some embodiments, the method may use a Min-Max normalization method to calculate the coupling score, as follows: D couple =(VV min ) / (V max -V min ); Among them, V min V max For all V ij Find the minimum / maximum value, and output D. couple ∈[0,1].

[0059] In the above embodiments, the method can quickly identify high-urgency tasks by calculating the urgency of tasks and the resource coupling degree between tasks respectively. At the same time, it can effectively quantify the resource competition and dependency relationship between tasks by analyzing the intersection and correlation of resource requirements between task pairs, thereby providing a comprehensive quantitative basis for subsequent task classification and resource allocation, and improving the system's response speed and resource utilization efficiency for urgent tasks.

[0060] In some embodiments, calculating the urgency score for each task in the dynamic task set includes: Extract the task features of each task in the dynamic task set; the task features shall include at least the security impact feature, the deadline feature, and the user level feature. The urgency score for each task is calculated based on the learnable coefficients obtained from pre-training and the task characteristics.

[0061] In some embodiments, the method may first extract safety impact features (such as the degree of impact of the task on vehicle driving safety), deadline features (the time node when the task must be completed), and user level features (based on user settings or system default user importance) from each task in the dynamic task set.

[0062] Then, the three features are weighted using learnable coefficients to obtain the urgency score. Urgency score = α × security impact feature + β × deadline feature + γ × user level feature (where α, β, and γ are learnable coefficients).

[0063] In some embodiments, the security impact features S of the extraction task are extracted. safety This feature can be calculated using a security threat model, as shown below: S safety =0.7×I collision +0.3×I traffic ; Among them, I collision The collision probability is estimated as 0-1; I traffic The traffic violation risk index is 0-1.

[0064] In some embodiments, the deadline feature T deadline (Preset according to task type, such as 100ms for "emergency obstacle avoidance"); The initial values ​​of the learnable coefficients α, β, and γ are 0.5, 0.3, and 0.2, respectively (α is the weight of security impact, β is the weight of deadline, and γ is the weight of user level).

[0065] In some embodiments, the training data comes from 50,000 vehicle task logs (including features, execution latency, and resource consumption) from the past 3 months. The coefficients are optimized using gradient descent, and the loss function is: L = 0.7 × (average latency of critical tasks / 100ms) + 0.3 × (1 - η) resource ); Where, η resource The resource utilization rate is 0-1; The learning rate η = 0.01.

[0066] In some embodiments, after 5 iterations, if |L t+1 -L t | <0.01 Stop; The formula for calculating the urgency score is: E=α·S safety +β·T -1 deadline +γ·λ user Among them, T-1 deadline Counting back to the deadline amplifies time sensitivity; The final output is E∈[0,1].

[0067] In the above embodiments, the method can extract multi-dimensional features such as the security impact of the task, the deadline and the user level, and perform weighted calculations by combining them with pre-trained learnable coefficients, so as to achieve scientific quantification and dynamic adjustment of the urgency of the task.

[0068] In some embodiments, calculating the resource requirement matrix for each task includes: Identify the task type for each task; Calculate the task cost for each task based on the task type; the task cost includes computational cost and communication cost. Generate a resource requirement matrix for each task based on the task overhead.

[0069] In some embodiments, the method may first determine the type of task, such as object detection task, path planning task, user entertainment task, etc.

[0070] In some embodiments, the computational and communication overheads of different types of tasks are calculated in different ways.

[0071] For example, the computational cost of object detection is related to image resolution and model complexity, while the communication cost is related to the amount of data in the detection results; the computational cost of path planning is related to map complexity and planning range, while the communication cost is relatively small.

[0072] In some embodiments, the method can calculate the computational overhead and communication overhead according to the computational rules corresponding to the task type, and then integrate the two to form a resource requirement matrix.

[0073] In some embodiments, the method can identify task types (such as object detection, path planning, etc.) and calculate costs by type.

[0074] In some embodiments, the computational cost C of the target detection task comp,det =0.02 × image resolution (megapixels) + 0.5 × model depth (e.g., a 1920 × 1080 image corresponds to 2 megapixels, and the ResNet-18 model has a depth of 18), communication overhead C comm,det =0.1 × number of targets detected (each target's data size is approximately 10KB); Path planning task computational cost C comp,plan =0.3 × map complexity (0-1) + 0.2 × planning distance (km), communication overhead C comm,plan =0.05 × map update data volume (MB); generate resource requirement matrix R=[C comp Ccomm ](C comp Corresponding units: CPU cores per second (ms); C comm The corresponding unit is Mbps.

[0075] In the above embodiments, the method can first identify the task type to clarify the direction of resource demand, then calculate the computational and communication overhead of the task in a targeted manner, and finally generate a resource demand matrix, thereby achieving accurate quantification and clear representation of the resources required for each task.

[0076] In some embodiments, calculating the intersection of inter-task resource requirements and the task correlation factor for each task pair includes: Determine the intersection of resource requirements between tasks for each task pair based on the resource requirement matrix; Obtain the task feature matrix for each task; wherein, the task feature matrix includes at least the task type code, the normalized value of computation overhead, the normalized value of communication overhead, the number of historical co-occurrences, the resource overlap rate, and the security correlation degree; The task correlation factor for each task pair is generated by using a pre-defined graph neural network and task feature matrix.

[0077] In some embodiments, the method can determine the resource requirement matrix for each task and find the intersection of resource requirements between tasks (i.e., the resource portions commonly needed by different tasks). Simultaneously, a task correlation factor is generated using a graph neural network, which reflects the degree of logical or functional correlation between tasks (e.g., the object detection task and trajectory prediction task have a high correlation).

[0078] In some embodiments, the input to the graph neural network is a 6-dimensional task feature matrix (task type encoding 0-4, C...). comp Normalized value, C comm Normalized value, historical co-occurrence count, resource overlap rate, and security correlation degree (0-1) are processed through a 2-layer GCN, and the output layer is activated by Sigmoid to obtain the correlation factor I. taskrel (0-1, the higher the value, the stronger the association).

[0079] In the above embodiments, the method can accurately calculate the resource intersection between task pairs through the resource demand matrix, and use the task feature matrix combined with graph neural network to deeply mine the semantic and statistical associations between tasks, thereby quantifying the coupling relationship between tasks from both the resource dimension and the feature dimension.

[0080] In some embodiments, the task classification for each task is determined based on the urgency score and the target coupling score, including: Obtain the pre-set urgency threshold and coupling threshold; The task classification for each task is determined based on the urgency threshold, coupling threshold, urgency score, and target coupling score. The task is classified into one of the following levels: real-time critical level, elastic optimization level, or background batch processing level.

[0081] In some embodiments, the method may pre-set high and low thresholds for urgency, as well as high and low thresholds for coupling.

[0082] In some embodiments, for each task, its urgency score and coupling score are compared with corresponding thresholds. If the urgency score is greater than the high urgency threshold, or the coupling score is greater than the high coupling threshold, the task requires priority processing and is classified as real-time critical. If the urgency score is between the low and high urgency thresholds, and the coupling score is between the low and high coupling thresholds, it is classified as elastic optimization, and can be optimized when resources allow. Otherwise, it is classified as background batch processing, and can be processed when the system is idle. This dual-threshold classification method comprehensively considers the urgency and resource relevance of tasks, making task classification more reasonable.

[0083] In some embodiments, the method may set an urgency threshold θ high =0.85, θ low =0.3, coupling threshold δ high =0.9、δ low =0.4; Read tasks E and D couple If E > θ high Or D couple >δ high Classified as real-time critical level; if θ low <E≤θ high And δ low <D couple ≤δ high Classified as elastic optimization level; if E≤θ low And D couple ≤δ low It is classified as a background batch processing level, and the hierarchical results are written to the task attributes.

[0084] In the above embodiments, the method can clearly classify tasks into real-time critical level, elastic optimization level or background batch processing level based on preset urgency and coupling thresholds, combined with the urgency score and target coupling score of each task, so as to achieve fine-grained hierarchical management of tasks with different priorities.

[0085] In some embodiments, resource allocation is performed for each task according to task hierarchy, including: The available computing resources of the vehicle-mounted computing platform are statistically analyzed. Based on the task classification, the first real-time critical task and the second non-real-time critical task are statistically analyzed in the dynamic task set. Allocate a preset proportion of available computing resources to the first task and pre-allocate a preset amount of memory space to the first task; Determine the current remaining available resources; The remaining available resources are allocated to the second task according to the preset allocation ratio.

[0086] In some embodiments, for tasks classified as real-time critical, no less than 70% of the available computing resources are allocated, and 200MB of memory is pre-allocated to ensure that there are sufficient resources to support them and to avoid task delays or failures due to insufficient resources.

[0087] In some embodiments, for elastic optimization level and background batch processing level tasks, the remaining computing resources are allocated in a certain proportion, which can be determined based on factors such as the urgency of the task and resource requirements.

[0088] In some embodiments, the method can statistically analyze the available resources of an in-vehicle computing platform (such as NVIDIA Orin), including the number of remaining CPU cores, the proportion of GPU computing power, the amount of free memory (in MB), and the available bandwidth (in Mbps), providing a data basis for subsequent differentiated allocation.

[0089] In some embodiments, priority allocation is performed for real-time critical tasks to ensure the resource requirements of core tasks are met: Allocate no less than 70% of the available CPU resources; Allocate no less than 70% of the GPU computing power; Pre-allocate 200MB of memory (this capacity is set based on historical data; the average memory requirement for critical tasks is 150-180MB, and reserving redundancy can avoid task interruption due to insufficient resources). Guarantee at least 70% available bandwidth.

[0090] In some embodiments, after the real-time critical tasks are allocated, the remaining resources are allocated to elastic optimization tasks at a ratio of 20%-30%. The allocation ratio needs to be combined with the task's urgency score (E) and target coupling score (Dc). ouple The weighted calculation is performed, with both weights set to 0.5, to ensure that the allocation priority matches the urgency of the task and the resource relevance.

[0091] In some embodiments, background batch processing tasks are allocated only to all remaining resources, and the resource ratio does not exceed 10% of the total available resources, so as to avoid occupying core resources and affecting the operation of critical tasks.

[0092] In some embodiments, the entire resource allocation process is executed through the resource management interface of the vehicle operating system (OS), while the instruction latency must be controlled within 5ms to ensure the real-time performance of resource scheduling and avoid affecting task execution efficiency due to scheduling delay.

[0093] In the above embodiments, the method can ensure that real-time critical tasks have sufficient computing and memory resources to meet high real-time requirements by prioritizing the resource needs of real-time critical tasks; at the same time, the remaining resources are allocated to non-real-time critical tasks according to a preset ratio, so as to achieve efficient use of resources while ensuring system security and avoiding resource waste and task blocking.

[0094] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below. In some embodiments, such as Figure 2 As shown, the onboard mission resource allocation method includes: S201: Real-time acquisition of onboard sensor data, vehicle operating status parameters, and user interaction requests, followed by standardized processing to generate a dynamic task set.

[0095] S202. Extract the security impact characteristics, deadline characteristics, and user level characteristics of the task, and calculate the urgency score by weighting with learnable coefficients.

[0096] S203. Calculate the computational and communication overhead based on the task type and generate a resource requirement matrix.

[0097] S204. Based on the intersection of resource requirements between tasks and the correlation factor generated by the graph neural network, the coupling score is calculated using normalization.

[0098] S205. If the urgency score is greater than the high urgency threshold or the coupling score is greater than the high coupling value, it is classified as real-time critical level; if the urgency score is between the low and high urgency thresholds and the coupling score is between the low and high coupling values, it is classified as elastic optimization level; otherwise, it is classified as background batch processing level.

[0099] S206. Allocate no less than 70% of available computing resources to real-time critical tasks and pre-allocate 200MB of memory. The remaining resources are allocated to other tasks proportionally.

[0100] S207. Collect environmental data every 200 milliseconds, adjust the learnable coefficients using gradient descent, and re-execute steps S202 to S206.

[0101] In some embodiments, the method can collect environmental data about the vehicle, such as traffic flow, road conditions, and network status, every 200 milliseconds. Then, based on changes in the environmental data, the learnable coefficients α, β, and γ in S202 are adjusted using gradient descent to adapt the urgency score calculation to the new environment. Afterward, steps S202 to S206 are re-executed, i.e., the urgency score, resource requirement matrix, and coupling score are recalculated, and task classification and resource allocation are performed.

[0102] In these embodiments, this periodic dynamic adjustment solves the problem that traditional methods have fixed parameters and cannot adapt to environmental changes, ensuring that task processing remains efficient and reasonable in different scenarios.

[0103] In some embodiments, environmental data is collected every 200ms: V2X obtains the number of vehicles within a 500-meter radius and calculates the vehicle density as density = number of vehicles / 0.5km (normalized to 0-1). GPS and vehicle speed calculation efficiency = current vehicle speed / road speed limit (speed limit comes from high-precision map, normalized to 0-1); The network module obtains the actual speed and calculates the communication quality. quality = actual rate / 100Mbps (normalized 0-1); Environment vector E env =[density,efficiency,quality], if efficiency<0.3, α+0.1, γ-0.1; If quality < 0.4, β + 0.1; after adjustment, α + β + γ = 1, triggering S202-S206 to be re-executed.

[0104] In some embodiments, the execution process of the vehicle-mounted task resource allocation method is as follows: 1) Dynamic task set generation The system detects a vehicle braking suddenly 30 meters ahead using an onboard LiDAR (1024×64 resolution) (80 point cloud clusters), generating "Collision Warning (T1)" and "Track Correction (T2)" tasks. The CAN bus detects that the vehicle speed has dropped to 20km / h, generating "Follow Distance Adjustment (T3)" task. When the user triggers the voice command "Query Congestion Cause (T4)," the system simultaneously triggers regular tasks such as "Navigation Route Update (T5)" and "Sensor Data Log Upload (T6)," ultimately forming a dynamic set containing 10 tasks.

[0105] 2) Calculation of urgency score Based on learnable coefficients (α=0.6, β=0.3, γ=0.1, with automatic increase in safety weights in congested scenarios), task features are extracted and urgency is calculated (the score is normalized and takes a value between 0 and 1): T1 (Collision Warning): Safety impact characteristic 0.9 (collision probability 0.8), deadline 100ms, user level default 5, urgency E=0.6×0.9+0.3×(1 / 100)+0.1×5≈0.54+0.003+0.5=1.043 (normalized 0.92). T4 (User Query): Security impact characteristic 0.1, deadline 3000ms, user level 8, urgency E=0.6×0.1+0.3×(1 / 3000)+0.1×8≈0.06+0.0001+0.8=0.8601 (normalized to 0.41).

[0106] 3) Resource Demand Matrix Generation Calculate overhead by task type and generate a matrix (overhead unit: CPU cores) (ms, communication overhead unit: Mbps) T1 (Collision warning, target detection): Computational overhead = 0.02 × 200 (megapixels) + 0.5 × 18 (model depth) = 13, Communication overhead = 0.1 × 5 (number of detected targets) = 0.5, Matrix R = [13, 0.5]; T5 (Navigation path update, path planning class): Computational overhead = 0.3 × 0.8 (map complexity) + 0.2 × 5 (planning distance) = 1.24, Communication overhead = 0.05 × 2 (map data volume) = 0.1, Matrix R = [1.24, 0.1].

[0107] 4) Coupling score calculation Combining the intersection of resource demands with the correlation factors generated by the graph neural network, a coupling score (value 0-1) is obtained after normalization: T1 and T2 (trajectory correction): The intersection of resource requirements is "LiDAR data cache", the correlation factor is 0.85 (both belong to the security control category), the original coupling value = 0.6 (resource overlap rate) × 0.85 ≈ 0.51, and the normalized Dcouple = 0.92; T4 and T6 (log upload): No overlap in resource requirements, correlation factor 0.1, normalized Dcouple=0.15.

[0108] 5) Task classification results Based on thresholds (high urgency threshold 0.85, low threshold 0.3; high coupling threshold 0.9, low threshold 0.4), the following classifications are applied: Real-time criticality levels (E > 0.85 or Dcouple > 0.9): T1 (E = 0.92), T2 (Dcouple = 0.92), T3 (E = 0.88); Elastic optimization level (0.3 < E ≤ 0.85 and 0.4 < Dcouple ≤ 0.9): T5 (E = 0.65, Dcouple = 0.6), T7 (V2X data interaction); Background batch processing level: T4, T6, and the remaining 6 tasks.

[0109] 6) Resource allocation execution Real-time available resources (6-core CPU, 1000MB memory, 50Mbps bandwidth) based on the in-vehicle computing platform (NVIDIA Orin, 10-core CPU, 2000MB memory) are allocated in a tiered and differentiated manner: Real-time critical level: Allocate 5 CPU cores (83% of available CPU), 200MB pre-allocated memory, and 35Mbps bandwidth (70% of available bandwidth). Elastic Optimized Level: Allocates 1 CPU core (17% of available CPU), 100MB of memory, and 10Mbps of bandwidth; Background batch processing level: Allocate remaining resources (0 CPU cores, 700MB memory, 5Mbps bandwidth).

[0110] 7) Dynamic adjustment and process re-execution Environmental data is collected and optimized periodically every 200ms. Environmental data update: Traffic efficiency remains at 0.2 (congestion has not eased), and communication quality has dropped to 0.3 (network has deteriorated). Learnable coefficient adjustment: β (deadline weight) increased from 0.3 to 0.4, α (security impact weight) remained at 0.6, and γ (user level weight) decreased to 0.0 (total weights remained at 1). Process re-execution: The urgency (T1 urgency E is increased to 0.95), coupling degree and hierarchy are recalculated, and the final resource allocation remains stable.

[0111] In some embodiments, this method can combine "urgency" and "resource coupling" as two dimensions to evaluate task priority. Urgency is calculated by weighting security impact, deadline, and user level features using learnable coefficients (α, β, γ); resource coupling is obtained by normalizing the intersection of correlation factors generated by a graph neural network and task resource requirements. This two-dimensional evaluation addresses the limitations of traditional single-dimensional evaluations (such as evaluations based solely on cycles).

[0112] In some embodiments, the method can use a preset dual threshold of urgency (θ) high =0.85, θ low=0.3) and coupling degree double threshold (δ high =0.9、δ low =0.4), tasks are categorized into real-time critical, elastic optimization, and background batch processing levels. The categorization rules are directly linked to the dual-dimensional assessment results, ensuring that safety-critical tasks and tasks with strong resource coupling receive priority access to resources.

[0113] In some embodiments, environmental data such as traffic flow, road conditions, and network status are collected every 200 milliseconds and quantified into an environmental vector. The learnable coefficients (α, β, γ) for calculating urgency are dynamically adjusted using gradient descent, and the adjustment logic is strongly correlated with environmental features (e.g., increasing the safety weight α during congestion). This enables the system to adapt to different scenarios, solving the environmental adaptability defects of traditional fixed-parameter systems.

[0114] Figure 3 A schematic diagram of an onboard mission resource allocation device is shown. It should be understood that this device is related to... Figure 1 The method executed in the middle corresponds to the steps involved in the aforementioned method. The specific functions and effects of the device can be found in the description above. To avoid repetition, detailed descriptions are omitted here.

[0115] The vehicle-mounted mission resource allocation device includes: Acquisition unit 310 is used to collect multi-source data of vehicles in real time; The generation unit 320 is used to generate a dynamic task set based on multi-source data; The calculation unit 330 is used to calculate the urgency score and target coupling score of each task in the dynamic task set; Unit 340 is used to determine the task level of each task based on the urgency score and the target coupling score; Resource allocation unit 350 is used to allocate resources to each task according to task hierarchy.

[0116] In some embodiments, the acquisition unit 310 includes: The acquisition subunit 311 is used to acquire sensor perception data of the vehicle's surrounding environment in real time through on-board sensors. The first acquisition subunit 312 is used to acquire operating status parameters from the vehicle's internal systems; The receiving subunit 313 is used to receive user interaction requests issued by the user through voice, touch or other means; The aggregation subunit 314 is used to aggregate sensor sensing data, operating status parameters, and user interaction requests to obtain multi-source data.

[0117] In some embodiments, the generation unit 320 includes: The standardization subunit 321 is used to standardize multi-source data to obtain standardized data. Generating subunit 322 is used to generate a dynamic task set based on preset rules and standardized data.

[0118] In some embodiments, the computing unit 330 includes: The calculation subunit 331 is used to calculate the urgency score of each task in the dynamic task set; The computational subunit 331 is also used to calculate the resource requirement matrix for each task; The first determining subunit 332 is used to determine multiple task pairs based on a dynamic task set; The calculation subunit 331 is also used to calculate the intersection of inter-task resource requirements and task correlation factors for each task pair; The computational subunit 331 is also used to calculate the original coupling score of each task pair based on the task correlation factor and the intersection of resource requirements between tasks. Normalization subunit 333 is used to normalize the original coupling score to obtain the normalized coupling score for each task pair. The first determining subunit 332 is also used to determine the target coupling score of each task based on the normalized coupling score of each task pair.

[0119] In some embodiments, the calculation subunit 331 is specifically used to extract the task features of each task in the dynamic task set; wherein, the task features include at least security impact features, deadline features, and user level features; The urgency score for each task is calculated based on the learnable coefficients obtained from pre-training and the task characteristics.

[0120] In some embodiments, the computing subunit 331 is also used to identify the task type of each task; Calculate the task cost for each task based on the task type; the task cost includes computational cost and communication cost. Generate a resource requirement matrix for each task based on the task overhead.

[0121] In some embodiments, the computation subunit 331 is specifically used to determine the inter-task resource requirement intersection of each task pair based on the resource requirement matrix. Obtain the task feature matrix for each task; wherein, the task feature matrix includes at least the task type code, the normalized value of computation overhead, the normalized value of communication overhead, the number of historical co-occurrences, the resource overlap rate, and the security correlation degree; The task correlation factor for each task pair is generated by using a pre-defined graph neural network and task feature matrix.

[0122] In some embodiments, the determining unit 340 includes: The second acquisition subunit 341 is used to acquire a pre-set urgency threshold and coupling threshold; The second determining subunit 342 is used to determine the task level of each task based on the urgency threshold, coupling threshold, urgency score and target coupling score; The task is classified into one of the following levels: real-time critical level, elastic optimization level, or background batch processing level.

[0123] In some embodiments, the resource allocation unit 350 includes: The statistics subunit 351 is used to count the available computing resources of the vehicle computing platform; The statistics subunit 351 is also used to statistically analyze the first real-time critical level task and the second non-real-time critical level task in the dynamic task set according to the task classification. Allocation subunit 352 is used to allocate a preset proportion of available computing resources to the first task and pre-allocate a preset size of memory space for the first task; The third determining subunit 353 is used to determine the current remaining available resources; The allocation subunit 352 is also used to allocate the currently remaining available resources to the second task according to a preset allocation ratio.

[0124] like Figure 4 As shown, this application provides an electronic device 400, which includes a processor 401 and a memory 402. The processor 401 and the memory 402 are interconnected and communicate with each other through a communication bus 403 and / or other forms of connection mechanism (not shown). The memory 402 stores a computer program that can be executed by the processor 401. When the computing device is running, the processor 401 executes the computer program to perform the method in any of the aforementioned optional implementations.

[0125] This application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the method in any of the aforementioned optional implementations.

[0126] The computer-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0127] This application provides a computer program product, which includes a computer program that, when run by a processor, executes the method in any of the aforementioned optional implementations.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and not to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and they should all be covered within the scope of the claims and specification of this application. In particular, as long as there is no conflict, the various technical features mentioned in the embodiments can be combined in any way. This application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for allocating vehicle-mounted task resources, characterized in that, include: Real-time collection of multi-source vehicle data; A dynamic task set is generated based on the multi-source data; Calculate the urgency score and target coupling score for each task in the dynamic task set; Based on the urgency score and the target coupling score, the task level of each task is determined; Resource allocation is performed for each task according to the task classification.

2. The on-board task resource allocation method according to claim 1, characterized in that, The real-time acquisition of multi-source vehicle data includes: The vehicle uses onboard sensors to collect sensor data about the surrounding environment in real time. Obtain operating status parameters from the vehicle's internal systems; Receive user interaction requests via voice, touch, or other means; By summarizing the sensor data, the operating status parameters, and the user interaction requests, multi-source data is obtained.

3. The on-board task resource allocation method according to claim 1, characterized in that, The step of generating a dynamic task set based on the multi-source data includes: The multi-source data is standardized to obtain standardized data; A dynamic task set is generated based on preset rules and the standardized data.

4. The on-board mission resource allocation method according to claim 1, characterized in that, The calculation of the urgency score and target coupling score for each task in the dynamic task set includes: Calculate the urgency score for each task in the dynamic task set; Calculate the resource requirement matrix for each of the tasks; Multiple task pairs are determined based on the dynamic task set; Calculate the intersection of resource requirements and task correlation factor for each task pair; Based on the task relevance factor and the intersection of resource requirements between tasks, calculate the original coupling score for each task pair; The original coupling score is normalized to obtain the normalized coupling score for each task pair. The target coupling score for each task is determined based on the normalized coupling score for each task pair.

5. The on-board task resource allocation method according to claim 4, characterized in that, The calculation of the urgency score for each task in the dynamic task set includes: Extract the task features of each task in the dynamic task set; wherein, the task features include at least security impact features, deadline features, and user level features; An urgency score for each task is calculated based on the learnable coefficients obtained from pre-training and the task characteristics.

6. The on-board task resource allocation method according to claim 4, characterized in that, The calculation of the resource requirement matrix for each task includes: Identify the task type for each of the tasks; Calculate the task cost for each task based on the task type; wherein the task cost includes computational cost and communication cost; Generate a resource requirement matrix for each task based on the task overhead.

7. The on-board task resource allocation method according to claim 4, characterized in that, The calculation of the intersection of inter-task resource requirements and task correlation factor for each task pair includes: The task-to-task resource requirement intersection for each task pair is determined based on the resource requirement matrix. Obtain the task feature matrix for each task; wherein the task feature matrix includes at least the task type code, the normalized value of computation overhead, the normalized value of communication overhead, the number of historical co-occurrences, the resource overlap rate, and the security correlation degree. The task correlation factor for each task pair is generated using a preset graph neural network and the task feature matrix.

8. The method for allocating vehicle-mounted mission resources according to claim 1, characterized in that, The step of determining the task level for each task based on the urgency score and the target coupling score includes: Obtain the pre-set urgency threshold and coupling threshold; The task level of each task is determined based on the urgency threshold, the coupling threshold, the urgency score, and the target coupling score; The task is classified into one of the following levels: real-time critical level, elastic optimization level, and background batch processing level.

9. The on-board task resource allocation method according to claim 8, characterized in that, The step of allocating resources to each task according to the task hierarchy includes: The available computing resources of the vehicle-mounted computing platform are statistically analyzed. Based on the task classification, the first real-time critical task and the second non-real-time critical task in the dynamic task set are statistically analyzed. Allocate a preset proportion of the available computing resources to the first task, and pre-allocate a preset size of memory space to the first task; Determine the current remaining available resources; The remaining available resources are allocated to the second task according to a preset allocation ratio.

10. A vehicle-mounted task resource allocation device, characterized in that, The vehicle-mounted task resource allocation device includes: The data acquisition unit is used to collect multi-source data from vehicles in real time. The generation unit is used to generate a dynamic task set based on the multi-source data; A calculation unit is used to calculate the urgency score and target coupling score of each task in the dynamic task set; A determining unit is configured to determine the task level of each task based on the urgency score and the target coupling score; A resource allocation unit is used to allocate resources to each task according to the task hierarchy.

11. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory being used to store a computer program, and the processor running the computer program to cause the electronic device to perform the vehicle-mounted task resource allocation method according to any one of claims 1 to 9.

12. A readable storage medium, characterized in that, The readable storage medium stores a computer program, which, when executed by a processor, performs the vehicle-mounted task resource allocation method according to any one of claims 1 to 9.

13. A computer program product, characterized in that, The computer program product includes a computer program, which, when executed by a processor, performs the vehicle task resource allocation method according to any one of claims 1 to 9.

Images