An intelligent driving edge collaborative computing system for highways

Abstract

This invention discloses an edge collaborative computing system for intelligent driving on highways, relating to the field of data processing technology. It includes an information acquisition module: obtaining real-time tasks and past tasks based on information to be processed and previously processed information acquired by the intelligent driving vehicle. This invention uses a dynamic routing method to real-time route the corresponding processing nodes of real-time tasks according to the target level and judgment results, thereby improving the processing efficiency and accuracy of real-time tasks. The process of collaboratively processing real-time tasks based on the routing results with the processing nodes avoids the problem that fixed hierarchical partitioning and static task allocation are difficult to adapt to rapid changes in dynamic network conditions, latency budgets, and task importance. This also reduces the possibility of unacceptable delays or false alarms at the edge due to resource contention for urgent tasks.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to an intelligent driving edge collaborative computing system applied to highways. Background Technology

[0002] The intelligent driving edge collaborative computing system is a system that works collaboratively among computing resources such as vehicles, roadside equipment, and cloud servers. Through local and distributed computing capabilities, it performs rapid and reliable processing of perception, decision-making, and control to achieve a high level of driving assistance and automation functions. In collaborative computing, task allocation is particularly important.

[0003] A patent publication number CN112306673A discloses an SDN system for cloud computing and edge computing collaboration. The system comprises a base layer and an application layer. The base layer manages central cloud devices and edge cloud devices respectively, and discovers link information between them through a link layer discovery protocol, forming a network topology between the central cloud devices and edge cloud devices. The central cloud devices are used for cloud computing, and the edge cloud devices are used for edge computing. The application layer includes resource collaboration, security collaboration, and service collaboration, used to establish connections between the central cloud devices and edge cloud devices. The SDN system described in this specification can provide unified management of the central cloud and edge cloud, enabling efficient collaborative operation and providing a better user experience.

[0004] The aforementioned and existing technologies have the following drawbacks in edge collaborative computing systems: fixed hierarchical partitioning and static task allocation are difficult to adapt to the rapid changes in dynamic network states, latency budgets, and task importance, which may lead to unacceptable delays or false alarms for urgent tasks even at the edge due to resource contention. At the same time, the comprehensive optimization capability for multi-task concurrency is insufficient, making it difficult to achieve an efficient trade-off of "rapid response first, followed by in-depth analysis" in complex multi-scenario situations. Therefore, this invention is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent driving edge collaborative computing system for highways, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent driving edge collaborative computing system applied to highways, the system comprising:

[0007] Information acquisition module: Based on the information to be processed and the information previously processed by the intelligent driving vehicle, obtain the real-time task and the previous task, and obtain the processing node;

[0008] Information processing module: The processing nodes include edge computing nodes and cloud computing nodes, which process real-time tasks based on the processing nodes;

[0009] The rating module: Based on real-time tasks, the rating method judges and classifies the processing time of real-time tasks according to past tasks to obtain the result information, which includes the target rating, task time and task timeliness.

[0010] The collaborative adjustment module coordinates the way processing nodes handle real-time tasks through collaborative adjustment methods to obtain demand coordination, and processes real-time tasks based on the coordination of processing nodes with demand coordination.

[0011] Factor Judgment Module: Based on the real-time task, the module uses a factor judgment method to determine whether the processing time of the real-time task by the cloud computing node meets the task time requirement, and obtains the judgment result.

[0012] Dynamic task allocation module: Based on real-time tasks, target levels, and judgment results, the module performs real-time task allocation using dynamic allocation methods to obtain allocation results and target tasks. Based on the allocation results, the module coordinates with processing nodes to perform collaborative processing of real-time tasks.

[0013] Enriched processing modules: Based on the target task, the processing nodes enrich the target task processed by the edge computing nodes by using enriched processing methods.

[0014] Furthermore, the method for determining the level includes: splitting previous tasks into several previous sub-tasks, obtaining the processing time of previous sub-tasks at processing nodes to obtain previous sub-timeliness, obtaining the demand response time of previous sub-tasks, establishing the association between previous sub-tasks, previous sub-timeliness, and demand response time, establishing an auxiliary directory to store previous sub-timeliness, previous sub-tasks, demand response time, and association, classifying real-time tasks into levels based on real-time tasks using a partitioning method to obtain target level, task time, and task timeliness, and integrating target level, task time, and task timeliness to obtain result information.

[0015] Furthermore, the partitioning method includes: traversing the auxiliary directory of real-time tasks and performing similarity matching with previous sub-tasks to obtain matching results, the matching results including matching objects and matching values, a preset matching threshold, when the matching value exceeds the matching threshold, extracting the demand response time and previous sub-timeliness corresponding to the matching object from the auxiliary directory based on the association relationship to obtain the first matching time and the second matching time, the first matching time being the task time and the second matching time being the task timeliness, determining the number of real-time tasks, when the number of real-time tasks is single, the real-time task is the first priority, the first priority being the target level, when the number of real-time tasks is multiple, sorting the real-time tasks based on the number of task times to obtain a sorting result, adding levels to the real-time tasks based on the sorting result to obtain the target level, when the number of real-time tasks is multiple and there are real-time tasks whose matching values ​​do not exceed the matching threshold, then the real-time tasks whose matching values ​​do not exceed the matching threshold are promoted to the first priority.

[0016] Furthermore, the collaborative adjustment method includes: setting a preset time segmentation threshold, determining the relationship between the task time of the real-time task and the time segmentation threshold, processing the real-time task through cloud computing nodes when the task time of the real-time task is less than the time segmentation threshold, and processing the real-time task through edge computing nodes when the task time of the real-time task is greater than the time segmentation threshold or the real-time task is the first priority, and integrating the allocation of processing nodes to process real-time tasks to obtain demand coordination.

[0017] Furthermore, the factor judgment method includes: real-time monitoring of information upload speed to obtain real-time upload speed, obtaining a fixed value by the difference between task time and task timeliness, setting a speed threshold based on the fixed value and the information size of the real-time task, judging the relationship between the real-time upload speed and the speed threshold to obtain a judgment result, when the real-time upload speed is greater than the speed threshold, the judgment result is fed back as the time for the cloud computing node to process the real-time task conforms to the task time, when the real-time upload speed is less than the speed threshold, the judgment result is fed back as the time for the cloud computing node to process the real-time task does not conform to the task time.

[0018] Furthermore, the dynamic task allocation method includes: when the judgment result indicates that the processing time of the real-time task by the cloud computing node does not match the task time, the real-time task is processed through the edge computing node and marked as a first selected task; the target level of the first selected task is increased; the first selected task is processed through the edge computing node as a first switching result; based on the real-time task, the processing method of the real-time task is switched in real time according to the content of the real-time task using an emergency extraction method to obtain a second switching result and a second selected task; the first switching result and the second switching result are integrated to obtain a task allocation result, and the first selected task and the second selected task are the target tasks.

[0019] Furthermore, the emergency extraction method includes: acquiring emergency processing vocabulary, establishing a repository for storing emergency processing vocabulary, determining whether emergency processing vocabulary exists in the repository in a real-time task, when emergency processing vocabulary exists in the repository in a real-time task, extracting the real-time task containing the emergency processing vocabulary to obtain a second selected task, increasing the target level of the second selected task, processing the second selected task through an edge computing node, and simultaneously processing the second selected task through an edge computing node to obtain a second switching result.

[0020] Furthermore, the enrichment processing method includes: elevating the target task to the first priority and processing it through a cloud computing node, integrating the processing results of the target task on the edge computing node and the cloud computing node, and outputting the target task that has completed the enrichment processing on the edge computing node.

[0021] Compared with the prior art, the beneficial effects of the present invention are:

[0022] This intelligent driving edge collaborative computing system applied to highways uses a factor judgment method to determine whether the processing time of real-time tasks by cloud computing nodes matches the task time of the real-time task. The judgment result is used to facilitate the operation of the subsequent dynamic traffic allocation method. The dynamic traffic allocation method allocates the corresponding processing nodes of the real-time task in real time according to the target level of the real-time task and the judgment result, so as to improve the processing efficiency and accuracy of real-time tasks. Based on the traffic allocation result, the processing nodes coordinate to process the real-time task. This process avoids the problem that fixed hierarchical division and static task allocation are difficult to adapt to the rapid changes in dynamic network conditions, latency budgets and task importance. It also reduces the possibility of unacceptable delays or false alarms in urgent tasks at the edge due to resource competition.

[0023] Meanwhile, by setting up a variety of processing methods, we can perform in-depth analysis of the target tasks processed by edge computing nodes according to usage requirements, thereby enriching the processing results of the target tasks and achieving an efficient balance between rapid response and in-depth analysis. By setting up a classification method, we can classify real-time tasks by level and fuzzily determine the processing time through auxiliary directories. By using past sub-times, including past edge times and past cloud times, we can facilitate in-depth analysis of the timeliness of different processing nodes in processing tasks.

[0024] Meanwhile, when there are multiple real-time tasks and the matching value of a real-time task does not exceed the matching threshold, the real-time task is given the highest priority. This allows for the priority processing of tasks that have not yet appeared, thereby increasing the alertness of task processing. At the same time, the processing result of the task can be stored in an auxiliary directory to improve the data richness of the auxiliary directory. By judging the relationship between the task time of the real-time task and the time segmentation threshold, the processing nodes for processing real-time tasks can be adjusted to improve the processing efficiency of the processing nodes for processing real-time tasks, so as to quickly respond to real-time tasks within the required time. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the overall process of the present invention;

[0026] Figure 2 This is a schematic diagram of the grade determination method of the present invention;

[0027] Figure 3 This is a schematic diagram of the grade determination method of the present invention;

[0028] Figure 4 This is a schematic diagram of the emergency extraction method of the present invention;

[0029] Figure 5 This is a schematic diagram of the sorting structure for multiple real-time tasks according to the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] Intelligent driving on highways is closely related to edge collaborative computing. Intelligent driving requires highly reliable, low-latency perception, decision-making, and control capabilities. Edge collaborative computing rapidly processes, merges, and shares data on nearby edge servers / devices or cloud servers, enabling low-latency information collaboration, safety warnings, and collaborative decision-making between vehicles and between vehicles and infrastructure. This improves lane-level collaborative parking, platooning, and conflict prevention capabilities, while reducing reliance on the cloud and enhancing system scalability and robustness. In short, edge collaborative computing provides the real-time performance and collaborative efficiency required for intelligent driving on highways, and is a key support for achieving efficient, safe, and large-scale intelligent transportation.

[0032] like Figures 1-5As shown, the present invention provides a technical solution: an intelligent driving edge collaborative computing system applied to highways, the system comprising:

[0033] Information acquisition module: Based on the information to be processed and the information previously processed by the intelligent driving vehicle, obtain the real-time task and the previous task, and obtain the processing node;

[0034] It is important to note that both real-time tasks and past tasks refer to information that needs to be processed, i.e., data information generated during the intelligent driving process of the vehicle that needs to be processed. Task types include, but are not limited to, emergency tasks and daily tasks. Emergency tasks include sudden obstacle avoidance, forward accident warning, lane change response, etc.; daily tasks include route planning, traffic flow prediction, vehicle status monitoring, etc. Real-time tasks are obtained by acquiring information that needs to be processed during the intelligent driving process of the intelligent driving vehicle, while past tasks are obtained by acquiring information that needs to be processed during the intelligent driving process of the vehicle.

[0035] Information processing module: The processing nodes include edge computing nodes and cloud computing nodes, which process real-time tasks based on the processing nodes;

[0036] It is important to note that processing nodes are the nodes that process tasks, including edge computing nodes and cloud computing nodes. Compared to cloud computing nodes, edge computing nodes have a faster response speed, but their information processing accuracy is not as deep. Edge computing nodes, such as servers in vehicles, roadside units (RSUs), or MEC servers, are often deployed on highways or in service areas. They provide low-latency, high-reliability local computing resources for handling urgent tasks and critical decisions. Cloud computing nodes, or cloud computing centers, are typically used to centrally process non-real-time tasks, providing capabilities such as big data analysis and global path optimization to enable collaboration between autonomous vehicles and traffic system-level optimization. The process of processing real-time tasks based on processing nodes involves transmitting real-time tasks to processing nodes, processing them, and outputting response results, enabling the autonomous vehicle to complete the real-time task processing.

[0037] The rating module: Based on real-time tasks, the rating method judges and classifies the processing time of real-time tasks according to past tasks to obtain the result information, which includes the target rating, task time and task timeliness.

[0038] It is important to note that the grading method is used to determine the grading efficiency of real-time tasks based on the processing efficiency of previous tasks. The result information is obtained by judging and grading the processing efficiency of real-time tasks. The target grading is the corresponding grading of the real-time task. Processing nodes prioritize processing real-time tasks with higher target grading. The task time is the response time that the intelligent driving vehicle needs to receive from the processing node to process the real-time task. The task efficiency is the time required for the processing node to complete the processing of the real-time task.

[0039] The collaborative adjustment module coordinates the way processing nodes handle real-time tasks through collaborative adjustment methods to obtain demand coordination, and processes real-time tasks based on the coordination of processing nodes with demand coordination.

[0040] It is important to note that by setting up a collaborative adjustment method to distribute the processing of real-time tasks by the processing nodes, the processing accuracy can be improved while ensuring response efficiency, and the working efficiency of the processing nodes can be increased at the same time.

[0041] Factor Judgment Module: Based on the real-time task, the module uses a factor judgment method to determine whether the processing time of the real-time task by the cloud computing node meets the task time requirement, and obtains the judgment result.

[0042] It is important to note that the judgment result is obtained by using a factor judgment method to determine whether the time for cloud computing nodes to process real-time tasks matches the task time of the real-time task, so as to facilitate the subsequent operation of the dynamic traffic distribution method.

[0043] Dynamic task allocation module: Based on real-time tasks, target levels, and judgment results, the module performs real-time task allocation using dynamic allocation methods to obtain allocation results and target tasks. Based on the allocation results, the module coordinates with processing nodes to perform collaborative processing of real-time tasks.

[0044] It is important to note that by setting a dynamic traffic splitting method, the corresponding processing nodes of real-time tasks are split in real time according to the target level and judgment results of the real-time tasks. This improves the processing efficiency and accuracy of real-time tasks. The process of coordinating the processing nodes to process real-time tasks based on the traffic splitting results avoids the situation where fixed hierarchical division and static task allocation are difficult to adapt to the rapid changes in dynamic network conditions, latency budgets and task importance. This also reduces the possibility of unacceptable delays or false alarms for urgent tasks at the edge due to resource contention.

[0045] Enriched processing modules: Based on the target task, the processing nodes enrich the target task processed by the edge computing nodes by using enriched processing methods.

[0046] It is important to note that by setting up a variety of processing methods, the target tasks processed by edge computing nodes can be analyzed in depth according to usage requirements, thereby enriching the processing results of the target tasks and achieving an efficient trade-off between rapid response and in-depth analysis.

[0047] like Figure 2As shown, the method for determining the level includes: splitting previous tasks into several previous sub-tasks, obtaining the processing time of previous sub-tasks at processing nodes to obtain previous sub-timeliness, obtaining the demand response time of previous sub-tasks, establishing the association between previous sub-tasks, previous sub-timeliness, and demand response time, establishing an auxiliary directory to store previous sub-timeliness, previous sub-tasks, demand response time, and association, classifying real-time tasks into levels based on real-time tasks using a partitioning method to obtain target level, task time, and task timeliness, and integrating target level, task time, and task timeliness to obtain result information.

[0048] It is important to note that obtaining the processing time of past subtasks at the processing node involves two steps: First, obtaining the processing time of past subtasks at the processing node (i.e., obtaining the time taken by the processing node to process past subtasks); second, obtaining the required response time of past subtasks (i.e., obtaining the time range within which the autonomous vehicle needs to process the results—the specific timeframe can be set according to actual usage). Third, establishing the correlation between past subtasks, their corresponding processing times, and required response times involves finding the corresponding processing times and required response times for each past subtask and establishing this correlation. This process can be performed when obtaining past sub-timeframes and demand response times. By setting the classification method, real-time tasks can be classified and their processing timeframes can be fuzzily determined through the auxiliary directory. Past sub-timeframes include past edge timeframes and past cloud timeframes. Past edge timeframes refer to the processing timeframes of past sub-tasks on edge computing nodes, while past cloud timeframes refer to the processing timeframes of past sub-tasks on cloud computing nodes. By dividing past sub-timeframes into past edge timeframes and cloud timeframes, it is easy to conduct in-depth comparisons of the processing timeframes of different processing nodes.

[0049] like Figure 2 and Figure 5 As shown, the partitioning method includes: traversing the auxiliary directory based on real-time tasks and performing similarity matching with previous sub-tasks to obtain matching results. The matching results include matching objects and matching values. A preset matching threshold is used. When the matching value exceeds the matching threshold, the requirement response time and previous sub-timeliness corresponding to the matching object are extracted from the auxiliary directory based on the association relationship to obtain the first matching time and the second matching time. The first matching time is the task time, and the second matching time is the task timeliness. The number of real-time tasks is determined. When there is only one real-time task, the real-time task is given the first priority, which is the target level. When there are multiple real-time tasks, the real-time tasks are sorted based on the number of task times to obtain a sorting result. Based on the sorting result, the real-time tasks are added to obtain the target level. When there are multiple real-time tasks and there are real-time tasks whose matching values ​​do not exceed the matching threshold, the real-time tasks whose matching values ​​do not exceed the matching threshold are promoted to the first priority.

[0050] It's important to note that the process of obtaining matching results based on the real-time task's traversal of the auxiliary directory and similarity matching with previous subtasks involves selecting the real-time task and its previous subtask in the auxiliary directory for similarity matching, extracting the previous subtask with the highest similarity to obtain the matching object and matching value. The matching value is the similarity result. The process of setting a matching threshold involves pre-setting the matching threshold based on actual usage. The higher the matching threshold, the more accurate the subsequently obtained task time, processing efficiency, and target level. By prioritizing real-time tasks when there are multiple tasks and the matching value of a real-time task does not exceed the matching threshold, the real-time task is given first priority. This allows for the priority processing of tasks that have not yet appeared, increasing the alertness of task processing. Simultaneously, the processing result of this task can be stored in the auxiliary directory to improve the data richness of the auxiliary directory. Figure 5 This represents the process of sorting multiple real-time tasks.

[0051] like Figure 3 As shown, the collaborative adjustment method includes: setting a preset time segmentation threshold, judging the relationship between the task time of the real-time task and the time segmentation threshold, when the task time of the real-time task is less than the time segmentation threshold, the real-time task is processed by cloud computing nodes, when the task time of the real-time task is greater than the time segmentation threshold or the real-time task is the first priority, the real-time task is processed by edge computing nodes, and the allocation of processing nodes to process real-time tasks is integrated to obtain demand coordination.

[0052] It is important to note that the time segmentation threshold is a specific time value, preset according to actual conditions. Specifically, it can be the maximum time for processing real-time tasks. By judging the relationship between the task time of a real-time task and the time segmentation threshold, the processing nodes handling real-time tasks are adjusted to improve their processing efficiency, enabling rapid response to real-time tasks within the required timeframe. Figure 3 As shown, moving to the right indicates that the task time is greater than the time segmentation threshold, while moving to the left indicates that the person's time is less than the time segmentation threshold.

[0053] like Figure 1 As shown, the factor judgment method includes: real-time monitoring of information upload speed to obtain real-time upload speed, obtaining a fixed value by the difference between task time and task timeliness, setting a speed threshold based on the fixed value and the information size of the real-time task, judging the relationship between the real-time upload speed and the speed threshold to obtain the judgment result. When the real-time upload speed is greater than the speed threshold, the judgment result is that the time for the cloud computing node to process the real-time task is in line with the task time. When the real-time upload speed is less than the speed threshold, the judgment result is that the time for the cloud computing node to process the real-time task is not in line with the task time.

[0054] It is important to note that when processing real-time tasks through cloud servers, information transmission is required. By monitoring the information upload speed in real time and comparing the difference between the task time and the task timeliness, a judgment result can be obtained. This allows for a fuzzy assessment of whether the cloud computing nodes can complete the real-time task within the task time, thereby improving response accuracy. The process of setting a speed threshold based on a fixed value and the size of the real-time task information involves calculating the speed threshold within a fixed time frame in conjunction with the size of the real-time task information being transmitted.

[0055] like Figure 1 As shown, the dynamic task allocation method includes: when the judgment result indicates that the processing time of the real-time task by the cloud computing node does not match the task time, the real-time task is processed through the edge computing node and marked as a first selected task. The target level of the first selected task is increased, and the first selected task is processed through the edge computing node as a first switching result. Based on the real-time task, the processing method of the real-time task is switched in real time according to the content of the real-time task using an emergency extraction method to obtain a second switching result and a second selected task. The first switching result and the second switching result are integrated to obtain the allocation result, and the first selected task and the second selected task are the target tasks.

[0056] It is important to note that when the processing time of a real-time task by a cloud computing node does not match the task time, processing the real-time task through an edge computing node can improve information processing efficiency. By setting an emergency extraction method, the processing method of the real-time task can be switched according to the content of the real-time task to avoid missing urgent information and improve the efficiency of processing truly urgent real-time tasks. At the same time, real-time tasks can be marked to facilitate the subsequent implementation of richer processing methods.

[0057] like Figure 4 As shown, the emergency extraction method includes: acquiring emergency processing vocabulary, establishing a repository to store emergency processing vocabulary, determining whether emergency processing vocabulary exists in the repository in a real-time task, extracting the real-time task containing the emergency processing vocabulary to obtain a second selected task, increasing the target level of the second selected task, processing the second selected task through an edge computing node, and simultaneously processing the second selected task through an edge computing node to obtain a second switching result.

[0058] It is important to note that the process of acquiring emergency response vocabulary involves obtaining vocabulary that matches the expression format in the real-time task. This can be done by staff adding and setting their own emergency response vocabulary, such as sudden obstacle avoidance, forward accident warning, lane change response, etc. The specific emergency response vocabulary is determined based on the actual usage and does not necessarily have to be in text form. It can also be set according to the code expression in the real-time task.

[0059] like Figure 1 As shown, the enriched processing method includes: elevating the target task to the first priority and processing it through cloud computing nodes, integrating the processing results of the target task on edge computing nodes and cloud computing nodes, and outputting the target task that has been enriched by edge computing nodes.

[0060] It is important to note that real-time tasks processed by edge computing nodes are marked and renamed as target tasks. The target tasks are then processed by cloud computing nodes, which can enrich the accuracy of target task processing and reflect the efficient trade-off between collaborative computing and rapid response followed by in-depth analysis.

[0061] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended embodiments and their equivalents.

Claims

1. An intelligent driving edge collaborative computing system applied to a highway, the system comprising: an information acquisition module: based on an intelligent driving vehicle, obtaining real-time tasks and past tasks from information to be processed and past processed information, and obtaining processing nodes; characterized in that the system further comprises: an information processing module: the processing nodes comprise edge computing nodes and cloud computing nodes, and the real-time tasks are processed based on the processing nodes; a level determination module: based on the real-time tasks, the processing time of the real-time tasks is judged and divided according to the past tasks by a level determination method to obtain result information, the result information comprising a target level, a task time and a task time efficiency; a collaborative adjustment module: the way of processing the real-time tasks by the processing nodes is cooperated to obtain a demand cooperation by a collaborative adjustment method, and the real-time tasks are processed based on the demand cooperation of the processing nodes; a factor judgment module: based on the real-time tasks, whether the time of processing the real-time tasks by the cloud computing nodes meets the task time is judged to obtain a judgment result by a factor judgment method; a dynamic shunting module: based on the real-time tasks, the target level and the judgment result, real-time shunting is performed by a dynamic shunting method to obtain a shunting result and a target task, and the real-time tasks are cooperatively processed based on the shunting result and the processing nodes; a processing enrichment module: based on the target task and the processing nodes, the target task processed by the edge computing nodes is enriched by an enrichment processing method; the level determination method comprises: splitting the past tasks to obtain a plurality of past subtasks, obtaining past subtask processing time efficiency of the past subtasks in the processing nodes to obtain past subtask efficiency, obtaining demand response time of the past subtasks, establishing an association relationship among the past subtasks, the past subtask efficiency and the demand response time, establishing an auxiliary directory for storing the past subtask efficiency, the past subtasks, the demand response time and the association relationship, and based on the real-time tasks, the real-time tasks are divided into levels by a division method to obtain the target level, the task time and the task time efficiency, and the target level, the task time and the task time efficiency are integrated to obtain the result information; the division method comprises: based on the real-time tasks, similarity matching is performed between the auxiliary directory and the past subtasks to obtain a matching result, the matching result comprising a matching object and a matching value, a preset matching threshold is set, when the matching value exceeds the matching threshold, the demand response time and the past subtask efficiency corresponding to the matching object in the auxiliary directory are extracted based on the association relationship to obtain a first matching time and a second matching time, the first matching time is the task time, and the second matching time is the task time efficiency, the number of real-time tasks is judged, when the number of real-time tasks is single, the real-time task is a first priority, the first priority is the target level, when the number of real-time tasks is multiple, the real-time tasks are sorted based on the number of task times to obtain a sorting result, and the target level is obtained by adding the level of the real-time tasks in reverse based on the sorting result, and when the number of real-time tasks is multiple and there is a real-time task whose matching value does not exceed the matching threshold, the real-time task whose matching value does not exceed the matching threshold is promoted to the first priority. 2.The intelligent driving edge collaborative computing system applied to the expressway according to claim 1, wherein: The cooperative regulation method comprises the following steps: presetting a time segmentation threshold, judging the size relationship between the task time of the real-time task and the time segmentation threshold, processing the real-time task by a cloud computing node when the task time of the real-time task is less than the time segmentation threshold, processing the real-time task by an edge computing node when the task time of the real-time task is greater than the time segmentation threshold or the real-time task is of a first priority, and obtaining demand cooperation by integrating the allocation of the processing nodes for processing the real-time task. 3.The intelligent driving edge collaborative computing system applied to the expressway according to claim 1, characterized in that: The factor judgment method comprises the following steps: obtaining a real-time uploading speed by monitoring the information uploading speed in real time, obtaining a fixed value by obtaining the difference between the task time and the task time limit, setting a speed threshold based on the fixed value and the information size of the real-time task, obtaining a judgment result by judging the size relationship between the real-time uploading speed and the speed threshold, and feeding back the judgment result as the time for processing the real-time task by the cloud computing node being in conformity with the task time when the real-time uploading speed is greater than the speed threshold, and feeding back the judgment result as the time for processing the real-time task by the cloud computing node being not in conformity with the task time when the real-time uploading speed is less than the speed threshold. 4.The intelligent driving edge collaborative computing system applied to the expressway of claim 1, wherein: The dynamic shunting method comprises the following steps: when the judgment result is fed back as the time for processing the real-time task by the cloud computing node being not in conformity with the task time, processing the real-time task by the edge computing node and marking the real-time task to obtain a first selected task, improving the target level of the first selected task, processing the first selected task by the edge computing node to obtain a first switching result, obtaining a second switching result and a second selected task by real-time switching of the processing mode of the real-time task according to the content of the real-time task based on the real-time task by an emergency extraction method, and integrating the first switching result and the second switching result to obtain a shunting result, the first selected task and the second selected task being target tasks. 5.The intelligent driving edge collaborative computing system applied to the expressway according to claim 4, characterized in that: The emergency extraction method comprises the following steps: obtaining an emergency processing vocabulary, establishing a storage library for storing the emergency processing vocabulary, judging whether the emergency processing vocabulary in the storage library exists in the real-time task, obtaining the second selected task by extracting the real-time task with the emergency processing vocabulary in the storage library when the emergency processing vocabulary in the storage library exists in the real-time task, improving the target level of the second selected task, processing the second selected task by the edge computing node, and simultaneously processing the second selected task by the edge computing node to obtain the second switching result. 6.The intelligent driving edge collaborative computing system applied to the expressway according to claim 1, characterized in that: The rich processing method comprises the following steps: improving the target task to the first priority and processing the target task by the cloud computing node, integrating the processing results of the target task in the edge computing node and the cloud computing node, and outputting the target task processed by the rich edge computing node.

Images