Intelligent driving task unloading method based on mobile edge computing

Abstract

The invention discloses an intelligent driving task unloading method based on mobile edge computing. The intelligent driving task unloading method comprises the following steps of S1, environment perception and task identification, wherein the surrounding environment of a vehicle is perceived in real time through a vehicle-mounted sensor; and the intelligent driving task needing to be processed is identified according to the sensed environment information. According to the method, unloading task evaluation is divided into simple tasks and intensive tasks, cooperative unloading work of the vehicle-mounted equipment and the edge server is carried out, efficient unloading and execution of the intelligent driving tasks are achieved, the performance and the response speed of the intelligent driving system are improved, and through task segmentation, priority determination and dynamic adjustment of the execution sequence, the efficiency of the intelligent driving system is improved. The overall execution efficiency is optimized, further efficient and real-time processing of the intelligent driving task is achieved, the optimal dynamic adjustment strategy of the edge server is additionally selected according to the information monitored in real time when the load capacity is large, smooth and rapid execution of task unloading is ensured, and the unloading efficiency and stability are further optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent driving task offloading, and particularly to an intelligent driving task offloading method based on mobile edge computing. Background Art

[0002] Intelligent driving task offloading refers to the technology of offloading the computing tasks generated during the intelligent driving process to an edge server for processing. With the advent of the 5G era and the development of autonomous driving technology, the amount of data generated by intelligent driving vehicles and the application requirements have increased sharply. These data need to be processed quickly within a short time to meet the real-time and accuracy requirements of autonomous driving. As a new computing mode, edge computing deploys computing resources at the network edge, close to the data source, which can significantly reduce data transmission latency and improve processing efficiency. Therefore, it has become an important technical support for intelligent driving task offloading.

[0003] For the existing intelligent driving task offloading, by directly offloading the tasks to the edge server connected to the vehicle network or mobile communication network for processing, the following deficiencies exist when in use: 1. Since some tasks are simple tasks and can be directly completed on the in-vehicle device, uniformly offloading them to the edge server will inevitably greatly increase the processing volume of the edge server, resulting in relative latency and lacking the unified collaborative work between the two; 2. Multiple tasks do not distinguish priorities, which will inevitably reduce the task processing efficiency; 3. In the event of a load failure of the edge server, there is a lack of dynamic adjustment strategies to cope with emergencies, and it is difficult to ensure the smooth and rapid execution of task offloading. In view of this, the present application proposes an intelligent driving task offloading method based on mobile edge computing. Summary of the Invention

[0004] Based on the technical problems existing in the background art, the present invention proposes an intelligent driving task offloading method based on mobile edge computing.

[0005] An intelligent driving task offloading method based on mobile edge computing proposed by the present invention includes the following steps:

[0006] S1: Environment perception and task recognition: Real-time perceive the vehicle surrounding environment through in-vehicle sensors; and according to the perceived environmental information, recognize the intelligent driving tasks that need to be processed;

[0007] S2: Task evaluation and classification: Evaluate the computing requirements of the recognized tasks; and classify the tasks into computationally intensive tasks and simple tasks according to the computing requirements;

[0008] S3: Edge Server Discovery and Selection: Discover nearby edge servers and their computing capabilities and network status information through the vehicle network or mobile communication network; and select the most suitable edge server to establish a connection for task offloading according to the computing requirements, real-time requirements of the task, and the available resources of the edge server.

[0009] S4: Task Offloading Decision: Determine the proportion of task offloading according to factors such as the vehicle's moving speed, network bandwidth, and the processing capacity of the edge server; and formulate specific offloading strategies, including task splitting, data transmission methods, and task execution order.

[0010] S5: Task Offloading and Execution: Split the task into multiple subtasks according to the offloading strategy, and transmit the subtasks to be offloaded to the selected edge server through the wireless network; and execute the offloaded subtasks on the edge server, while the in-vehicle device executes the remaining tasks. The edge server uses its powerful computing power to quickly complete the task and returns the result to the in-vehicle device.

[0011] S6: Result Reception and Integration: The in-vehicle device receives the task execution results returned by the edge server; and integrates the results returned by the edge server with the results executed on the in-vehicle device to form a complete intelligent driving decision or output.

[0012] S7: Real-time Monitoring and Adjustment: During the task offloading and execution process, real-time monitor the vehicle's motion state, network status, and the status of the edge server to ensure the smooth progress of task offloading; and dynamically adjust the task offloading strategy according to the real-time monitored information.

[0013] S8: End and Feedback: When all tasks are executed and integrated, it marks the end of the intelligent driving task offloading process; and provide feedback and learning on the task offloading process, analyze the efficiency, cost, and reliability of task offloading, and provide optimization suggestions and experience for future task offloading.

[0014] Preferably, in the S1, the in-vehicle sensors include radar, camera, GPS, etc., the perceived vehicle surrounding environment includes road conditions, positions of other vehicles, traffic signals, etc., and the intelligent driving tasks include path planning, obstacle detection, traffic signal recognition, etc.

[0015] Preferably, in the S2, the task computing evaluation includes computing volume evaluation, computing complexity evaluation, and real-time requirement evaluation. The specific expressions used for evaluation are as follows:

[0016] S201: Computing Volume Evaluation: The computing volume is related to the amount of data processed by the task, the complexity of the algorithm, and the number of executions. The following formula is used to estimate the computing volume:

[0017] Computational amount = Image resolution × Frame rate × Algorithm complexity coefficient;

[0018] Among them, the algorithm complexity coefficient is an empirical value used to represent the computational amount of a specific algorithm when processing unit data;

[0019] S202: Computational complexity evaluation: Computational complexity is evaluated by the time complexity and space complexity of the algorithm. The time complexity represents the relationship between the algorithm execution time and the scale of the input data, and the space complexity represents the relationship between the storage space required for algorithm execution and the scale of the input data. The time complexity representation methods include O(n), O(n2), O(log n), where n represents the scale of the input data;

[0020] The time complexity is represented by the following formula:

[0021] T(n) = O(f(n));

[0022] Among them, T(n) represents the algorithm execution time, and f(n) is a function of n used to describe the relationship between the algorithm execution time and the scale of the input data; The space complexity is represented in the same way, with the difference being that it focuses on the usage of the storage space;

[0023] S203: Real-time requirement evaluation: The real-time requirement is set according to the specific requirements of the task, and whether the task meets the real-time requirement is evaluated by setting a response time threshold; For example, for the emergency braking system in an intelligent driving task, by setting a response time threshold (such as 100 milliseconds) and requiring that the execution time of the algorithm must be less than this threshold, the following formula is used to evaluate the real-time performance of the algorithm;

[0024]

[0025] If the real-time performance is greater than 1, it means the algorithm meets the real-time requirement; otherwise, it means the algorithm does not meet the real-time requirement.

[0026] Preferably, in the above S4, the proportion of task offloading is divided into simple tasks that can be directly completed on the in-vehicle device and intensive tasks that need to be executed on the edge server. Task segmentation is to split out the computationally intensive tasks, and the data transmission methods include 5G and WiFi, which are used to transmit the separated intensive tasks to the edge server with high computational requirements for execution.

[0027] Preferably, in the above S5, the specific logical steps for the edge server to execute the offloaded multiple subtasks are as follows:

[0028] S501: Determine task priority: Determine the execution priority of the task according to the real-time requirement, computational amount, and dependency relationship of the task;

[0029] S502: Arrange the execution order: Arrange the execution of tasks on the in-vehicle computing platform and the edge server in descending order of priority.

[0030] S503: Dynamically adjust the execution order: Dynamically adjust the execution order of tasks according to changes in vehicle speed, network bandwidth, and edge server processing capacity to optimize the overall execution efficiency.

[0031] Preferably, in the above S6, the specific logical steps for integrating the results returned by the edge server and the results executed on the in-vehicle device are as follows:

[0032] S601: Data reception: The in-vehicle device receives the task execution results returned by the edge server through a communication network (5G, WIFI).

[0033] S602: Data preprocessing: Check the received data to ensure its integrity and correctness; perform time synchronization processing on the data to ensure that the results returned by the edge server are consistent in time with the results executed on the in-vehicle device; and perform format conversion or unit unification on the data as needed for subsequent processing.

[0034] S603: Result fusion: Integrate the results returned by the edge server and the results executed on the in-vehicle device; the fusion methods include weighted average, voting mechanism, and Kalman filter algorithm, and the specific selection depends on the characteristics of the data and application requirements; for example, in environmental perception, fuse the perception data returned by the edge server with the in-vehicle sensor data to obtain a more accurate perception result of the surrounding environment; in path planning, fuse the path planning suggestions provided by the edge server with the path planning algorithm on the in-vehicle device to obtain a better path planning decision.

[0035] S604: Decision generation: Generate the final intelligent driving decision or output based on the fused results.

[0036] Preferably, in the above S603, the expression used for result fusion is:

[0037] R 最后 = f(R 边缘 , R 车辆 , θ);

[0038] Where, R 最后 represents the final integrated result, R 边缘 represents the result returned by the edge server, R 车辆 represents the result executed on the in-vehicle device, and θ represents the parameters or weights in the fusion algorithm.

[0039] Preferably, in S7, the dynamic adjustment of the task offloading strategy includes adjusting the offloading ratio and selecting other edge servers to cope with emergencies or optimize the task execution efficiency. The logical steps for selecting other edge servers are as follows:

[0040] S701: Evaluate the status of the current edge server: Real-time monitor key metrics such as the load condition, resource usage, and response latency of the current edge server. When problems such as overload, increased response latency, or insufficient resources occur in the current edge server, select other edge servers.

[0041] S702: Search for available edge servers: Search for available edge servers around the vehicle or within a wider network range, and obtain a list of available servers by interacting with the edge computing platform or querying the status information of the edge servers.

[0042] S703: Evaluate candidate servers: Evaluate each candidate edge server, considering factors such as its remaining resources, current load, and communication latency with the vehicle. The server with more remaining resources, lower current load, and smaller communication latency has a higher evaluation value.

[0043] S704: Select the optimal server: Based on the evaluation results, select the edge server with the highest evaluation value as the new offloading target.

[0044] The selection process is through the following expression:

[0045] Select the optimal edge server = argmax(S_i)[a * (R_i / L_i)-β*D_i];

[0046] Where S_i represents the i-th edge server, R_i represents the remaining resources of the server, L_i represents the current load of the server, D_i represents the communication latency with the vehicle, and α and β are weight coefficients used to balance the importance of resource utilization, load condition, and communication latency in the decision-making.

[0047] Compared with the existing technologies, the beneficial effects of the present invention are:

[0048] The present invention divides the offloading task evaluation into simple tasks that can be directly offloaded using in-vehicle devices and intensive tasks that require the use of edge servers for offloading, and conducts collaborative offloading work between in-vehicle devices and edge servers to achieve efficient offloading and execution of intelligent driving tasks, improve the performance and response speed of the intelligent driving system; cooperate with task segmentation, determination of priorities, and dynamic adjustment of the execution order to optimize the overall execution efficiency, achieve further efficient and real-time processing of intelligent driving tasks, and further improve the processing efficiency; and then cooperate with the dynamic adjustment strategy of selecting the optimal edge server according to the information monitored in real time when the load is large to cope with emergencies or further optimize the task execution efficiency, ensure the smooth and rapid execution of task offloading, and further optimize the offloading efficiency and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 FIG. is a flowchart of an intelligent driving task offloading method based on mobile edge computing proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] The present invention will be further described below in conjunction with specific embodiments.

[0051] Embodiment

[0052] Referring to Figure 1 , this embodiment proposes an intelligent driving task offloading method based on mobile edge computing, including the following steps:

[0053] S1: Environmental perception and task recognition: Real-time perception of the vehicle's surrounding environment through in-vehicle sensors; and according to the perceived environmental information, identify the intelligent driving tasks that need to be processed; where the in-vehicle sensors include radar, cameras, GPS, etc., the perceived vehicle surrounding environment includes road conditions, positions of other vehicles, traffic signals, etc., and the intelligent driving tasks include path planning, obstacle detection, traffic signal recognition, etc.;

[0054] S2: Task evaluation and classification: Conduct a computational requirement evaluation on the identified tasks; and classify the tasks into computationally intensive tasks and simple tasks according to the computational requirements; where computationally intensive tasks usually require a large amount of computing resources, while simple tasks can be quickly completed on in-vehicle devices; where the task computational evaluation includes computational volume evaluation, computational complexity evaluation, and real-time requirement evaluation, and the specific expressions used for evaluation are:

[0055] S201: Computational volume evaluation: The computational volume is related to the amount of data processed by the task, the complexity of the algorithm, and the number of executions. The following formula is used to estimate the computational volume:

[0056] Computational volume = image resolution × frame rate × algorithm complexity coefficient;

[0057] Among them, the algorithm complexity coefficient is an empirical value used to represent the amount of computation of a specific algorithm when processing unit data;

[0058] S202: Computational Complexity Evaluation: Computational complexity is evaluated through the time complexity and space complexity of the algorithm. The time complexity represents the relationship between the execution time of the algorithm and the scale of the input data, and the space complexity represents the relationship between the storage space required for the algorithm execution and the scale of the input data. The methods for representing the time complexity are O(n), O(n2), O(log n), where n represents the scale of the input data;

[0059] The time complexity is represented by the following formula:

[0060] T(n) = O(f(n));

[0061] Among them, T(n) represents the algorithm execution time, and f(n) is a function of n used to describe the relationship between the algorithm execution time and the scale of the input data; The space complexity is represented in the same way, with the difference being the focus on the usage of the storage space;

[0062] S203: Real-time Requirement Evaluation: The real-time requirement is set according to the specific needs of the task, and the task is evaluated for meeting the real-time requirement by setting a response time threshold; For example, for the emergency braking system in an intelligent driving task, by setting a response time threshold (such as 100 milliseconds) and requiring the execution time of the algorithm to be less than this threshold, the following formula is used to evaluate the real-time performance of the algorithm;

[0063]

[0064] If the real-time performance is greater than 1, it means the algorithm meets the real-time requirement; otherwise, it means the algorithm does not meet the real-time requirement;

[0065] S3: Edge Server Discovery and Selection: Through the vehicle network or mobile communication network, discover nearby edge servers and their computing capabilities and network status information; And based on the computing requirements, real-time requirements of the task, and the available resources of the edge server, select the most suitable edge server to establish a connection for task offloading;

[0066] S4: Task Offloading Decision: Determine the proportion of task offloading based on factors such as the vehicle's moving speed, network bandwidth, and the processing capacity of the edge server; And formulate specific offloading strategies, including task splitting, data transmission methods, and task execution order; Among them, the proportion of task offloading is divided into simple tasks that can be directly completed on in-vehicle devices and intensive tasks that need to be executed on the edge server. Task splitting is to split out the computationally intensive tasks, and the data transmission methods include 5G and WiFi, which are used to transmit the separated intensive tasks to the edge server with a large amount of computation for execution;

[0067] S5: Task offloading and execution: Split the task into multiple subtasks according to the offloading strategy, and transmit the subtasks to be offloaded to the selected edge server through the wireless network; and execute the offloaded subtasks on the edge server, while the vehicle-mounted device executes the remaining tasks. The edge server uses its powerful computing power to quickly complete the tasks and returns the results to the vehicle-mounted device. The specific logical steps for the edge server to execute multiple offloaded subtasks are as follows:

[0068] S501: Determine task priority: Determine the execution priority of the task according to the real-time requirements, computational complexity, and dependencies of the task;

[0069] S502: Arrange the execution order: Arrange the execution of tasks on the vehicle-mounted computing platform and the edge server in order from highest to lowest priority;

[0070] S503: Dynamically adjust the execution order: Dynamically adjust the execution order of tasks according to changes in vehicle speed, network bandwidth, and edge server processing capacity to optimize the overall execution efficiency;

[0071] Through the above strategy, it is possible to reasonably determine the task offloading ratio according to factors such as vehicle speed, network bandwidth, and edge server processing capacity, and formulate offloading strategies for specific task splitting, data transmission methods, and task execution order, which helps to achieve efficient and real-time processing of intelligent driving tasks and improve processing efficiency;

[0072] S6: Result reception and integration: The vehicle-mounted device receives the task execution results returned by the edge server; and integrates the results returned by the edge server with the results executed on the vehicle-mounted device to form a complete intelligent driving decision or output. The specific logical steps for integrating the results returned by the edge server with the results executed on the vehicle-mounted device are as follows:

[0073] S601: Data reception: The vehicle-mounted device receives the task execution results returned by the edge server through the communication network (5G, WIFI);

[0074] S602: Data preprocessing: Verify the received data to ensure its integrity and correctness; perform time synchronization processing on the data to ensure that the results returned by the edge server are consistent with the results executed on the vehicle-mounted device in terms of time; and perform format conversion or unit unification on the data as needed for subsequent processing;

[0075] S603: Result Fusion: Fuse the results returned by the edge server with the results executed on the in-vehicle device; the fusion methods include weighted average, voting mechanism, and Kalman filtering algorithm, and the specific selection depends on the characteristics of the data and application requirements; for example, in environmental perception, fuse the perception data returned by the edge server with the in-vehicle sensor data to obtain a more accurate perception result of the surrounding environment; in path planning, fuse the path planning suggestions provided by the edge server with the path planning algorithm on the in-vehicle device to obtain a better path planning decision;

[0076] S604: Decision Generation: Generate the final intelligent driving decision or output based on the fused results; including vehicle control instructions, path planning results, obstacle avoidance strategies, etc.;

[0077] The expression used for result fusion is:

[0078] R 最后 = f(R 边缘 , R 车辆 , θ);

[0079] Where, R 最后 represents the final integrated result, R 边缘 represents the result returned by the edge server, R 车辆 represents the result executed on the in-vehicle device, and θ represents the parameters or weights in the fusion algorithm;

[0080] Through the above reasonable steps of data reception, preprocessing, result fusion, and decision generation, as well as appropriate calculation formulas, the stability and safety of the intelligent driving system can be ensured;

[0081] S7: Real-time Monitoring and Adjustment: During the task offloading and execution process, real-time monitor the motion state of the vehicle, network condition, and the state of the edge server to ensure the smooth progress of task offloading; and dynamically adjust the task offloading strategy according to the real-time monitored information; the dynamic adjustment of the task offloading strategy includes adjusting the offloading ratio and selecting other edge servers to cope with emergencies or optimize the task execution efficiency; the logical steps for selecting other edge servers are as follows:

[0082] S701: Evaluate the current edge server status: Real-time monitor key indicators such as the load situation, resource usage, and response latency of the current edge server; when problems such as overload, increased response latency, or insufficient resources occur in the current edge server, select other edge servers;

[0083] S702: Search for available edge servers: Search for available edge servers within the vehicle's vicinity or a wider network range, and obtain a list of available servers by interacting with the edge computing platform or querying the status information of the edge servers;

[0084] S703: Evaluate candidate servers: Evaluate each candidate edge server, considering factors such as its remaining resources, current load, and communication latency with the vehicle; the server with more remaining resources, lower current load, and smaller communication latency has a higher evaluation value.

[0085] S704: Select the optimal server: According to the evaluation results, select the edge server with the highest evaluation value as the new offloading target.

[0086] The selection process is through the following expression:

[0087] Select the optimal edge server = argmax(S_i)[a * (R_i / L_i) - β * D_i];

[0088] where S_i represents the i-th edge server, R_i represents the remaining resources of the server, L_i represents the current load of the server, D_i represents the communication latency with the vehicle, and α and β are weight coefficients used to balance the importance of resource utilization, load situation, and communication latency in the decision-making.

[0089] Through the above logical steps and formulas, the optimal edge server can be dynamically selected according to the real-time monitored information to ensure the smooth progress of task offloading and optimize the task execution efficiency. This dynamic adjustment strategy can cope with emergencies such as the failure or overload of the current edge server, and can also optimize the task offloading decision according to the changes in network conditions and device loads, further optimizing the offloading efficiency and stability.

[0090] S8: End and feedback: When all tasks are executed and integrated, it marks the end of the intelligent driving task offloading process; and feedback and learning are carried out on the task offloading process, analyzing the efficiency, cost, and reliability of task offloading, providing optimization suggestions and experience for future task offloading.

[0091] In this embodiment, by dividing the offloading tasks into simple tasks that can be directly offloaded using in-vehicle devices and intensive tasks that require edge servers for offloading, the collaborative offloading work of in-vehicle devices and edge servers is carried out to achieve the efficient offloading and execution of intelligent driving tasks; by cooperating with task segmentation, determining priorities, and dynamically adjusting the execution order, the overall execution efficiency is optimized, realizing the further efficient and real-time processing of intelligent driving tasks and further improving the processing efficiency; and then by cooperating with the dynamic adjustment strategy of selecting the optimal edge server according to the real-time monitored information when the load is large to cope with emergencies or further optimize the task execution efficiency, ensuring the smooth and fast execution of task offloading, and further optimizing the offloading efficiency and stability.

[0092] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes, shall be covered by the protection scope of the present invention.

Claims

1. A method for intelligent driving task offloading based on mobile edge computing, characterized in that: The following steps are involved: S1: Environmental perception and task identification: Real-time perception of the vehicle's surrounding environment through on-board sensors; and identification of intelligent driving tasks that need to be processed based on the perceived environmental information; S2: Task evaluation and classification: Evaluate the computational requirements of the identified tasks and classify the tasks into computationally intensive tasks and simple tasks according to the computational requirements; S3: Edge server discovery and selection: Discover nearby edge servers and their computing capabilities and network status information through the Internet of Vehicles or mobile communication networks; and select the most suitable edge server to establish a connection for task offloading based on the computing requirements, real-time requirements, and available resources of the edge server; S4: Task offloading decision: Determine the proportion of task offloading based on the vehicle's moving speed, network bandwidth, and edge server processing capacity; and formulate specific offloading strategies, including task segmentation, data transmission method, and task execution order; S5: Task offloading and execution: The task is divided into multiple subtasks according to the offloading strategy, and the subtasks to be offloaded are transmitted to the selected edge server through the wireless network; the offloaded subtasks are executed on the edge server, while the vehicle-mounted device executes the remaining tasks. The edge server uses its powerful computing power to quickly complete the task and returns the result to the vehicle-mounted device; S6: Result reception and integration: The vehicle-mounted device receives the task execution results returned by the edge server; and integrates the results returned by the edge server with the results executed on the vehicle-mounted device to form a complete intelligent driving decision or output; S7: Real-time monitoring and adjustment: During the task offloading and execution process, the vehicle's motion status, network status, and edge server status are monitored in real time to ensure the smooth progress of task offloading; and the task offloading strategy is dynamically adjusted based on the real-time monitoring information; S8: End and feedback: When all tasks are executed and integrated, it marks the end of the intelligent driving task offloading process; feedback and learning are conducted on the task offloading process to analyze the efficiency, cost, and reliability of task offloading, and provide optimization suggestions and experience reference for future task offloading.

2. The method for offloading intelligent driving tasks based on mobile edge computing according to claim 1, characterized in that: In S1, the vehicle-mounted sensors include radar, camera, and GPS. The perceived vehicle surrounding environment includes road conditions, positions of other vehicles, and traffic signals. The intelligent driving tasks include path planning, obstacle detection, and traffic signal recognition.

3. The method for offloading intelligent driving tasks based on mobile edge computing according to claim 1, characterized in that: In S2, the task calculation evaluation includes the calculation amount evaluation, the calculation complexity evaluation, and the real-time requirement evaluation, and the specific evaluation expressions used are: S201: Calculation Amount Evaluation: The calculation amount is related to the amount of data processed by the task, the complexity of the algorithm, and the number of executions. The following formula is used to estimate the calculation amount: Calculation amount = image resolution × frame rate × algorithm complexity coefficient; Among them, the algorithm complexity coefficient is an empirical value used to indicate the amount of calculation of a specific algorithm when processing unit data; S202: Computational complexity evaluation: The computational complexity is evaluated by the time complexity and space complexity of the algorithm. The time complexity represents the relationship between the algorithm execution time and the input data size. The space complexity represents the relationship between the storage space required for the algorithm execution and the input data size. The time complexity can be expressed in O(n), O(n2), and O(log n), where n represents the size of the input data. The time complexity is expressed using the following formula: T(n) = O(f(n)); Where T(n) represents the algorithm execution time, and f(n) is a function of n, which is used to describe the relationship between the algorithm execution time and the input data size. The space complexity is expressed in the same way, but the difference is that it focuses on the usage of storage space. S203: Real-time requirement evaluation: The real-time requirement is set according to the specific needs of the task, and the response time threshold is set to evaluate whether the task meets the real-time requirement. For the emergency braking system in the intelligent driving task, a response time threshold is set and the execution time of the algorithm must be less than this threshold. The following formula is used to evaluate the real-time performance of the algorithm; If the real-time performance is greater than 1, it means that the algorithm meets the real-time requirements; otherwise, it means that the algorithm does not meet the real-time requirements.

4. The method for offloading intelligent driving tasks based on mobile edge computing according to claim 1 is characterized in that: In S4, the ratio of task offloading is divided into simple tasks that can be completed directly on the vehicle-mounted equipment and intensive tasks that need to be executed on the edge server. Task segmentation is to segment the computing-intensive tasks. The data transmission methods include 5G and WiFi, which are used to transmit the separated intensive tasks to the edge server with large computing amount for execution.

5. The method for offloading intelligent driving tasks based on mobile edge computing according to claim 1, characterized in that: In S5, the specific logical steps of the edge server executing the multiple subtasks of offloading are as follows: S501: Determine task priority: Determine the execution priority of the task according to the real-time requirements, computational load and dependency of the task; S502: Arrange execution order: arrange execution of tasks on the vehicle computing platform and the edge server in order from high to low priority; S503: Dynamically adjust the execution order: According to the changes in vehicle movement speed, network bandwidth and edge server processing capacity, dynamically adjust the execution order of tasks to optimize the overall execution efficiency.

6. The method for offloading intelligent driving tasks based on mobile edge computing according to claim 1, characterized in that: In S6, the specific logical steps of integrating the result returned by the edge server with the result executed on the vehicle-mounted device are as follows: S601: Data reception: The vehicle-mounted device receives the task execution result returned by the edge server through the communication network; S602: Data preprocessing: Verify the received data to ensure the integrity and correctness of the data; perform time synchronization on the data to ensure that the result returned by the edge server is consistent with the result executed on the vehicle-mounted device in time; And convert the data format or unify the units as needed for subsequent processing; S603: Result fusion: Fusion of the result returned by the edge server and the result executed on the vehicle-mounted device; S604: Decision generation: Based on the fusion results, generate the final intelligent driving decision or output.

7. The method for offloading intelligent driving tasks based on mobile edge computing according to claim 6 is characterized in that: In S603, the expression used for result fusion is: R 最后 =f(R 边缘 ,R 车辆 ,θ); Among them, R 最后 represents the final integration result, R 边缘 Indicates the result returned by the edge server, R 车辆 represents the result executed on the vehicle-mounted device, and θ represents the parameter or weight in the fusion algorithm.

8. The method for offloading intelligent driving tasks based on mobile edge computing according to claim 1, characterized in that: In S7, dynamically adjusting the task offloading strategy includes adjusting the offloading ratio and selecting other edge servers to cope with emergencies or optimize task execution efficiency; the logical steps of selecting other edge servers are as follows: S701: Evaluate the current edge server status: monitor the load, resource usage, and key indicators of response delay of the current edge server in real time; if the current edge server is overloaded, the response delay increases, or the resources are insufficient, select another edge server; S702: Searching for available edge servers: searching for available edge servers around the vehicle or in a wider network range, and obtaining a list of available servers by interacting with the edge computing platform or querying the status information of the edge server; S703: Evaluate candidate servers: Evaluate each candidate edge server, taking into account its remaining resources, current load, and communication delay with the vehicle; the server with more remaining resources, lower current load, and smaller communication delay has a higher evaluation value; S704: Select the best server: According to the evaluation result, select the edge server with the highest evaluation value as the new offloading target; The selection process is expressed as follows: Select the optimal edge server = argmax(S_i)[a*(R_i / L_i)-β*D_i]; Where S_i represents the i-th edge server, R_i represents the remaining resources of the server, L_i represents the current load of the server, D_i represents the communication delay with the vehicle, and α and β are weight coefficients used to balance the importance of resource utilization, load conditions, and communication delay in decision making.

Images