Vehicle-road cooperative intelligent driving method and system based on calculation unloading, and medium

Abstract

The invention discloses a vehicle-road cooperative intelligent driving method and system based on calculation unloading and a medium, and the method comprises the steps: obtaining a plurality of intelligent driving calculation tasks of a target vehicle, and determining the real-time requirements, the safety influence degree and the calculation complexity of each intelligent driving calculation task; determining the task priority of each intelligent driving calculation task according to the real-time requirement, the safety influence degree and the calculation complexity; obtaining the current device state of each roadside device of the area where the target vehicle is located, and performing task allocation decision on each intelligent driving calculation task according to the current device state and the task priority to obtain a target roadside device corresponding to each intelligent driving calculation task; each intelligent driving calculation task is unloaded to the corresponding target roadside equipment for calculation, and a calculation result of the target roadside equipment is fed back to the target vehicle. The method improves the reliability and safety of intelligent driving, and can be widely applied to the technical field of intelligent driving.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent driving, and in particular to a vehicle-road collaborative intelligent driving method, system and medium based on computing offloading. Background Art

[0002] At present, intelligent driving mainly adopts the combination of multi-sensor fusion and a high-computing-power intelligent driving central integrated controller, usually based on the "perception - decision - control" architecture mode, specifically as follows:

[0003] Perception layer: cameras, lidar, millimeter-wave radars, ultrasonic sensors, etc.

[0004] Decision layer: The data from different sensors are fused to obtain more comprehensive and accurate environmental information; deep learning algorithms, such as convolutional neural networks (CNNs), are used to process camera images and lidar point cloud data to identify various target objects and determine their position, category, attitude and other information. According to the perceived environmental information and the target location, a safe and efficient driving path is planned for the vehicle.

[0005] Control layer: By controlling the throttle opening of the engine or the output power of the motor, the vehicle is accelerated, decelerated and driven at a constant speed; by controlling the steering system, the steering angle of the vehicle is precisely controlled to make the vehicle drive along the planned path; by controlling the braking system of the vehicle, the vehicle is decelerated and stopped.

[0006] However, the above intelligent driving solutions have the following problems:

[0007] 1) Limited perception ability: Single-vehicle intelligence mainly relies on on-vehicle sensors such as lidar, cameras, millimeter-wave radars, etc. for environmental perception, but the detection range, accuracy and reliability of these sensors have certain limitations. In bad weather (such as rain, snow, fog) or complex scenarios (such as blocked vision, multi-vehicle occlusion), the perception ability will be significantly reduced, resulting in incomplete or inaccurate cognition of the surrounding environment.

[0008] 2) High computing resource requirements: In order to process a large amount of sensor data in real time and perform complex decision-making calculations, single-vehicle intelligence requires a high-performance on-vehicle computing platform, which not only increases the hardware cost, but also brings problems such as power consumption and heat dissipation.

[0009] 3) Insufficient decision optimization: Single-vehicle intelligence mainly makes decisions based on its own perception and algorithms, and it is difficult to obtain global traffic information. Therefore, there are limitations in aspects such as traffic flow optimization and intersection passing efficiency improvement, which may lead to unnecessary stops, waits or unreasonable route planning.

[0010] 4) High cost: To achieve a high level of autonomous driving capabilities, vehicle intelligence requires a variety of high-precision sensors and high-performance computing platforms, which significantly increases the manufacturing cost of the vehicle.

[0011] The above problems need to be solved urgently. Summary of the Invention

[0012] An object of the present invention is to solve at least to a certain extent one of the technical problems existing in the prior art.

[0013] To this end, an object of an embodiment of the present invention is to provide a vehicle-road collaborative intelligent driving method based on computing offloading. This method offloads the computing workload of the vehicle end to the roadside for execution based on the current device status of the roadside device and the task priority of the intelligent driving computing task. On the one hand, it reduces the computing power pressure on the vehicle end, reduces the use of high-computing-power chips on the vehicle end, and reduces the cost per vehicle. On the other hand, more powerful computing resources can be deployed on the roadside, and more complete perception information can be obtained, so that the calculation of intelligent driving-related tasks can be completed more quickly and accurately, improving the reliability and safety of intelligent driving.

[0014] Another object of an embodiment of the present invention is to provide a vehicle-road collaborative intelligent driving system based on computing offloading.

[0015] To achieve the above technical objectives, the technical solutions adopted in the embodiments of the present invention include:

[0016] In the first aspect, an embodiment of the present invention provides a vehicle-road collaborative intelligent driving method based on computing offloading, including the following steps:

[0017] Obtain multiple intelligent driving computing tasks of the target vehicle, and determine the real-time requirements, safety impact degree, and computing complexity of each of the intelligent driving computing tasks;

[0018] Determine the task priority of each of the intelligent driving computing tasks according to the real-time requirements, the safety impact degree, and the computing complexity;

[0019] Obtain the current device status of each roadside device in the area where the target vehicle is located, make a task allocation decision for each of the intelligent driving computing tasks according to the current device status and the task priority, and obtain the target roadside device corresponding to each of the intelligent driving computing tasks;

[0020] Offload each of the intelligent driving computing tasks to the corresponding target roadside device for calculation, and feed back the calculation result of the target roadside device to the target vehicle.

[0021] Further, in an embodiment of the present invention, obtaining multiple intelligent driving computing tasks of the target vehicle and determining the real-time requirements, safety impact degree, and computing complexity of each of the intelligent driving computing tasks specifically include:

[0022] Obtain a computing task list through the intelligent driving system of the target vehicle, and extract multiple intelligent driving computing tasks according to the computing task list;

[0023] Determine the task type, application scenario, algorithm type, and required computing resources of each of the intelligent driving computing tasks;

[0024] Determine the real-time requirements of each of the intelligent driving computing tasks according to the application scenario;

[0025] Determine the safety impact degree of each of the intelligent driving computing tasks according to the task type;

[0026] Determine the computing complexity of each of the intelligent driving computing tasks according to the algorithm type and the required computing resources.

[0027] Further, in an embodiment of the present invention, determining the task priority of each of the intelligent driving computing tasks according to the real-time requirements, the safety impact degree, and the computing complexity specifically includes:

[0028] Perform weighted summation on the real-time requirements, the safety impact degree, and the computing complexity according to a preset weight factor to obtain a priority evaluation score for each of the intelligent driving computing tasks;

[0029] Determine the task priority of each of the intelligent driving computing tasks according to the priority evaluation score.

[0030] Further, in an embodiment of the present invention, obtaining the current device status of each target roadside device in the area where the target vehicle is located, and making a task allocation decision for each of the intelligent driving computing tasks according to the current device status and the task priority to obtain the target roadside device corresponding to each of the intelligent driving computing tasks specifically includes:

[0031] Determine multiple roadside devices according to the real-time position of the target vehicle and a preset signal coverage range;

[0032] Obtain the current device status of each roadside device, where the current device status includes current idle computing resources and current communication delay;

[0033] According to the current device status and the task priority, use the particle swarm algorithm to make a task allocation decision for each of the intelligent driving computing tasks to obtain an optimal task allocation scheme;

[0034] Determine the target roadside device corresponding to each of the intelligent driving computing tasks according to the optimal task allocation scheme.

[0035] Further, in an embodiment of the present invention, the task allocation decision for each of the intelligent driving computing tasks is performed by using a particle swarm optimization algorithm according to the current device state and the task priority to obtain an optimal task allocation scheme, which specifically includes:

[0036] Randomly generate a plurality of initial task allocation matrices according to the intelligent driving computing tasks and the target roadside devices, and initialize the particle swarm according to the initial task allocation matrices, where the search space dimension of the particle swarm is the same as the number of the intelligent driving computing tasks, the coordinate values of the particle in each dimension respectively represent the serial number values of the roadside devices to which each intelligent driving computing task is allocated, and the velocity of the particle represents the adjustment direction of the task allocation;

[0037] Calculate the fitness value of each particle according to a preset fitness function;

[0038] Update the individual optimal position and the global optimal position of each particle according to the fitness value;

[0039] Adjust the velocity and position of each particle according to the individual optimal position and the global optimal position, and return to the step of calculating the fitness value of each particle according to the preset fitness function until a preset convergence condition is reached, and determine the optimal task allocation scheme according to the global optimal position.

[0040] Further, in an embodiment of the present invention, the fitness function is:

[0041]

[0042] where F represents the fitness value, W i represents the current idle computing resource of the roadside device to which the intelligent driving computing task i is allocated, L i represents the current communication delay of the roadside device to which the intelligent driving computing task i is allocated, P i represents the task priority of the intelligent driving computing task i, n represents the number of the intelligent driving computing tasks, and α, β, and γ represent preset weight parameters.

[0043] Further, in an embodiment of the present invention, the vehicle-road collaborative intelligent driving method further includes the following steps:

[0044] When it is monitored that the device load rate of the target roadside device reaches a preset first threshold, migrate the intelligent driving computing task corresponding to the target roadside device to an adjacent roadside device for calculation.

[0045] In a second aspect, an embodiment of the present invention provides a vehicle-road collaborative intelligent driving system based on computing offloading, including:

[0046] A computing task acquisition module, configured to acquire multiple intelligent driving computing tasks of a target vehicle, and determine the real-time requirement, safety impact degree, and computing complexity of each of the intelligent driving computing tasks;

[0047] A task priority determination module, configured to determine the task priority of each of the intelligent driving computing tasks according to the real-time requirement, the safety impact degree, and the computing complexity;

[0048] A task allocation decision module, configured to acquire the current device status of each roadside device in the area where the target vehicle is located, and make a task allocation decision on each of the intelligent driving computing tasks according to the current device status and the task priority, so as to obtain the target roadside device corresponding to each of the intelligent driving computing tasks;

[0049] A computing task offloading module, configured to offload each of the intelligent driving computing tasks to the corresponding target roadside device for computing, and feed back the computing result of the target roadside device to the target vehicle.

[0050] In a third aspect, an embodiment of the present invention provides a vehicle-road collaborative intelligent driving device based on computing offloading, including:

[0051] At least one processor;

[0052] At least one memory, configured to store at least one program;

[0053] When the at least one program is executed by the at least one processor, the at least one processor is caused to implement the above-mentioned vehicle-road collaborative intelligent driving method based on computing offloading.

[0054] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, in which a program executable by a processor is stored, and the program executable by the processor is used to execute the above-mentioned vehicle-road collaborative intelligent driving method based on computing offloading when executed by the processor.

[0055] The advantages and beneficial effects of the present invention will be partially given in the following description, partially become obvious from the following description, or be understood through the practice of the present invention:

[0056] In an embodiment of the present invention, multiple intelligent driving computing tasks of a target vehicle are obtained, and the real-time requirements, safety impact degrees, and computing complexities of each intelligent driving computing task are determined. The task priorities of each intelligent driving computing task are determined according to the real-time requirements, safety impact degrees, and computing complexities. The current device states of each roadside device in the area where the target vehicle is located are obtained. Task allocation decisions are made for each intelligent driving computing task according to the current device states and task priorities, and the target roadside devices corresponding to each intelligent driving computing task are obtained. Each intelligent driving computing task is offloaded to the corresponding target roadside device for computing, and the computing results of the target roadside device are fed back to the target vehicle. In the embodiment of the present invention, based on the current device state of the roadside device and the task priority of the intelligent driving computing task, the computing workload at the vehicle end is offloaded to the roadside for execution. On the one hand, it reduces the computing power pressure at the vehicle end, reduces the use of high-computing-power chips at the vehicle end, and reduces the single-vehicle cost. On the other hand, more powerful computing resources can be deployed on the roadside, and more complete perception information can be obtained, so that the computing of intelligent driving-related tasks can be completed more quickly and accurately, improving the reliability and safety of intelligent driving. Description of the Drawings

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following introduces the drawings required to be used in the embodiments of the present invention. It should be understood that the drawings introduced below only conveniently and clearly illustrate some embodiments of the technical solutions in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative efforts.

[0058] Figure 1 It is a flowchart of the steps of a vehicle-road collaborative intelligent driving method based on computing offloading provided by an embodiment of the present invention;

[0059] Figure 2 It is a structural block diagram of a vehicle-road collaborative intelligent driving system based on computing offloading provided by an embodiment of the present invention;

[0060] Figure 3 It is a structural block diagram of a vehicle-road collaborative intelligent driving device based on computing offloading provided by an embodiment of the present invention. Detailed Embodiments

[0061] Embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as a limitation of the present invention. For the step numbers in the following embodiments, they are only set for the convenience of elaboration and explanation, and no limitation is imposed on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0062] In the description of the present invention, the meaning of "a plurality" is two or more. If the first and second are described, it is only for the purpose of distinguishing technical features and should not be construed as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features or implicitly indicating the sequence relationship of the indicated technical features. In addition, unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art of this technical field.

[0063] Referring to Figure 1 , an embodiment of the present invention provides a vehicle-road collaborative intelligent driving method based on computing offloading, which specifically includes the following steps:

[0064] S101. Obtain multiple intelligent driving computing tasks of the target vehicle, and determine the real-time requirements, safety impact degree, and computing complexity of each intelligent driving computing task;

[0065] S102. Determine the task priority of each intelligent driving computing task according to the real-time requirements, safety impact degree, and computing complexity;

[0066] S103. Obtain the current device status of each roadside device in the area where the target vehicle is located, and make a task allocation decision for each intelligent driving computing task according to the current device status and task priority to obtain the target roadside device corresponding to each intelligent driving computing task;

[0067] S104. Offload each intelligent driving computing task to the corresponding target roadside device for calculation, and feedback the calculation result of the target roadside device to the target vehicle.

[0068] Specifically, the embodiment of the present invention is implemented based on a vehicle-road collaborative architecture, including roadside devices, on-vehicle devices, and cloud systems. By making a computing offloading decision to judge the main computing unit of the operation, the computing power is reasonably allocated to ensure that the calculation is completed and the result is feedback in the shortest time, so as to ensure driving safety.

[0069] The roadside devices include intelligent traffic lights, cameras, millimeter-wave radars, lidars, etc. These devices are responsible for collecting information such as road conditions, traffic flow, and weather conditions. In addition, the roadside computing unit integrates a high-computing-power computing unit and uses computing offloading technology to share the computing power pressure of the vehicle end.

[0070] The in-vehicle devices are installed on the vehicle and cover sensing sensors (such as cameras, ultrasonic sensors), communication devices, and control systems. It can not only sense and send the state of the vehicle itself, such as vehicle speed, steering angle, etc., but also receive the information sent by the roadside devices.

[0071] The cloud system has powerful data storage, computing, and decision-making functions, including a decision-making platform and a command center. It can integrate the data uploaded by the roadside devices and in-vehicle devices for in-depth analysis and processing.

[0072] In the embodiments of the present invention, the computing tasks of intelligent driving are carefully classified, such as image recognition and target detection in environmental perception tasks, global path planning and local path planning in path planning tasks, acceleration, deceleration, and steering decisions in decision control tasks, etc. The priorities of each task are determined according to the real-time requirements, safety impact degree, and computing complexity of the tasks. Tasks with high real-time requirements, large safety impact degree, and large computing complexity are given high priorities and are preferentially offloaded to the roadside devices with rich computing resources and fast response speeds for execution. The cloud system provides computing power support for each edge computing node (roadside device) based on cloud-edge collaboration to ensure that the computing is completed and the results are fed back in the shortest time, guaranteeing driving safety.

[0073] In the embodiments of the present invention, the computing amount of the vehicle end is offloaded to the roadside for execution based on the current device state of the roadside device and the task priorities of the intelligent driving computing tasks. On the one hand, it reduces the computing power pressure of the vehicle end, reduces the use of high-computing-power chips at the vehicle end, and reduces the cost of a single vehicle. On the other hand, more powerful computing resources can be deployed on the roadside, and more complete sensing information can be obtained, so that the computing of intelligent driving-related tasks can be completed more quickly and accurately, improving the reliability and safety of intelligent driving.

[0074] Further as an optional implementation manner, multiple intelligent driving computing tasks of the target vehicle are obtained, and the real-time requirements, safety impact degree, and computing complexity of each intelligent driving computing task are determined, which specifically include:

[0075] S1011. Obtain a computing task list through the intelligent driving system of the target vehicle, and extract multiple intelligent driving computing tasks according to the computing task list;

[0076] S1012. Determine the task type, application scenario, algorithm type, and required computing resources of each intelligent driving computing task;

[0077] S1013. Determine the real-time requirements for each intelligent driving computing task according to the application scenario;

[0078] S1014. Determine the safety impact degree of each intelligent driving computing task according to the task type;

[0079] S1015. Determine the computing complexity of each intelligent driving computing task according to the algorithm type and the required computing resources.

[0080] Specifically, extract multiple computing tasks from the intelligent driving system of the target vehicle. The task list can be obtained through the communication interfaces inside the vehicle, such as the CAN bus, Ethernet, etc. For example, the environmental perception tasks (including camera image recognition, radar data processing), path planning tasks, decision-making and control tasks, etc. in the intelligent driving system. Analyze the nature and application scenario of the tasks to determine the real-time requirements. For the environmental perception tasks applied to the emergency obstacle avoidance scenario, since a response needs to be made within an extremely short time, its real-time requirements are very high, usually requiring the calculation to be completed within dozens of milliseconds. For some tasks with low real-time requirements, such as the path planning task applied to in-vehicle map update, it can be completed within a relatively long time, and the real-time requirements are low. Evaluate the impact degree of the tasks on the vehicle driving safety according to the task type. Among them, the environmental perception and decision-making and control tasks are directly related to the vehicle driving safety. Once a calculation error or delay occurs, it may lead to serious safety accidents, so the safety impact degree is high. While some auxiliary tasks, such as the computing tasks of the in-vehicle entertainment system, have a relatively small impact on the vehicle driving safety, and the safety impact degree is low. Determine the computing complexity according to the computing resources required by the tasks and the algorithm type. For example, complex deep learning algorithms are used for image recognition, which require a large amount of computing resources and a long computing time, and the computing complexity is high. While some simple logical judgment tasks have a relatively low computing complexity.

[0081] Further as an optional implementation manner, determine the task priorities of each intelligent driving computing task according to the real-time requirements, safety impact degree, and computing complexity. Specifically, it includes:

[0082] S1021. Perform weighted summation on the real-time requirements, safety impact degree, and computing complexity according to the preset weight factors to obtain the priority evaluation scores of each intelligent driving computing task;

[0083] S1022. Determine the task priorities of each intelligent driving computing task according to the priority evaluation scores.

[0084] Specifically, considering three factors: real-time requirement, safety impact degree, and computational complexity, a weighted evaluation model is established. For example, the weight of the real-time requirement can be set to 0.5, the weight of the safety impact degree to 0.3, and the weight of the computational complexity to 0.2. For each task, according to the corresponding values of the real-time requirement, safety impact degree, and computational complexity, combined with the weighted evaluation model, calculate the task priority evaluation score. The higher the priority evaluation score of a task, the higher the task priority and the more it needs to be processed preferentially.

[0085] Further as an optional implementation manner, obtain the current device status of each target roadside device in the area where the target vehicle is located, and make a task allocation decision for each intelligent driving computing task according to the current device status and task priority, to obtain the target roadside device corresponding to each intelligent driving computing task, which specifically includes:

[0086] S1031. Determine multiple roadside devices according to the real-time position of the target vehicle and the preset signal coverage range;

[0087] S1032. Obtain the current device status of each roadside device, where the current device status includes the current idle computing resource and the current communication delay;

[0088] S1033. According to the current device status and task priority, use the particle swarm algorithm to make a task allocation decision for each intelligent driving computing task to obtain the optimal task allocation scheme;

[0089] S1034. Determine the target roadside device corresponding to each intelligent driving computing task according to the optimal task allocation scheme.

[0090] Specifically, obtain the real-time coordinates of the target vehicle through in-vehicle GPS, roadside radar (such as millimeter-wave radar), or V2X communication, call the preset electronic map data, and combine the deployment topology of roadside devices (RSU, MEC) to filter out all devices covering the area where the vehicle is located. Considering environmental interference factors (such as building occlusion, weather impact), the effective coverage range can be dynamically adjusted through historical communication quality data or real-time channel measurement (such as the feedback of the LTE-V2X module of the RSU).

[0091] Obtain information such as CPU / GPU utilization rate, remaining memory, and task queue length through the API interface of the roadside device, and calculate the current idle computing resource of the roadside device; use Ping test or timestamp synchronization (such as IEEE 802.1AS protocol) to measure the end-to-end delay from the device to the vehicle end to obtain the current communication delay.

[0092] Based on the current device status and task priorities, different task allocation schemes are used as particles in the particle swarm, and the corresponding fitness function and constraint conditions are set. Task allocation decisions are made based on the particle swarm algorithm to obtain the optimal task allocation scheme. For example, for the highest-priority environmental perception task, it is preferentially allocated to the roadside device that is closest to the target vehicle and has the most idle computing resources.

[0093] According to the task allocation decision results, each intelligent driving computing task is offloaded to the corresponding roadside device through a wireless communication link. During the offloading process, it is necessary to ensure the accurate transmission and integrity of the task data.

[0094] After the roadside device completes the computing task, the computing result is fed back to the target vehicle through a wireless communication link. After receiving the computing result, the target vehicle applies it to the intelligent driving system to achieve the corresponding functions. For example, the computing result of the environmental perception task can be used for the vehicle's obstacle avoidance decision-making and path planning.

[0095] Further as an optional implementation manner, based on the current device status and task priorities, the particle swarm algorithm is used to make task allocation decisions for each intelligent driving computing task to obtain the optimal task allocation scheme, which specifically includes:

[0096] S10331. According to the intelligent driving computing tasks and target roadside devices, randomly generate multiple initial task allocation matrices, and initialize the particle swarm according to the initial task allocation matrices. Among them, the search space dimension of the particle swarm is the same as the number of intelligent driving computing tasks, and the coordinate values of the particle in each dimension respectively represent the serial number values of the roadside devices to which each intelligent driving computing task is allocated, and the velocity of the particle represents the adjustment direction of the task allocation;

[0097] S10332. Calculate the fitness values of each particle according to the preset fitness function;

[0098] S10333. Update the individual optimal position and global optimal position of each particle according to the fitness value;

[0099] S10334. Adjust the velocity and position of each particle according to the individual optimal position and global optimal position, and return to the step of calculating the fitness values of each particle according to the preset fitness function until the preset convergence condition is reached, and determine the optimal task allocation scheme according to the global optimal position.

[0100] Further as an optional implementation manner, the fitness function is:

[0101]

[0102] Where F represents the fitness value, W iDenote the current idle computing resources of the roadside device to which the intelligent driving computing task i is assigned, L i Denote the current communication delay of the roadside device to which the intelligent driving computing task i is assigned, P i Denote the task priority of the intelligent driving computing task i. n represents the number of intelligent driving computing tasks, and α, β, and γ represent preset weight parameters.

[0103] Specifically, the process of making task allocation decisions based on the particle swarm algorithm is as follows:

[0104] 1. Generate an initial task allocation matrix and initialize the particle swarm

[0105] Randomly generate an initial task allocation matrix: First, clarify the set of intelligent driving computing tasks T = {t1, t2, …, t n} and the set of target roadside devices R = {r1, r2, …, r m}. Using a random number generation algorithm, for each intelligent driving computing task t i (i = 1, 2, …, n), randomly select a roadside device from the set of target roadside devices R as its initial allocation object, and repeat this operation k times to obtain k initial task allocation matrices. The size of each matrix is n×m, and the element a ij in the matrix represents whether the task t i is allocated to the roadside device r j (a ij = 1 indicates allocation, and a ij = 0 indicates non - allocation).

[0106] Initialize the particle swarm: The particle swarm consists of k particles, each particle representing a task allocation scheme. The search space dimension of the particle swarm is set to the number n of intelligent driving computing tasks. For each particle p, the coordinate value x pi in each dimension i (i = 1, 2, …, n) represents the serial number value of the roadside device to which the task t i is allocated, and this serial number value is between 1 and m. At the same time, assign an initial value to the velocity v pi of each particle. The velocity represents the adjustment direction of task allocation, and its initial value can be randomly generated within a preset small range.

[0107] 2. Calculate the fitness value of each particle

[0108] Define the fitness function: The fitness function F is used to evaluate the quality of the task allocation scheme represented by each particle. The design of the fitness function needs to comprehensively consider multiple factors, such as the current idle computing resources, communication delay, and task priority of the roadside device, etc. In the embodiment of the present invention, the fitness function is defined as:

[0109]

[0110] Among them, F represents the fitness value, and W i represents the current idle computing resource of the roadside device to which the intelligent driving computing task i is assigned, and L i represents the current communication delay of the roadside device to which the intelligent driving computing task i is assigned, and P i represents the task priority of the intelligent driving computing task i, n represents the number of intelligent driving computing tasks, and α, β, and γ represent preset weight parameters.

[0111] Calculating the fitness value: For each particle p in the particle swarm, according to the task allocation scheme represented by its coordinate value, calculate the value of the fitness function F. It can be seen that when the current idle computing resource of the roadside device to which a certain task i is assigned is more and the communication delay is lower, its contribution value αW i -βL i -γP i to the fitness value F is larger, and the fitness value F is the sum of the contribution values αW i -βL i -γP i of each task. Therefore, for tasks with a larger task priority, it is necessary to allocate roadside devices with as much current idle computing resource and as low communication delay as possible to maximize the fitness value.

[0112] 3. Updating the individual optimal position and the global optimal position

[0113] Updating the individual optimal position: For each particle p, compare its current fitness value with the previously recorded individual optimal fitness value. If the current fitness value is greater than the previously recorded individual optimal fitness value, update the individual optimal position to the position of the current particle and update the individual optimal fitness value to the current fitness value.

[0114] Updating the global optimal position: Among the individual optimal positions of all particles, find the particle with the largest fitness value, take its position as the global optimal position, and record its fitness value.

[0115] 4. Adjusting the speed and position of the particles

[0116] Adjusting the particle speed: According to the individual optimal position p best and the global optimal position g best of the particle, use the following formula to adjust the speed of the particle:

[0117] v pi (t + 1) = w × v pi (t) + c1 × r1 × (p best,i - x pi (t)) + c2 × r2 × (gbest,i -x pi (t))

[0118] where v pi (t) is the velocity of particle p in the i-th dimension (corresponding to the i-th task) at the t-th iteration, w is the inertia weight, c1 and c2 are learning factors, r1 and r2 are random numbers in the interval [0, 1], p best,i and g best,i are the values of the individual optimal position and the global optimal position in the i-th dimension respectively, and x pi (t) is the coordinate value of particle p in the i-th dimension at the t-th iteration.

[0119] Adjust the particle position: Update the position x of the particle according to the adjusted velocity pi (t + 1) = x pi (t) + v pi (t + 1). It should be noted that the updated position may exceed the range [1, m] of the roadside device numbers. In this case, boundary processing is required to limit it within the range [1, m].

[0120] It should be noted that when adjusting the particle velocity and position, certain constraint conditions also need to be set. For example, it is specified that the idle computing resources of each roadside device after being assigned tasks shall not be lower than a threshold, that is, (current idle computing resources - total computing resource requirements of the assigned intelligent driving computing tasks) is greater than the second threshold, so as to ensure that each roadside device will not be overloaded.

[0121] 5. Judge the convergence condition and determine the optimal task allocation scheme

[0122] Judge the convergence condition: Repeat steps 2 to 4 until the preset convergence condition is met. The convergence condition can be reaching the maximum number of iterations, the change in the fitness value being less than a certain threshold, etc.

[0123] Determine the optimal task allocation scheme: When the convergence condition is reached, the task allocation scheme represented by the global optimal position g best is the optimal task allocation scheme. According to the coordinate values of the global optimal position, it is possible to determine the roadside device to which each intelligent driving computing task is assigned.

[0124] Furthermore, as an optional implementation manner, the vehicle-road collaborative intelligent driving method further includes the following steps:

[0125] When it is monitored that the device load rate of the target roadside device reaches the preset first threshold, migrate the intelligent driving computing task corresponding to the target roadside device to an adjacent roadside device for calculation.

[0126] Specifically, the embodiments of the present invention adopt a dynamic load balancing strategy, that is, when the load of a single device exceeds 80%, the computing migration between adjacent devices is triggered to improve the efficiency of intelligent driving task computing.

[0127] The above has described the method steps of the embodiments of the present invention. It can be understood that the embodiments of the present invention unload the computing workload of the vehicle end to the roadside for execution based on the current device state of the roadside device and the task priority of the intelligent driving computing task. On the one hand, it reduces the computing power pressure on the vehicle end, reduces the use of high-computing-power chips on the vehicle end, and reduces the cost of a single vehicle. On the other hand, more powerful computing resources can be deployed on the roadside, and more complete perception information can be obtained, so that the computing of intelligent driving-related tasks can be completed more quickly and accurately, improving the reliability and safety of intelligent driving.

[0128] Compared with the prior art, the embodiments of the present invention also have the following advantages:

[0129] 1) In terms of computing power

[0130] Reduce the requirements for in-vehicle hardware: Traditional intelligent driving requires a high-performance computing platform in the vehicle to process a large amount of data, with high costs, high power consumption, and difficult heat dissipation. However, this technology can unload some computing tasks to the roadside or cloud server, and only devices that can meet the basic computing and communication requirements need to be configured in the vehicle, reducing the in-vehicle hardware cost and design difficulty.

[0131] Improve computing efficiency and real-time performance: The computing resources of the roadside and cloud server are powerful and can process the computing tasks of multiple vehicles in parallel, without being restricted by the vehicle space and energy. In complex scenarios, data processing and analysis can be quickly completed, providing timely and accurate decision-making instructions for the vehicle.

[0132] 2) In terms of perception ability

[0133] Expand the perception range: Traditional intelligent driving mainly relies on in-vehicle sensors, and the perception range is limited to a certain area around the vehicle. Vehicle-road collaborative intelligent driving can obtain environmental information at a farther distance and wider angle through roadside sensors, such as cameras and radars installed at intersections, bridges, tunnels, etc., and can discover potential dangers in advance and notify the vehicle.

[0134] Enhance the reliability of perception: In the vehicle-road collaborative system, the multi-sensors of the vehicle and the roadside complement and verify each other. When the performance of in-vehicle sensors decreases due to factors such as bad weather, the roadside sensors can continue to provide accurate information, improving the overall reliability and stability of the system's perception, and reducing misjudgment and missed judgment situations.

[0135] 3) In terms of cost and scalability

[0136] Reduce overall costs: Although building roadside infrastructure requires certain investment, from the perspective of the entire transportation system, it can reduce the hardware costs of bicycles and make intelligent driving technology more cost-effective. With the improvement of roadside infrastructure and economies of scale, the unit cost is expected to be further reduced.

[0137] Easy to expand and upgrade: The functions can be easily expanded and the performance can be improved by adding roadside devices, upgrading cloud algorithms and software, etc., to adapt to the changing traffic demands and technological developments.

[0138] Refer to Figure 2 , an embodiment of the present invention provides a vehicle-road collaborative intelligent driving system based on computing offloading, including:

[0139] A computing task acquisition module, configured to acquire multiple intelligent driving computing tasks of a target vehicle, and determine the real-time requirements, safety impact degrees, and computing complexities of each intelligent driving computing task;

[0140] A task priority determination module, configured to determine the task priorities of each intelligent driving computing task according to the real-time requirements, safety impact degrees, and computing complexities;

[0141] A task allocation decision module, configured to acquire the current device states of each roadside device in the area where the target vehicle is located, and make a task allocation decision on each intelligent driving computing task according to the current device states and task priorities, to obtain the target roadside device corresponding to each intelligent driving computing task;

[0142] A computing task offloading module, configured to offload each intelligent driving computing task to the corresponding target roadside device for computing, and feed back the computing results of the target roadside device to the target vehicle.

[0143] The content in the above method embodiments is applicable to the system embodiments of the present invention. The functions specifically implemented by the system embodiments of the present invention are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0144] Refer to Figure 3 , an embodiment of the present invention provides a vehicle-road collaborative intelligent driving device based on computing offloading, including:

[0145] At least one processor;

[0146] At least one memory, configured to store at least one program;

[0147] When the above at least one program is executed by the above at least one processor, the above at least one processor implements the above vehicle-road collaborative intelligent driving method based on computing offloading.

[0148] The content in the above method embodiments is applicable to the device embodiments of the present invention. The functions specifically implemented in the device embodiments of the present invention are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those in the above method embodiments.

[0149] An embodiment of the present invention further provides a computer-readable storage medium, in which a program executable by a processor is stored. The program executable by the processor is used to execute the above method for vehicle-road collaborative intelligent driving based on computing offloading when executed by the processor.

[0150] A computer-readable storage medium according to an embodiment of the present invention can execute a method for vehicle-road collaborative intelligent driving based on computing offloading provided by an embodiment of the method of the present invention, can execute any combination of implementation steps of the method embodiment, and has the corresponding functions and beneficial effects of the method.

[0151] An embodiment of the present invention also discloses a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes Figure 1 the method shown.

[0152] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order mentioned in the operation diagrams. For example, depending on the functions / operations involved, two consecutive blocks shown may actually be executed substantially simultaneously or the above blocks can sometimes be executed in the reverse order. In addition, the embodiments presented and described in the flowcharts of the present invention are provided by way of example for the purpose of providing a more comprehensive understanding of the technology. The disclosed method is not limited to the operations and logical flows presented herein. Alternative embodiments are contemplated, in which the order of various operations is changed and the sub-operations described as part of a larger operation are executed independently.

[0153] In addition, although the present invention has been described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the above functions and / or features may be integrated in a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It should also be understood that a detailed discussion of the actual implementation of each module is not necessary for understanding the present invention. Rather, considering the attributes, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the modules will be understood within the ordinary skills of an engineer. Therefore, those skilled in the art can implement the present invention as set forth in the claims without undue experimentation. It should also be understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0154] If the above functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the above methods in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs that can store program codes.

[0155] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a predefined sequence of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in conjunction with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0156] More specific examples (nonexhaustive list) of computer-readable media include the following: an electrical connection (electronic device) having one or more wirings, a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which the above programs can be printed, because the above programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory.

[0157] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), and the like.

[0158] In the above description of this specification, the descriptions referring to the terms "one embodiment / example", "another embodiment / example", or "certain embodiments / examples", etc., mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0159] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the claims and their equivalents.

[0160] The above has specifically described the preferred embodiments of the present invention, but the present invention is not limited to the above embodiments. Those skilled in the art can also make various equivalent deformations or substitutions without departing from the spirit of the present invention, and these equivalent deformations or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A vehicle-road cooperative intelligent driving method based on computation offloading, characterized in that: The following steps are involved: Acquire multiple intelligent driving computing tasks of the target vehicle, and determine the real-time requirements, safety impact, and computing complexity of each of the intelligent driving computing tasks; Determining the task priority of each of the intelligent driving computing tasks according to the real-time requirement, the safety impact and the computing complexity; Acquire the current device status of each roadside device in the area where the target vehicle is located, make a task allocation decision for each intelligent driving computing task according to the current device status and the task priority, and obtain the target roadside device corresponding to each intelligent driving computing task; Each of the intelligent driving calculation tasks is offloaded to the corresponding target roadside equipment for calculation, and the calculation result of the target roadside equipment is fed back to the target vehicle.

2. The vehicle-road cooperative intelligent driving method based on computation offloading according to claim 1, characterized in that: The step of obtaining multiple intelligent driving computing tasks of the target vehicle and determining the real-time requirements, safety impact, and computing complexity of each intelligent driving computing task specifically includes: Obtaining a computing task list through the intelligent driving system of the target vehicle, and extracting a plurality of the intelligent driving computing tasks according to the computing task list; Determine the task type, application scenario, algorithm type and required computing resources of each of the intelligent driving computing tasks; Determining the real-time requirements of each of the intelligent driving computing tasks according to the application scenario; Determining the safety impact of each of the intelligent driving computing tasks according to the task type; The computational complexity of each of the intelligent driving computing tasks is determined according to the algorithm type and the required computing resources.

3. The vehicle-road cooperative intelligent driving method based on computation offloading according to claim 1, characterized in that: The step of determining the task priority of each of the intelligent driving computing tasks according to the real-time requirement, the safety impact, and the computing complexity specifically includes: The real-time requirement, the safety impact and the computational complexity are weighted and summed according to a preset weight factor to obtain a priority evaluation score for each of the intelligent driving computing tasks; The task priority of each of the intelligent driving computing tasks is determined according to the priority evaluation score.

4. The vehicle-road cooperative intelligent driving method based on computation offloading according to claim 1, characterized in that: The step of obtaining the current device status of each target roadside device in the area where the target vehicle is located, making a task allocation decision for each intelligent driving computing task according to the current device status and the task priority, and obtaining the target roadside device corresponding to each intelligent driving computing task specifically includes: Determining a plurality of roadside devices according to the real-time position of the target vehicle and a preset signal coverage range; Acquire the current device status of each roadside device, wherein the current device status includes current idle computing resources and current communication delay; According to the current device status and the task priority, a particle swarm algorithm is used to make task allocation decisions for each of the intelligent driving computing tasks to obtain an optimal task allocation solution; The target roadside equipment corresponding to each of the intelligent driving computing tasks is determined according to the optimal task allocation scheme.

5. The vehicle-road cooperative intelligent driving method based on computation offloading according to claim 4 is characterized in that: The method of using a particle swarm algorithm to make task allocation decisions for each of the intelligent driving computing tasks according to the current device state and the task priority to obtain an optimal task allocation solution specifically includes: According to the intelligent driving computing task and the target roadside equipment, a plurality of initial task allocation matrices are randomly generated, and a particle swarm is initialized according to the initial task allocation matrix, wherein the search space dimension of the particle swarm is the same as the number of the intelligent driving computing tasks, the coordinate values ​​of the particles in each dimension respectively represent the serial number values ​​of the roadside equipment to which each intelligent driving computing task is assigned, and the speed of the particles represents the adjustment direction of the task allocation; Calculate the fitness value of each particle according to the preset fitness function; Update the individual optimal position and the global optimal position of each particle according to the fitness value; The speed and position of each particle are adjusted according to the individual optimal position and the global optimal position, and the step of calculating the fitness value of each particle according to a preset fitness function is returned until a preset convergence condition is reached, and the optimal task allocation scheme is determined according to the global optimal position.

6. The vehicle-road cooperative intelligent driving method based on computation offloading according to claim 5, characterized in that: The fitness function is: Among them, F represents the fitness value, W i represents the current idle computing resources of the roadside equipment to which the intelligent driving computing task i is assigned, L i represents the current communication delay of the roadside equipment to which the intelligent driving computing task i is assigned, P i represents the task priority of intelligent driving computing task i, n represents the number of intelligent driving computing tasks, and α, β and γ represent preset weight parameters.

7. A vehicle-road cooperative intelligent driving method based on computation offloading according to any one of claims 1 to 6, characterized in that: The vehicle-road cooperative intelligent driving method further includes the following steps: When it is monitored that the equipment load rate of the target roadside equipment reaches a preset first threshold, the intelligent driving calculation task corresponding to the target roadside equipment is migrated to an adjacent roadside equipment for calculation.

8. A vehicle-road cooperative intelligent driving system based on computation offloading, characterized in that: include: A computing task acquisition module, used to acquire multiple intelligent driving computing tasks of a target vehicle and determine the real-time requirements, safety impact and computing complexity of each of the intelligent driving computing tasks; A task priority determination module, used to determine the task priority of each of the intelligent driving computing tasks according to the real-time requirement, the safety impact and the computational complexity; A task allocation decision module is used to obtain the current device status of each roadside device in the area where the target vehicle is located, and make a task allocation decision for each intelligent driving computing task according to the current device status and the task priority to obtain the target roadside device corresponding to each intelligent driving computing task; The computing task unloading module is used to unload each of the intelligent driving computing tasks to the corresponding target roadside equipment for computing, and feed back the computing results of the target roadside equipment to the target vehicle.

9. A vehicle-road cooperative intelligent driving device based on computation offloading, characterized in that: include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements a vehicle-road collaborative intelligent driving method based on computational offloading as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a program executable by a processor, characterized in that: The program executable by the processor is used to execute a vehicle-road collaborative intelligent driving method based on computational offloading as described in any one of claims 1 to 7 when executed by the processor.

Images