Vehicle accident emergency response method, electronic equipment and storage medium

Abstract

The invention provides a vehicle accident emergency response method, electronic equipment and a storage medium, and the method comprises the steps: obtaining a target driving path of a first vehicle for a preset cooperative task, and an original driving path of a second vehicle, carrying out the risk propagation of a dynamic risk field, and obtaining a target driving path of the first vehicle; obtaining a first risk exposure degree of the first vehicle in the target driving path and a second risk exposure degree of the second vehicle in the original driving path, obtaining a first collaborative value of the first vehicle according to the vehicle information of the first vehicle, and obtaining a second collaborative value of the second vehicle according to the vehicle information of the second vehicle; and obtaining task allocation information according to the first risk exposure degree, the second risk exposure degree, the first cooperative value and the second cooperative value, and sending a task instruction to the corresponding vehicle, so that the corresponding vehicle executes the target cooperative task. Therefore, when the target vehicle has an accident, the surrounding vehicles can provide emergency cooperative response, and the intelligent and precise requirements of accident emergency response are met.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation technology, and more specifically, to a vehicle accident emergency response method, electronic device, and storage medium. Background Technology

[0002] With the increasing number of motor vehicles, problems such as traffic congestion, frequent accidents, and low traffic efficiency are becoming increasingly prominent. Therefore, it is particularly important to improve the emergency response to vehicle accidents and ensure driving safety and optimize the travel experience.

[0003] In related technologies, emergency response to accidents typically relies on a linear process of sensing, transmission, and notification. In this process, the vehicle uses a sensing layer to collect vehicle data, and when an accident is determined based on the vehicle data, the transmission layer transmits the vehicle's location to the dispatch center. The dispatch center then uses a notification layer to send an accident notification to the rescue center so that rescue can be obtained in a timely manner.

[0004] However, the aforementioned response mechanisms lack coordination and are insufficient to meet the needs for intelligent and precise emergency response to vehicle accidents. Summary of the Invention

[0005] In view of this, embodiments of this application provide a vehicle accident emergency response method, electronic device, and storage medium to address the problem that existing response mechanisms lack coordination and are unable to meet the needs of intelligent and precise vehicle accident emergency response.

[0006] In a first aspect, embodiments of this application provide a vehicle accident emergency response method applied to a cloud device, the method comprising: Initialize the dynamic risk field of the target vehicle where a vehicle accident has occurred; The first vehicle and the second vehicle are determined from multiple vehicles within a preset range of the target vehicle, and the target driving path of the first vehicle for the preset collaborative task and the original driving path of the second vehicle are obtained. Risk propagation is performed on the dynamic risk field to obtain the first risk exposure of the first vehicle under the target driving path and the second risk exposure of the second vehicle under the original driving path. The risk exposure is used to indicate the risk of the vehicle being affected by the vehicle accident when the vehicle is traveling along the corresponding driving path. Based on the vehicle information of the first vehicle, a first collaborative value of the first vehicle is obtained, and based on the vehicle information of the second vehicle, a second collaborative value of the second vehicle is obtained. The collaborative value is used to indicate the emergency response collaborative capability of the corresponding vehicle. Based on the first risk exposure, the second risk exposure, the first collaboration value, and the second collaboration value, iterative optimization is performed to obtain task allocation information, which is used to indicate the allocation of the preset collaboration task to the corresponding vehicle among the multiple vehicles. Send a task instruction to the corresponding vehicle so that the corresponding vehicle performs the preset collaborative task.

[0007] In an optional implementation, the step of performing risk propagation on the dynamic risk field to obtain a first risk exposure of the first vehicle under the target driving path and a second risk exposure of the second vehicle under the original driving path includes: Based on the diffusion coefficient corresponding to the road network where the target vehicle is located and the preset attenuation coefficient, the dynamic risk field is spatially diffused and temporally attenuated to obtain a risk field grid map. Each grid in the risk field grid map stores the risk intensity of the corresponding location at each time. Based on the risk field grid map, obtain the first risk exposure and the second risk exposure.

[0008] In an optional implementation, obtaining the first risk exposure and the second risk exposure based on the risk field grid map includes: Based on the risk field grid map, risk is accumulated along the target driving path to obtain the first risk exposure. Based on the risk field grid map, risk is accumulated along the original driving path to obtain the second risk exposure.

[0009] In an optional implementation, the step of iteratively optimizing based on the first risk exposure, the second risk exposure, the first collaboration value, and the second collaboration value to obtain task allocation information includes: Construct a multi-objective optimization model, which includes: a total risk optimization model and a total value optimization model; Based on the first risk exposure, the second risk exposure, the first collaborative value, and the second collaborative value, the multi-objective optimization model is solved iteratively to obtain the task allocation information.

[0010] In an optional implementation, the step of performing multi-objective iterative solution on the multi-objective optimization model based on the first risk exposure, the second risk exposure, the first collaborative value, and the second collaborative value to obtain the task allocation information includes: Based on the first risk exposure, the second risk exposure, the first risk indicator factor, the second risk indicator factor, the first collaborative value, the second collaborative value, the first decision variable, and the second decision variable, the multi-objective optimization model is solved iteratively to obtain the task allocation information. The risk indicator factor is used to indicate whether the corresponding driving path passes through the preset risk area, and the decision variable is used to indicate whether the vehicle is assigned the preset collaborative task.

[0011] In an optional implementation, obtaining the first collaborative value of the first vehicle based on the vehicle information of the first vehicle, and obtaining the second collaborative value of the second vehicle based on the vehicle information of the second vehicle, includes: The first collaborative value is obtained based on the rescue equipment information and personnel skill information of the first vehicle, the estimated arrival time for the preset collaborative task, and the urgency of the transportation task. The second collaborative value is obtained based on the rescue equipment information, personnel skill information, and urgency of the transportation mission of the second vehicle.

[0012] In an optional implementation, the method further includes: Acquire the motion data and video data of the target vehicle; Based on the motion data, obtain the temporal motion characteristics that meet the preset abnormal conditions; Based on the abnormal time points of the temporal motion characteristics, extract the target video segment from the video data; Based on the temporal motion characteristics and the target video segment, a preset Bayesian network is used to determine the probability of an accident. If the probability of the accident exceeds a preset threshold, then it is determined that the target vehicle has been involved in the accident.

[0013] In an optional implementation, the method further includes: A backbone feature extraction network is used to extract features from the target video segment to obtain target image features; A multi-task sub-network is used to identify the confidence level of the vehicle accident, the confidence level of the accident risk, and the status of the personnel based on the features of the target image. Emergency suggestions are matched from a preset rule base based on the confidence level of the vehicle accident and the confidence level of the accident risk. Based on the vehicle accident, the accident risk, the personnel status, and the emergency recommendations, structured emergency data is generated; The structured emergency data is sent to the corresponding vehicle.

[0014] Secondly, embodiments of this application also provide an electronic device, including: a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the method described in any of the first aspects.

[0015] Thirdly, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the method described in any of the first aspects.

[0016] This application provides a vehicle accident emergency response method, electronic device, and storage medium. The method includes: acquiring the target driving path of a first vehicle for a preset collaborative task and the original driving path of a second vehicle; performing risk propagation on a dynamic risk field to obtain a first risk exposure of the first vehicle on the target driving path and a second risk exposure of the second vehicle on the original driving path; acquiring a first collaborative value of the first vehicle based on its vehicle information; acquiring a second collaborative value of the second vehicle based on its vehicle information; acquiring task allocation information based on the first risk exposure, second risk exposure, first collaborative value, and second collaborative value; and sending task instructions to the corresponding vehicles to enable them to execute the target collaborative task. Thus, when an accident occurs to the target vehicle, surrounding vehicles can provide emergency collaborative response, meeting the needs for intelligent and precise accident emergency response. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 1 ; Figure 2 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 2 ; Figure 3 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 3 ; Figure 4 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 4 ; Figure 5 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 5 ; Figure 6 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 6 ; Figure 7 This is a schematic diagram of the structure of the vehicle accident emergency response device provided in the embodiments of this application; Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0020] To address the issue that existing vehicle accident response mechanisms lack coordination and fail to meet the demands for intelligent and precise emergency response, this application, based on risk exposure and collaborative value, obtains task allocation information. This information instructs the allocation of pre-defined collaborative tasks to corresponding vehicles among multiple vehicles, enabling these vehicles to execute the pre-defined collaborative tasks. Thus, when a target vehicle is involved in an accident, surrounding vehicles can provide emergency collaborative responses, meeting the requirements for intelligent and precise accident emergency response.

[0021] Figure 1 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 1 In this embodiment, the executing entity can be a cloud device.

[0022] like Figure 1 As shown, the method may include: S101. Initialize the dynamic risk field of the target vehicle where a vehicle accident has occurred.

[0023] The target vehicle is a vehicle involved in a traffic accident, such as a single-vehicle skid, a tire blowout, or a fire.

[0024] Determine the severity level of the vehicle accident, and determine the risk intensity R0 based on the severity level. For example, R0 = 1 for minor accidents, R0 = 5 for general accidents, and R0 = 10 for major accidents. Obtain the time t0 when the vehicle accident occurs and the location of the accident (x0, y0) of the target vehicle.

[0025] Initialize and generate the dynamic risk field of the target vehicle, which is represented as: R(x, y, t+Δt) = R(x, y, t) + D· ²R(x, y, t) - λ·R(x, y, t) Where R(x, y, t) is the risk intensity of the vehicle at position (x, y) and time t, R(x, y, t+Δt) is the risk intensity of the vehicle at position (x, y) and time t+Δt, Δt is the time step, and D· ²R(x, y, t) represents the diffusion term, and D is the diffusion coefficient, used to simulate the speed of risk propagation (e.g., a larger D value on highways and a smaller D value on narrow streets). Traffic congestion reduces the diffusion coefficient D because the propagation of risk perception slows down. ² is the Laplace operator (related to the risk intensity of adjacent locations), representing spatial diffusion. λ·R(x, y, t) is the attenuation term, and λ is the attenuation coefficient, indicating that the risk intensity naturally decreases over time (e.g., the attenuation coefficient naturally decreases after the accident is dealt with and traffic resumes).

[0026] Among them, the dynamic risk field is a scalar field defined on the road geographic coordinates (x,y) and time t. Its generation is based on the diffusion-attenuation model, which simulates the risk spreading like ripples and weakening over time. The risk intensity changes with time and space, reflecting the degree of impact of the accident location on the surrounding area.

[0027] It should be noted that the risk intensity R0 is R(x0, y0, t0). At time t0, the accident only occurs at position (x0, y0), so the risk intensity is non-zero only at this position and zero at other positions.

[0028] S102. Determine the first vehicle and the second vehicle from multiple vehicles within a preset range of the target vehicle, and obtain the target driving path of the first vehicle for the preset collaborative task, and the original driving path of the second vehicle.

[0029] The target vehicle's preset range can be a range centered on the target vehicle (e.g., a range of 5 kilometers from the target vehicle). From multiple vehicles within this preset range, a first vehicle and a second vehicle are selected. The first vehicle is initially selected to perform a preset collaborative task, and the second vehicle is initially selected not to perform a preset collaborative task. The number of the first vehicle and the second vehicle may be, but is not limited to, one. There may be multiple preset collaborative tasks, with one task assigned to each vehicle.

[0030] Pre-set collaborative tasks may include upstream early warning tasks, rescue tasks, and video assistance tasks. The upstream early warning task refers to turning on high-brightness warning lights to slow down or divert upstream vehicles in advance, reducing traffic density and speed in the accident area, thereby slowing the spread of risk downstream along the road and avoiding secondary accidents. The rescue task refers to participating in rescue operations. For example, if a vehicle is carrying sand, it can directly participate in the rescue of a target vehicle with a fuel leak, using sand to absorb the leaked fuel. The video assistance task refers to collecting video from another perspective of the accident location and uploading it to the rescue center to assist in the rescue.

[0031] After a vehicle is assigned a preset collaborative task, the path that the vehicle needs to travel to perform the preset collaborative task can be obtained as the target driving path. For example, if a vehicle assigned a video assistance task needs to change course to a specific location (such as the opposite lane of the target vehicle's lane), then the target driving path is the path generated by the specific location and the vehicle's current location. As another example, if a vehicle assigned a rescue task needs to change course to the location of an accident, then the target driving path is the path generated by the location of the accident and the vehicle's current location.

[0032] In some embodiments, the target driving path and the original driving path can be predicted and determined based on the corresponding vehicle's driving speed, direction, road network topology, and driving intention (such as going straight or turning).

[0033] The original driving path of the second vehicle refers to the driving path of the second vehicle itself, which can be determined by the navigation information of the second vehicle.

[0034] S103. Perform risk propagation on the dynamic risk field to obtain the first risk exposure of the first vehicle under the target driving path and the second risk exposure of the second vehicle under the original driving path.

[0035] Risk exposure is used to indicate the risk of a vehicle being affected by a vehicle accident while traveling along a corresponding driving path, representing the total amount of risk that a vehicle may face in the future.

[0036] Risk propagation is performed along the road network where the target vehicle is located to obtain the risk intensity at different locations at different times. Then, the risk intensity of each path point on the target driving path is determined. Based on the risk intensity of each path point on the target driving path, the first risk exposure is obtained, as well as the risk intensity of the locations traversed by the original driving path. Based on the risk intensity of the locations traversed by the original driving path, the second risk exposure of the second vehicle on the original driving path is obtained.

[0037] The first risk exposure is used to indicate the risk of the first vehicle being affected by a vehicle accident while traveling along the target driving path, and the second risk exposure is used to indicate the risk of the second vehicle being affected by a vehicle accident while traveling along the original driving path.

[0038] It should be noted that the risk only spreads within the road network (digitalized road segments) and does not extend to areas outside the roads.

[0039] S104. Based on the vehicle information of the first vehicle, obtain the first collaborative value of the first vehicle, and based on the vehicle information of the second vehicle, obtain the second collaborative value of the second vehicle.

[0040] Among them, the collaborative value is used to indicate the emergency response collaborative capability of the corresponding vehicle. In other words, it is the emergency collaborative potential that the corresponding vehicle can provide to the target vehicle after a vehicle accident occurs.

[0041] The vehicle information for the first vehicle includes rescue equipment information and personnel skill information, and the vehicle information for the second vehicle also includes rescue equipment information and personnel skill information. The rescue equipment information may include, for example, whether the vehicle is carrying a fire extinguisher or a first aid kit, and the personnel skill information may include, for example, the driver's emergency skill rating certified by the platform.

[0042] Based on the vehicle rescue equipment information and personnel skill information of the first vehicle, the collaborative value of the first vehicle is calculated as the first collaborative value. Based on the vehicle rescue equipment information and personnel skill information of the second vehicle, the collaborative value of the second vehicle is calculated as the second collaborative value.

[0043] S105. Iteratively optimize based on the first risk exposure, the second risk exposure, the first collaborative value, and the second collaborative value to obtain task allocation information.

[0044] The task allocation information is used to indicate which vehicle among multiple vehicles will be assigned a preset collaborative task.

[0045] Based on the first and second risk exposures, the total risk is calculated. Based on the first and second collaborative values, the total value is calculated. The first and second vehicles are then re-determined from among multiple vehicles. The total risk and total value are iteratively calculated and repeated. With the goal of minimizing the total risk and maximizing the total value, an optimization algorithm is used to find a balanced solution (such as maximizing the total value while ensuring that the total risk does not exceed a certain threshold). The task allocation information is determined based on this balanced solution. In other words, the total risk and total value are calculated through continuous iteration (adjusting the allocation relationship between vehicles and preset collaborative tasks). An optimization algorithm is used to find the optimal compromise point between the total risk and total value. Finally, task allocation information is generated based on this compromise point.

[0046] Among them, the corresponding vehicle in the multiple vehicles is the vehicle used to perform the preset collaborative task as indicated by the task allocation information.

[0047] S106. Send a task instruction to the corresponding vehicle so that the corresponding vehicle can perform a preset collaborative task.

[0048] The communication link is used to send task instructions to the corresponding vehicles. For example, the instruction for the early warning task is sent to vehicle V1, the instruction for the rescue task is sent to vehicle V2, and the instruction for the video assistance task is sent to vehicle V3.

[0049] After receiving the task instruction, the corresponding vehicle can display the task information on the central control screen. After the driver views the task, clicks to confirm, and then begins to execute the task (such as turning on the warning lights, driving to the accident location, adjusting the camera angle, etc.). In addition, the execution instruction information can also be uploaded to the cloud device to achieve real-time status feedback and form a closed loop.

[0050] In some embodiments, the target vehicle, the first vehicle, and the second vehicle mentioned in this embodiment may be logistics transportation vehicles.

[0051] In this embodiment, task allocation information is obtained based on risk exposure and collaborative value. This information is used to instruct the allocation of preset collaborative tasks to corresponding vehicles among multiple vehicles, enabling those vehicles to execute the preset collaborative tasks. Thus, when the target vehicle is involved in an accident, surrounding vehicles can provide emergency collaborative response, meeting the requirements for intelligent and precise emergency response.

[0052] Figure 2 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 2 ,like Figure 2 As shown, in an optional implementation, step S103 above, which involves risk propagation of the dynamic risk field to obtain the first risk exposure of the first vehicle under the target driving path and the second risk exposure of the second vehicle under the original driving path, may include: S201. Based on the diffusion coefficient corresponding to the road network where the target vehicle is located and the preset attenuation coefficient, the dynamic risk field is spatially diffused and temporally attenuated to obtain a risk field grid map. Each grid in the risk field grid map stores the corresponding risk intensity.

[0053] The diffusion coefficient corresponding to the road network where the target vehicle is located refers to the diffusion coefficient of the roads on that network. The diffusion coefficient is related to the degree of road congestion; the more congested the road, the slower the risk propagation and the smaller the diffusion coefficient. Generally, highways have a larger diffusion coefficient, while alleyways have a smaller one. The preset attenuation coefficient decreases over time as the risk intensity decreases.

[0054] The dynamic risk field is spatially diffused according to the diffusion coefficient, and at the same time, the dynamic risk field is temporally decayed according to the preset decay coefficient to obtain a risk field grid map. Each grid in the risk field grid map stores the risk intensity of the corresponding location at each time. One grid stores the risk intensity of the next location at a certain time, that is, the risk intensity of a certain location at a certain time.

[0055] In other words, for each location on the road, the risk intensity at time t0 (x0, y0) is R(x0, y0, t0), and the risk intensity at other locations is 0. By spatially and temporally diffusing the dynamic risk field, the risk intensity of each location on the road at different times can be obtained.

[0056] S202. Based on the risk field grid map, obtain the first risk exposure and the second risk exposure.

[0057] From the risk field grid map, determine each first grid that matches each path point on the target driving path and the arrival time of each path point. Each first grid refers to the location and time corresponding to the risk intensity stored in it, which are the location of the path point on the target driving path and the time when the vehicle arrives at that path point.

[0058] From the risk field grid map, determine each second grid that matches each waypoint on the original driving path and the arrival time of each waypoint. Each second grid refers to the location and time corresponding to the risk intensity stored in it, which are the location of the waypoint on the original driving path and the time when the vehicle arrives at that waypoint.

[0059] In other words, within the grid of the risk field grid map, the grid that best matches the vehicle's waypoint in space and time is identified in order to extract the risk intensity corresponding to the vehicle's spatiotemporal path.

[0060] Then, based on the risk intensity stored in each first grid, the first risk exposure is obtained, and based on the risk intensity stored in each second grid, the second risk exposure is obtained.

[0061] In an optional implementation, step S202 above, obtaining the first risk exposure and the second risk exposure based on the risk field grid map, may include: Based on the risk field grid map, risk is accumulated along the target driving path to obtain the first risk exposure; based on the risk field grid map, risk is accumulated along the original driving path to obtain the second risk exposure.

[0062] Risk exposure is expressed as:

[0063] in, For vehicles The driving route, Indicates vehicle exist At any moment in the location The intensity of the risk, For spatial integration, For vehicles Risk exposure, indicating along the [missing information] Integrating the risk intensity, that is, the risk along... The accumulation.

[0064] The risk exposure is a continuous integral, but since the risk field is only known at discrete points in time and space, it can be approximated using numerical discretization. For example, the path can be divided into many segments, the risk of each segment can be approximated, and then the results can be summed.

[0065] Based on each path point on the target driving path, the target driving path is divided into multiple line segments (a path point can be located at the midpoint of a line segment). From the risk field grid map, the first grid that matches each path point on the target driving path and the arrival time corresponding to each path point is determined. Then, the product of the risk intensity stored in each first grid and the length of the corresponding line segment (the line segment on the target driving path where the path point is at the midpoint) is calculated, and the product is calculated. Finally, the products corresponding to all first grids are summed to obtain the first risk exposure.

[0066] Similarly, from the risk field grid map, each second grid that matches each waypoint on the original driving path and the arrival time corresponding to each waypoint is determined. Then, the product of the risk intensity stored in each second grid and the length of the corresponding line segment (the line segment of the original driving path where the waypoint is at the midpoint) is calculated, and the product is summed over all the products corresponding to the second grids to obtain the first risk exposure.

[0067] In this embodiment, the objectives are to minimize total risk and maximize total value. Minimizing total risk tends to keep vehicles away from high-risk areas, while maximizing total value aims to assign more tasks to capable vehicles. However, these vehicles may be located in or need to enter high-risk areas for rescue operations, leading to conflicts. Therefore, by solving a set of optimal trade-off solutions, the risk of vehicles being affected by accidents is taken into account while pursuing value maximization, preventing the pursuit of value maximization at the expense of total risk.

[0068] Figure 3 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 3 ,like Figure 3 As shown, in an optional implementation, step S105 above, obtaining task allocation information based on the first risk exposure, the second risk exposure, the first collaboration value, and the second collaboration value, may include: S301. Construct a multi-objective optimization model.

[0069] Among them, the multi-objective optimization model includes: the total risk optimization model and the total value optimization model.

[0070] The overall risk optimization model is expressed as:

[0071] in, For vehicles Risk exposure, For vehicles The risk indicator factor determines whether the driving path passes through a preset risk area. A preset risk area is defined as a region where the risk intensity at every location exceeds a preset threshold; that is, a high-risk area. Preset risk areas are determined based on a risk field grid map. If the vehicle... If the driving route passes through a pre-set risk area, then If the vehicle If the driving route does not pass through the preset risk area, then , This means calculating the product of the risk exposure of each vehicle and the risk indicator factor, and summing the products of all vehicles to obtain the total risk.

[0072] The total value optimization model is expressed as:

[0073] in, For vehicles The synergistic value, As a decision variable, used to indicate the vehicle Are pre-defined collaborative tasks assigned? (Such as early warning, rescue, video assistance), if the vehicle If a pre-defined collaborative task is assigned, then ,otherwise , This means calculating the product of the collaborative value of each vehicle and the decision variables, and summing the products of all vehicles to obtain the total value.

[0074] S302. Based on the first risk exposure, the second risk exposure, the first collaborative value, and the second collaborative value, the multi-objective optimization model is solved iteratively to obtain task allocation information.

[0075] Based on the first risk exposure, the second risk exposure, the first risk indicator factor, the second risk indicator factor, the first synergistic value, the second synergistic value, the first decision variable, and the second decision variable, the multi-objective optimization model is solved iteratively to obtain task allocation information.

[0076] Among them, the risk indicator factor is used to indicate whether the corresponding driving path passes through the preset risk area, and the decision variable is used to indicate whether the vehicle is assigned a preset collaborative task.

[0077] The first risk indicator factor is used to indicate whether the target driving path passes through a preset risk area, the second risk indicator factor is used to indicate whether the original driving path passes through a preset risk area, the first decision variable is used to indicate whether the first vehicle is assigned a preset collaborative task, and the second decision variable is used to indicate whether the second vehicle is assigned a preset collaborative task.

[0078] If the target driving path passes through a preset risk area, the first risk indicator factor is 1; otherwise, the risk indicator factor of the first vehicle is 0. If the original driving path passes through a preset risk area, the second risk indicator factor is 1; otherwise, the second risk indicator factor is 0.

[0079] Since the first vehicle is assigned a preset collaborative task and the second vehicle is not assigned a preset collaborative task, the first decision variable is 1 and the second decision variable is 0.

[0080] The total risk is obtained by using the total risk optimization model to calculate the product of the first synergy value and the first risk indicator factor, as well as the product of the second risk exposure and the second risk indicator factor, and summing the products of all vehicles.

[0081] Using the total value optimization model, the product of the first collaborative value and the second decision variable, as well as the product of the second collaborative value and the second decision variable, are calculated, and the products of all vehicles are summed to obtain the total value.

[0082] The first and second vehicles are re-determined from multiple vehicles. The total risk and total value are calculated iteratively. This process is repeated to adjust the allocation relationship between vehicles and preset collaborative tasks. With the goal of minimizing the total risk and maximizing the total value, the multi-objective optimization model is solved iteratively to find a balanced solution. The task allocation information is then determined based on this balanced solution.

[0083] In other words, by iteratively calculating the total risk and total value, and using optimization algorithms to find the optimal compromise between the total risk and total value, task allocation information is finally generated based on this compromise.

[0084] In some embodiments, under preset constraints, the multi-objective optimization model is solved iteratively to obtain task allocation information. The preset constraints may include, for example, that each preset collaborative task can only be assigned to one vehicle, that one vehicle can undertake at most one preset collaborative task, and communication range restrictions.

[0085] The communication range limit is used to indicate the maximum distance between vehicles that can reliably exchange data (such as location, speed, task status, risk field information, etc.). In other words, when multiple vehicles are working together to perform a task, they need to be able to communicate with each other, so the distance between vehicles cannot exceed this maximum distance.

[0086] In some embodiments, heuristic algorithms can be used for efficient approximate solutions, including the following steps: Step A: Transform the bi-objective problem into a single objective using Pareto optimization or a linear weighted method.

[0087] Step B: Initialize using a greedy algorithm, prioritizing collaborative value. Highest and highest risk exposure The most critical task is to allocate fewer vehicles.

[0088] Step C: Use local search or genetic algorithm to iteratively optimize the initial assignment and find a better solution while satisfying the constraints.

[0089] As an example, during the solution process, an optimization algorithm was used to find that vehicle V1 (short distance, short ETA, but no special equipment) could be assigned the upstream early warning task, which could significantly reduce the Ei of multiple vehicles after it, and its Vi score was high on this task. Vehicle V2 (carrying sand, longer ETA) was assigned the direct rescue task. Although its own Ei increased, its unique Vi (sand can handle leaks) contributed greatly to the total value. After iteration, the optimal solution was output: X_{V1, Early Warning}=1, X_{V2, Rescue}=1, X_{V3, Video}=1, thus outputting a set of optimal solutions. This allows us to obtain task allocation information.

[0090] In this embodiment, at the dynamic decision planning layer, by constructing a parameterized dynamic risk field and a quantified value function, the complex traffic emergency scenario is abstracted into a computable mathematical problem. Then, by defining a well-defined multi-objective optimization problem, the strategy problem of who to notify and what to do is transformed into a mathematical model that can obtain an approximate optimal solution.

[0091] Figure 4 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 4 ,like Figure 4 As shown, in an optional implementation, step S104, which involves obtaining the first collaborative value of the first vehicle based on the vehicle information of the first vehicle, and obtaining the second collaborative value of the second vehicle based on the vehicle information of the second vehicle, may include: S401. Based on the rescue equipment information and personnel skill information of the first vehicle, the estimated arrival time for the preset collaborative task, and the urgency of the transportation task, obtain the first collaborative value.

[0092] The vehicle information of the first vehicle includes: the rescue equipment information of the first vehicle, personnel skill information, the estimated arrival time for the preset collaborative task, and the urgency of the transportation task. For the rescue task, the estimated arrival time is the time when the vehicle arrives at the accident location. For the video assistance task, the estimated arrival time is the time when the vehicle arrives at the video acquisition location. For the upstream early warning task, the estimated arrival time is the time when the vehicle arrives at the early warning location.

[0093] The urgency of a transportation task refers to the degree of urgency of the task being transported by the vehicle itself.

[0094] Based on the rescue equipment information of the first vehicle, calculate the score of the vehicle-mounted rescue equipment. For example, a fire extinguisher corresponds to a score of 1, and a first aid kit corresponds to a score of 0.5. If both a fire extinguisher and a first aid kit are present, the score is 1.5. If neither is present, the score is 0.

[0095] Personnel skill information can include the emergency skill rating of drivers certified by the platform. The rating corresponds to the driver's emergency skill score. The priority of the task can be determined according to the urgency of the transportation task. If the first vehicle is transporting corrosive substances, the task urgency is high and the task priority can be 8. If the vehicle is empty, the task urgency is low and the task priority can be 0.

[0096] Using a value function, the value of the first collaboration is calculated based on the vehicle's onboard rescue equipment score, the driver's emergency skills score, the estimated arrival time for the preset collaborative task, and the task priority. The value function is expressed as follows:

[0097] in, For vehicles The synergistic value, As a preset reference time constant, For vehicles The preset arrival time, For vehicles Rating of vehicle-mounted rescue equipment For vehicles Driver emergency skills score For vehicles Task priority, , , , These are weighting coefficients, which can be selected based on the actual situation.

[0098] For example, if vehicle A has an ETA of 5 minutes, a resource score of 80, a driver emergency skills score of 90, and a task priority of 2 (lower), then... If the weighting coefficients are all 0.25, then the first synergistic value = 0.25 × (10 / 5) + 0.25 × 80 + 0.25 × 90 0.25 × 2 = 42.5.

[0099] S402. Obtain the second collaborative value based on the rescue equipment information, personnel skill information, and urgency of the transportation mission of the second vehicle.

[0100] Since the second vehicle has not been assigned a collaborative task, there is no estimated arrival time for the anticipated collaborative task. Therefore, the aforementioned value function is adopted, letting... =0, and calculate the second collaborative value based on the vehicle's onboard rescue equipment score, driver's emergency skills score, and task priority.

[0101] In this embodiment, the collaborative value is quantified by using multiple dimensions (timeliness, resources, skills, and current load) to facilitate subsequent iterations in determining task allocation information.

[0102] Figure 5 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 5 ,like Figure 5 As shown, in an optional implementation, the method may further include: S501, Obtain motion data and video data of the target vehicle.

[0103] S502. Based on the motion data, obtain the temporal motion characteristics that meet the preset motion anomaly conditions.

[0104] During the driving process of the target vehicle, motion data and video data are reported to the cloud device in real time. The motion data refers to the data of the Inertial Measurement Unit (IMU), including vehicle position, speed, acceleration, attitude (yaw rate, roll angle), and status (braking status, turn signal status, gear position, etc.).

[0105] Video data refers to the video stream captured by in-vehicle cameras (front-view, surround-view, or dashcam).

[0106] Anomaly identification is performed on the motion data, and abnormal data is extracted from the motion data. Abnormal data refers to data that does not meet the preset motion requirements. For example, if the roll angle is too large, the roll angle is determined to be abnormal data. Then, based on the abnormal data, motion features that meet the time sequence are generated. The time sequence motion features are motion features that meet the preset abnormal conditions in time sequence, such as a lateral impact lasting 300ms and accompanied by rotation.

[0107] S503. Extract the target video segment from the video data based on the abnormal time points of the temporal motion characteristics.

[0108] The abnormal time point of the temporal motion feature is the time point when the corresponding abnormal data occurs (start time point). Based on the abnormal time point, the target video segment is extracted from the video data. The start time of the target video segment is earlier than the abnormal time point. For example, video segments 10 seconds before and after the abnormal time point (a total of 20 seconds) are selected as the target video segments.

[0109] S504. Based on the temporal motion characteristics and the target video segment, a pre-set Bayesian network is used to determine the probability of the accident.

[0110] Bayesian networks are directed acyclic graph (DAG) models. Their topology is constructed based on the traffic accident logic and includes root nodes, intermediate nodes, and leaf nodes. The root node represents the cause, corresponding to IMU feature patterns (temporal motion features, such as high-intensity lateral impact waveforms) and visual events (such as the visual features of airbag deployment). The intermediate nodes represent the direct results caused by the cause, corresponding to the vehicle's physical state (such as airbag trigger signals). The leaf nodes represent the observation results, corresponding to the accident type (such as side collision) and the confidence level of the vehicle accident. The confidence level is used to characterize the probability that the accident type is true.

[0111] Historical accident data, including labeled temporal motion features, video clips, and vehicle bus data (vehicle physical state), is used to train a Bayesian network to learn the conditional probability table for each node in the network. For example, the probability P (airbag deployment | IMU side impact = high intensity, visual airbag event = yes) is learned, representing the probability that the airbag will be deployed under the condition that the IMU side impact is high intensity and the visual airbag event is yes. That is, the confidence level of airbag deployment when both the IMU and the vehicle camera are involved in a severe collision.

[0112] After constructing the Bayesian network, during inference, the real-time extracted temporal motion features and target video clips are input as evidence into the Bayesian network. Through Bayesian inference algorithms (such as variable elimination for precise inference or approximate inference suitable for lightweight inference), the learned conditional probability table is queried to calculate the accident probability (posterior probability) P (accident type = true | all evidence), which represents the probability that the accident type is true under the condition of all evidence (causal confidence).

[0113] S505. If the probability of an accident exceeds a preset threshold, then it is determined that the target vehicle has been involved in a vehicle accident.

[0114] If the accident probability is 0.92 and the preset threshold is 0.85, then the accident probability exceeds the preset threshold, and the target vehicle has a vehicle accident corresponding to the accident probability. For example, the accident probability of a single-vehicle skid accident is 0.92, which exceeds the threshold of 0.85, so a single-vehicle skid accident is confirmed to exist.

[0115] In this embodiment, the parameter learning and inference process of the Bayesian network can be referred to the existing related implementation process. The core of this embodiment is to use the Bayesian network to realize an objective, probability-based inference calculation process. Thus, at the information fusion and cognition layer, through the cross-modal causal probabilistic graphical model (Bayesian network), the inference relationship between different temporal motion features, visual events and accident states is formally defined at the algorithm level, realizing a high-order scene understanding that is interpretable based on probabilistic reasoning and improving the accuracy of vehicle accident recognition.

[0116] Figure 6 A flowchart illustrating the vehicle accident emergency response method provided in this application embodiment. Figure 6 ,like Figure 6 As shown, the method may further include: S601. A backbone feature extraction network is used to extract features from the target video segment to obtain the target image features.

[0117] The cloud-based analytics uses a shared backbone feature extraction network (such as the lightweight MobileNetV3) to connect multi-task sub-networks. These multi-task sub-networks can include an accident posture sub-network (accident posture classification head), a risk detection sub-network (risk detection head), and a personnel status detection sub-network (personnel status recognition head).

[0118] The accident pose subnetwork can use a convolutional neural network (CNN) and fully connected layers to identify vehicle accidents such as rollovers and rear-end collisions, and output the confidence level of the vehicle accident, such as a 92% confidence level for skidding.

[0119] The risk detection subnetwork uses object detection algorithms (such as YOLOv5s) to detect flames, smoke, liquids, and warning triangle lights, outputting the confidence level of the accident risk. For example, a fuel leak has an 88% confidence level, and vehicle body friction with guardrail has a 95% confidence level. Here, accident risk refers to the risk caused by an accident involving the vehicle.

[0120] The personnel status detection subnetwork uses a pose estimation model (such as OpenPose) combined with a recurrent neural network (RNN) to analyze personnel action sequences, such as waving, falling, or being trapped, and outputs personnel status, such as the driver being in the car, in a normal posture, or operating a mobile phone (estimated to be triggering an alarm).

[0121] S602. A multi-task sub-network is adopted to identify the confidence level of vehicle accidents, the confidence level of accident risks, and the status of personnel based on the features of the target image.

[0122] A backbone feature extraction network is used to extract features from the target video segment to obtain target image features. An accident pose sub-network is used to identify the confidence level of the vehicle accident based on the target image features. A risk detection sub-network is used to identify the confidence level of the accident risk based on the target image features. A personnel status detection sub-network is used to identify the personnel status based on the target image features.

[0123] S603. Based on the confidence level of the vehicle accident and the confidence level of the accident risk, match emergency suggestions from the preset rule base.

[0124] If the confidence level of a vehicle accident exceeds a preset first confidence level threshold, and the confidence level of the accident risk exceeds a preset second confidence level threshold, then emergency suggestions for the vehicle accident and accident risk are matched from a preset rule base based on the vehicle accident and accident risk. The preset rule base includes the correspondence between multiple vehicle accidents and multiple accident risks.

[0125] For example, based on skidding and fuel leakage, the emergency suggestion generated from the rule base is: There is a fuel leak on site, rescuers need to strictly prevent fire, and sand can be used to cover it.

[0126] S604. Generate structured emergency data based on vehicle accidents, accident risks, personnel status, and emergency recommendations.

[0127] Structured emergency data (structured summaries) includes: vehicle accidents, confidence levels of accident risk, personnel status, and emergency recommendations.

[0128] Structured emergency data can be represented as: { "event": "vehicle_side_slip", "Vehicle side slip" "risk": ["fuel_leakage", "friction_with_guardrail"], ["fuel leak", "friction with guardrail"] "person_status": "driver_in_cabin_normal", "The driver is in the cockpit and functioning normally." "Suggestion": "Fire prevention is required. Use sand for leakage if available." S605. Send structured emergency data to the corresponding vehicle.

[0129] By sending structured emergency data to the corresponding vehicle and displaying the task information and structured emergency data on the central control screen, the corresponding vehicle can be informed of the current status of the target vehicle in a timely manner, which facilitates the corresponding vehicle to carry out rescue.

[0130] In some embodiments, when a fuel leak occurs, four video frames containing the leak point, vehicle posture, and driver status can be selected to generate a keyframe mosaic, which is then sent to the corresponding vehicle.

[0131] In some embodiments, the vehicle performing the rescue mission, such as vehicle V2, is determined from the corresponding vehicles, and structured emergency data and keyframe mosaics are sent to vehicle V2.

[0132] It should be noted that a very lightweight model can run on the vehicle side to trigger and upload video data, while complex multi-task analysis is performed in the cloud to balance real-time performance and accuracy.

[0133] In this embodiment, the knowledge summary generation layer designs an edge-cloud collaborative multi-task neural network model to analyze time-consuming video segments, decompose them into multiple parallel and efficient sub-tasks (classification, detection, and recognition), and automatically synthesize structured knowledge summaries that can be efficiently processed by both humans and machines, thus realizing a qualitative change in information from raw data to decision-making knowledge.

[0134] Figure 7 This is a schematic diagram of the structure of a vehicle accident emergency response device provided in an embodiment of this application. This device can be integrated into a cloud device.

[0135] like Figure 7 As shown, the device may include: Initialization module 701 is used to initialize the dynamic risk field of the target vehicle where a vehicle accident has occurred; The determining module 702 is used to determine the first vehicle and the second vehicle from multiple vehicles within a preset range of the target vehicle, and to obtain the target driving path of the first vehicle for the preset collaborative task, and the original driving path of the second vehicle. The processing module 703 is used to propagate risks in the dynamic risk field to obtain the first risk exposure of the first vehicle under the target driving path and the second risk exposure of the second vehicle under the original driving path. The risk exposure is used to indicate the risk of the vehicle being affected by a vehicle accident when the vehicle is traveling along the corresponding driving path. The acquisition module 704 is used to acquire the first collaborative value of the first vehicle based on the vehicle information of the first vehicle, and to acquire the second collaborative value of the second vehicle based on the vehicle information of the second vehicle. The collaborative value is used to indicate the emergency response collaborative capability of the corresponding vehicle. The acquisition module 704 is also used to iteratively optimize based on the first risk exposure, the second risk exposure, the first collaborative value, and the second collaborative value to acquire task allocation information. The task allocation information is used to indicate the allocation of the preset collaborative task to the corresponding vehicle among the multiple vehicles. The sending module 705 is used to send task instructions to the corresponding vehicle so that the corresponding vehicle can perform a preset collaborative task.

[0136] In an optional implementation, the processing module 703 is specifically used for: Based on the diffusion coefficient corresponding to the road network where the target vehicle is located and the preset attenuation coefficient, the dynamic risk field is spatially diffused and temporally attenuated to obtain a risk field grid map. Each grid in the risk field grid map stores the risk intensity at the corresponding location at each time. Based on the risk field grid map, obtain the first risk exposure and the second risk exposure.

[0137] In an optional implementation, the processing module 703 is specifically used for: Based on the risk field grid map, risk is accumulated along the target driving path to obtain the first risk exposure level; Based on the risk field grid map, risk is accumulated along the original driving path to obtain the second risk exposure.

[0138] In an optional implementation, the acquisition module 704 is specifically used for: Construct a multi-objective optimization model, which includes: a total risk optimization model and a total value optimization model; Based on the first risk exposure, the second risk exposure, the first collaborative value, and the second collaborative value, the multi-objective optimization model is solved iteratively to obtain task allocation information.

[0139] In an optional implementation, the acquisition module 704 is specifically used for: Based on the first risk exposure, the second risk exposure, the first risk indicator factor, the second risk indicator factor, the first collaborative value, the second collaborative value, the first decision variable, and the second decision variable, the multi-objective optimization model is solved iteratively to obtain task allocation information. The risk indicator factor is used to indicate whether the corresponding driving path passes through the preset risk area, and the decision variable is used to indicate whether the vehicle is assigned a preset collaborative task.

[0140] In an optional implementation, the acquisition module 704 is specifically used for: The first collaborative value is obtained based on the rescue equipment information and personnel skill information of the first vehicle, the estimated arrival time for the preset collaborative task, and the urgency of the transportation task. The second collaborative value is obtained based on the rescue equipment information, personnel skill information, and urgency of the transportation mission of the second vehicle.

[0141] In an optional implementation, the acquisition module 704 is further configured to: Acquire motion and video data of the target vehicle; Based on the motion data, obtain the temporal motion characteristics that meet the preset abnormal conditions; Extract target video segments from video data based on abnormal time points of temporal motion characteristics; The determination module 702 is also used to determine the probability of an accident based on the temporal motion characteristics and the target video segment using a preset Bayesian network; The determination module 702 is also used to determine that the target vehicle has been involved in a vehicle accident if the probability of the accident exceeds a preset threshold.

[0142] In an optional implementation, the acquisition module 704 is further configured to: A backbone feature extraction network is used to extract features from the target video segment to obtain the target image features; The processing module 703 is also used to employ a multi-task sub-network to identify the confidence level of the vehicle accident, the confidence level of the accident risk, and the status of the personnel based on the features of the target image. The processing module 703 is also used to match emergency suggestions from a preset rule base based on the confidence level of the vehicle accident and the confidence level of the accident risk. The processing module 703 is also used to generate structured emergency data based on vehicle accidents, accident risks, personnel status, and emergency recommendations. The sending module 705 is also used to send structured emergency data to the corresponding vehicle.

[0143] The processing flow of each module in the device and the interaction flow between each module can be referred to the relevant descriptions in the above method embodiments, and will not be detailed here.

[0144] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. This device can be implemented through the aforementioned cloud device.

[0145] like Figure 8 As shown, the device may include a processor 801, a memory 802, and a bus 803. The memory 802 stores machine-readable instructions that can be executed by the processor 801. When the electronic device is running, the processor 801 communicates with the memory 802 through the bus 803, and the processor 801 executes the machine-readable instructions to perform the above-described method embodiments.

[0146] This application also provides a computer-readable storage medium storing a computer program, which is executed by a processor to perform the above-described method embodiments.

[0147] In this embodiment, the computer program, when run by the processor, can also execute other machine-readable instructions to perform other methods as described in the embodiments. For details on the specific execution steps and principles, please refer to the description of the embodiments, which will not be repeated here.

[0148] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0149] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0150] In addition, the functional units in the embodiments provided in this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0151] If the aforementioned functions are implemented as software functional 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 this application, in essence, or the part that contributes to the prior art, or a portion of the 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 to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0152] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0153] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application. All should be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A vehicle accident emergency response method, characterized in that, Applied to cloud devices, the method includes: Initialize the dynamic risk field of the target vehicle where a vehicle accident has occurred; The first vehicle and the second vehicle are determined from multiple vehicles within a preset range of the target vehicle, and the target driving path of the first vehicle for the preset collaborative task and the original driving path of the second vehicle are obtained. Risk propagation is performed on the dynamic risk field to obtain the first risk exposure of the first vehicle under the target driving path and the second risk exposure of the second vehicle under the original driving path. The risk exposure is used to indicate the risk of the vehicle being affected by the vehicle accident when the vehicle is traveling along the corresponding driving path. Based on the vehicle information of the first vehicle, a first collaborative value of the first vehicle is obtained, and based on the vehicle information of the second vehicle, a second collaborative value of the second vehicle is obtained. The collaborative value is used to indicate the emergency response collaborative capability of the corresponding vehicle. Based on the first risk exposure, the second risk exposure, the first collaboration value, and the second collaboration value, iterative optimization is performed to obtain task allocation information, which is used to indicate the allocation of the preset collaboration task to the corresponding vehicle among the multiple vehicles. Send a task instruction to the corresponding vehicle so that the corresponding vehicle performs the preset collaborative task.

2. The method according to claim 1, characterized in that, The step of propagating risk in the dynamic risk field to obtain the first risk exposure of the first vehicle under the target driving path and the second risk exposure of the second vehicle under the original driving path includes: Based on the diffusion coefficient corresponding to the road network where the target vehicle is located and the preset attenuation coefficient, the dynamic risk field is spatially diffused and temporally attenuated to obtain a risk field grid map. Each grid in the risk field grid map stores the risk intensity of the corresponding location at each time. Based on the risk field grid map, obtain the first risk exposure and the second risk exposure.

3. The method according to claim 2, characterized in that, The step of obtaining the first risk exposure and the second risk exposure based on the risk field grid map includes: Based on the risk field grid map, risk is accumulated along the target driving path to obtain the first risk exposure. Based on the risk field grid map, risk is accumulated along the original driving path to obtain the second risk exposure.

4. The method according to claim 1, characterized in that, The step of iteratively optimizing based on the first risk exposure, the second risk exposure, the first collaborative value, and the second collaborative value to obtain task allocation information includes: Construct a multi-objective optimization model, which includes: a total risk optimization model and a total value optimization model; Based on the first risk exposure, the second risk exposure, the first collaborative value, and the second collaborative value, the multi-objective optimization model is solved iteratively to obtain the task allocation information.

5. The method according to claim 4, characterized in that, The step of performing multi-objective iterative solution on the multi-objective optimization model based on the first risk exposure, the second risk exposure, the first collaborative value, and the second collaborative value to obtain the task allocation information includes: Based on the first risk exposure, the second risk exposure, the first risk indicator factor, the second risk indicator factor, the first collaborative value, the second collaborative value, the first decision variable, and the second decision variable, the multi-objective optimization model is solved iteratively to obtain the task allocation information. The risk indicator factor is used to indicate whether the corresponding driving path passes through the preset risk area, and the decision variable is used to indicate whether the vehicle is assigned the preset collaborative task.

6. The method according to claim 1, characterized in that, The step of obtaining a first collaborative value of the first vehicle based on the vehicle information of the first vehicle, and obtaining a second collaborative value of the second vehicle based on the vehicle information of the second vehicle, includes: The first collaborative value is obtained based on the rescue equipment information and personnel skill information of the first vehicle, the estimated arrival time for the preset collaborative task, and the urgency of the transportation task. The second collaborative value is obtained based on the rescue equipment information, personnel skill information, and urgency of the transportation mission of the second vehicle.

7. The method according to claim 1, characterized in that, The method further includes: Acquire the motion data and video data of the target vehicle; Based on the motion data, obtain the temporal motion characteristics that meet the preset abnormal conditions; Based on the abnormal time points of the temporal motion characteristics, extract the target video segment from the video data; Based on the temporal motion characteristics and the target video segment, a preset Bayesian network is used to determine the probability of an accident. If the probability of the accident exceeds a preset threshold, then it is determined that the target vehicle has been involved in the accident.

8. The method according to claim 7, characterized in that, The method further includes: A backbone feature extraction network is used to extract features from the target video segment to obtain target image features; A multi-task sub-network is used to identify the confidence level of the vehicle accident, the confidence level of the accident risk, and the status of the personnel based on the features of the target image. Emergency suggestions are matched from a preset rule base based on the confidence level of the vehicle accident and the confidence level of the accident risk. Based on the vehicle accident, the accident risk, the personnel status, and the emergency recommendations, structured emergency data is generated; The structured emergency data is sent to the corresponding vehicle.

9. An electronic device, characterized in that, include: The electronic device includes a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is in operation, the processor communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the method according to any one of claims 1 to 8.

Images