Traffic targeting regulation and control method based on vehicle and road cloud multi-source information fusion

Abstract

The invention provides a traffic targeted regulation and control method based on vehicle and road cloud multi-source information fusion, and the method comprises the steps: determining the attribute information of a new event according to vehicle end data, road side data and cloud end data; according to the attribute information, the current environment parameters of the environment where the newly-added event occurs and the road network characteristics of the road section where the newly-added event occurs, calculating the influence domain of the newly-added event at each future moment; candidate vehicles are determined according to the spatial distance between the current position of each vehicle and the boundary of the influence domain, and vehicles whose predicted trajectories will enter the influence domain in all future moments are screened out from all the candidate vehicles and determined as target vehicles influenced by newly added events; and pushing a corresponding regulation and control instruction to each target vehicle according to the vehicle identifier of each target vehicle. According to the invention, the pertinence and effectiveness of the regulation and control instruction are improved, and the information interference of irrelevant vehicles and the missed judgment of affected vehicles are avoided.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and more specifically, to a traffic targeted control method based on the fusion of multi-source information from vehicles, roads, and cloud. Background Technology

[0002] The core of traffic control lies in precisely intervening in affected vehicles based on the scope of a traffic incident's impact. Existing technologies mostly use the fixed radius method to divide the incident's impact domain, that is, estimating the impact range based on static geometric rules and indiscriminately pushing control information to all vehicles within that range.

[0003] However, the fixed radius division method in the existing technology does not integrate the severity of the event, environmental parameters and road network characteristics, resulting in rigid boundary of the influence domain and inability to match the dynamic evolution of the event. This causes the control range to be out of sync with the actual traffic situation, making it difficult to achieve precise targeted control. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a traffic targeted control method based on vehicle-road-cloud multi-source information fusion, which aims to overcome at least one of the above-mentioned defects.

[0005] Firstly, this application provides a traffic-targeted control method based on vehicle-road-cloud multi-source information fusion, including: Based on vehicle-side data, roadside data, and cloud data, determine the attribute information of newly added events; Based on the attribute information, the current environmental parameters of the environment in which the new event occurs, and the road network characteristics of the road segment where the new event is located, the influence domain of the new event at each future time is calculated. Candidate vehicles are determined based on the spatial distance between the current position of each vehicle and the boundary of the influence domain. Then, from all candidate vehicles, vehicles whose predicted trajectories will enter the influence domain in future moments are selected and identified as target vehicles affected by the new event. Based on the vehicle identifier of each target vehicle, corresponding control instructions are pushed to each target vehicle.

[0006] In one possible implementation, the attribute information of the newly added event is determined in the following way: Acquire the vehicle-side data and roadside data that conform to a unified semantic specification, and upload the vehicle-side data and roadside data to the corresponding edge nodes; Each edge node performs spatiotemporal registration of the received vehicle-side data and roadside data, and performs complementary verification based on the registered vehicle-side data and roadside data to identify mutually corroborating data pairs as candidate events. The cloud platform uses a spatiotemporal correlation analysis model to perform correlation analysis on candidate events reported by multiple edge nodes. By combining the weighted fusion of roadside device credibility weight and vehicle user credit level, conflicts are resolved and fused among multiple data sources pointing to the same real event, so as to obtain the attribute information of the newly added event.

[0007] In one possible implementation, the vehicle-side data and the roadside data are spatiotemporally registered in the following manner: The vehicle-side data and the roadside data are synchronized in time so that they can be compared on the same time reference. The vehicle-side data and the roadside data after time synchronization are transformed into map coordinates so that the vehicle-side data and the roadside data can be compared in the same spatial coordinate system.

[0008] In one possible implementation, it also includes: When the current environmental parameters indicate severe weather, the original image is preprocessed, including eliminating raindrop interference and improving nighttime perception accuracy. The preprocessed original image is input into a multi-scale event recognition model to extract motion trajectories and contour matching in order to identify roadside events under severe weather conditions.

[0009] In one possible implementation, the vehicle identification includes a social vehicle identification; Among them, the control instructions pushed to target vehicles with social vehicle identification include: Personalized route optimization and service area recommendations generated based on real-time traffic conditions, tourism resources, and weather data; and / or blind spot warning instructions generated based on at least one of the following warning methods: text pop-ups, voice reminders, or audible and visual alarms.

[0010] In one possible implementation, the vehicle identification includes an operating vehicle identification; The control instructions pushed to target vehicles with the vehicle identification number of commercial vehicles include: Safety prompts based on order hotspot pushes and optimal route planning generated from cloud-based order heatmaps, and / or safety prompts based on dynamic avoidance suggestions for curves and large vehicle blind spots generated from roadside perception data.

[0011] In one possible implementation, the vehicle identification includes an emergency vehicle identification; The control instructions pushed to target vehicles identified as emergency vehicles include: Priority passage instructions generated by the avoidance guidance assessment model and directed to other vehicles ahead, and / or route guidance instructions generated and displayed based on the roadside guidance screen linkage strategy for emergency dedicated lane planning.

[0012] Secondly, this application provides a traffic targeted control device based on vehicle-road-cloud multi-source information fusion, comprising: The vehicle-road-cloud semantic collaboration and fusion module is used to determine the attribute information of newly added events based on vehicle-side data, roadside data, and cloud data. The influence domain dynamic partitioning module is used to calculate the influence domain of the new event at each future time based on the attribute information, the current environmental parameters of the environment in which the new event occurs, and the road network characteristics of the road segment where the new event is located. The dynamic targeting control module is used to determine candidate vehicles based on the spatial distance between the current position of each vehicle and the boundary of the influence domain, and then filter out all vehicles whose predicted trajectories will enter the influence domain in future moments from all candidate vehicles, and determine them as the target vehicles affected by the new event. The multi-type vehicle classification and empowerment module is used to push corresponding control instructions to each target vehicle based on the vehicle identifier of each target vehicle.

[0013] Thirdly, this application also provides 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 when the machine-readable instructions are executed by the processor, the steps of the method described above are performed.

[0014] Fourthly, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the method described above.

[0015] This application provides a traffic targeted control method based on vehicle-road-cloud multi-source information fusion. The method includes: determining the attribute information of a new event based on vehicle-side data, roadside data, and cloud data; calculating the influence domain of the new event at each future time based on the attribute information, the current environmental parameters of the environment where the new event occurs, and the road network characteristics of the road segment where the new event is located; determining candidate vehicles based on the spatial distance between the current position of each vehicle and the boundary of the influence domain, and then selecting all vehicles whose predicted trajectories will enter the influence domain in future time moments from all candidate vehicles, identifying them as target vehicles affected by the new event; and pushing corresponding control instructions to each target vehicle based on its vehicle identifier. This application achieves accurate identification of target vehicles through dynamic influence domain division and dual-layer screening, improving the targeting and effectiveness of control instructions and avoiding information interference from irrelevant vehicles and missed identification of affected vehicles.

[0016] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. 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 a traffic-targeted control method based on vehicle-road-cloud multi-source information fusion provided in this application embodiment; Figure 2 A flowchart for determining the attribute information of a newly added event, provided in an embodiment of this application; Figure 3 This is a schematic diagram of the traffic targeted control device based on vehicle-road-cloud multi-source information fusion provided in an embodiment of this application; Figure 4 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. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0020] First, the applicable scenarios for this application will be introduced. This application can be applied to the field of data processing technology.

[0021] Research has found that existing vehicle-road cooperative systems mainly rely on the "wide-area broadcast" mode in traffic incident response. They send warning information indiscriminately to all vehicles within a fixed radius around the incident through roadside variable message signs or regional broadcasts, delineate the influence zone based on static geometric rules, filter target vehicles based on the instantaneous geographical location of vehicles, and push highly homogenized control instructions.

[0022] However, in terms of the way information is disseminated, warnings are indiscriminately issued to all vehicles within a fixed radius around the incident via roadside variable message signs or regional V2X broadcasts without distinguishing between vehicle type, driving intention, and task status. This directly leads to insufficient accuracy and timeliness of the control measures. A large number of irrelevant vehicles receive redundant information, while vehicles that are actually about to enter the danger zone may miss the best time to change course due to information overload, which seriously affects the efficiency of the road network.

[0023] In terms of methods for dividing the impact domain of an event, traditional methods estimate the impact range based on static geometric rules. They fail to establish a dynamic spatiotemporal model that integrates the evolution patterns of the event (such as the time required to handle the accident and the speed at which congestion dissipates) and environmental characteristics (such as the friction coefficient of icy and snowy roads in seasonally frozen areas and the curvature of mountain roads). This results in rigid boundaries of the impact domain and poor adaptability in complex environments. For example, the speed at which congestion spreads is much faster on icy and snowy days than at other times, making fixed threshold divisions prone to failure and causing the control range to lag behind the actual speed at which congestion spreads.

[0024] Regarding the technical means of target vehicle screening, the existing screening logic is based on the instantaneous geographical location of the vehicle, only determining whether the vehicle is "within the circle at this moment", ignoring the prediction of future spatiotemporal trajectory, and not integrating destination, historical driving habits and real-time speed vector. This also exacerbates the problem of insufficient accuracy - a large number of vehicles that only pass through the boundary without entering the core congestion area are misjudged as control targets, causing ineffective interference.

[0025] In terms of the generation and push mode of control instructions, the content of the instructions is highly homogenized, and only general "congestion / accident" status labels are pushed. There is a lack of a differentiated strategy generation mechanism based on vehicle business attributes (such as ride-hailing order acceptance efficiency and ambulance green wave demand). The push protocol relies heavily on standard broadcast frames and lacks targeted unicast or multicast adaptation. This directly leads to a lack of differentiated empowerment for service targets - the diversified needs of operating vehicles, emergency vehicles and social vehicles cannot be met, special vehicles have difficulty obtaining priority right-of-way, operating vehicles cannot obtain congestion avoidance and efficiency improvement suggestions, and the system service stickiness is low.

[0026] In the multi-source data processing approach of vehicle-road-cloud, vehicle trajectory, roadside perception, and cloud-based road network data are in a state of "data silos." There is a lack of unified semantic consistency representation among the multi-source data (e.g., the "sudden braking" perceived by the vehicle and the "queueing" perceived by the roadside cannot be effectively mapped to the same event), and there is a lack of conflict resolution mechanisms based on multi-source mutual trust. This leads to semantic conflicts and value loss in multi-source heterogeneous data—when vehicle-side and roadside information are inconsistent, the system decision wavers or directly discards data, failing to utilize the complementarity of multi-source information to improve perception accuracy, especially in adverse weather or perception-limited scenarios where data reliability is difficult to maintain. Furthermore, existing technologies typically "issue the command and that's it," lacking real-time monitoring and effect evaluation of vehicle response behavior. They cannot dynamically adjust the next round of strategies based on feedback from the road network status after regulation, resulting in the inability of the regulation model to self-iterate and optimize. This constitutes the fundamental reason for the lack of closed-loop feedback and continuous optimization capabilities.

[0027] Based on this, the embodiments of this application provide a traffic targeted control method based on the fusion of multi-source information from vehicles, roads and cloud, which significantly improves the ability to prevent traffic accidents, the efficiency of road network traffic and the execution rate of instructions.

[0028] Please see Figure 1 , Figure 1 This is a flowchart illustrating a traffic-targeted control method based on vehicle-road-cloud multi-source information fusion, provided as an embodiment of this application. Figure 1 As shown in the embodiments of this application, the traffic targeted control method based on vehicle-road-cloud multi-source information fusion includes: S101. Determine the attribute information of the newly added event based on vehicle-side data, roadside data, and cloud data.

[0029] This step first requires acquiring multi-source data that conforms to a unified semantic specification. A unified semantic specification refers to defining core terminology standards and data formats for vehicle-side, roadside, and cloud-side data by establishing a vehicle-roadside-cloud interoperability transmission protocol, thereby eliminating semantic ambiguity and enabling interoperability of the three-party data at the syntactic and semantic levels.

[0030] Please see Figure 2 , Figure 2A flowchart illustrating the determination of attribute information for newly added events, provided in an embodiment of this application. Includes: S201. Obtain vehicle-side data and roadside data that conform to the unified semantic specification, and upload the vehicle-side data and roadside data to the corresponding edge nodes.

[0031] Here, vehicle-side data may include vehicle ID, real-time location, speed, direction, acceleration, historical trajectory points, ABS trigger status, hazard light status, vehicle type, and user credit rating; roadside data may include raw video streams, radar point clouds, meteorological data, equipment self-test status, and equipment static credibility baseline.

[0032] S202. The received vehicle-side data and roadside data are spatiotemporally registered through each edge node, and complementary verification is performed based on the registered vehicle-side data and roadside data to identify mutually corroborating data pairs as candidate events.

[0033] The spatiotemporal registration of vehicle-side data and roadside data is performed using the following methods: Time synchronization is performed between vehicle-side data and roadside data to make them comparable on the same time reference. Map coordinate transformation is then performed on the time-synchronized vehicle-side data and roadside data to make them comparable in the same spatial coordinate system.

[0034] Specifically, time synchronization refers to synchronizing vehicle-side data with roadside data so that they can be compared on the same time reference, with an allowable error of no more than 10ms; map coordinate transformation refers to performing high-precision map coordinate transformation on the time-synchronized vehicle-side data and roadside data so that they can be compared in the same spatial coordinate system, and uniformly transforming the relative coordinates of vehicle-side GPS latitude and longitude and roadside radar to the high-precision map coordinate system.

[0035] Complementary verification refers to comparing and correlating vehicle-side data and roadside data after spatiotemporal registration according to time windows and spatial ranges. If the time difference between vehicle-side events and roadside events is less than a preset threshold, the spatial distance is less than a preset threshold, and the events are semantically related (e.g., there is a causal relationship between "emergency braking" on the vehicle and "queueing" on the roadside), then the two are determined to corroborate each other, and the data pair is identified as a candidate event. If only one side reports, it is marked as pending confirmation. If there is a semantic contradiction, it is marked as a conflict event and uploaded to the cloud.

[0036] S203. Using a cloud platform, a spatiotemporal correlation analysis model is used to perform correlation analysis on candidate events reported by multiple edge nodes. By combining the weighted fusion of roadside device credibility weight and vehicle user credit rating, conflict resolution and fusion are performed on multiple data sources pointing to the same real event to obtain the attribute information of the new event.

[0037] Here, the spatiotemporal correlation analysis model is used to determine whether candidate events reported by different edge nodes point to the same real event (e.g., multiple roadside cameras detect the same accident). A weighted fusion calculation is performed to determine the final confidence level. The dynamic confidence weights of roadside devices can be dynamically adjusted based on device health status and historical accuracy. Vehicle-side user credit ratings can be based on the vehicle's historical reporting accuracy. Through this mechanism, conflicts are resolved and fused among multiple data sources pointing to the same real event, ultimately determining the attribute information of newly added events, including event type, severity level, precise location, and confidence level.

[0038] In addition, the cloud platform maintains a real-time update closed loop, including second-level synchronization of emergencies (new events are immediately triggered for processing) and minute-level periodic status verification (recalculating the roadside equipment credibility weight and vehicle user credit rating every 1-5 minutes), ensuring the real-time nature of global information and the dynamic credibility of data sources.

[0039] As another example, conflict resolution methods can be replaced by probabilistic conflict resolution based on Bayesian networks or evidence synthesis methods based on DS evidence theory, with the core still being to eliminate semantic conflicts in multi-source data; data fusion architecture can be replaced by an edge-cloud two-level fusion architecture or a cloud-edge-device reverse scheduling fusion architecture, with the core still being multi-source collaboration to solve the problem of data silos.

[0040] In this way, through the collaborative processing of edge nodes and cloud platforms, the original vehicle-side and roadside data are transformed into high-quality, unambiguous new event attribute information, providing a reliable data foundation for subsequent regulation.

[0041] S102. Based on the attribute information, the current environmental parameters of the environment where the new event occurs, and the road network characteristics of the road segment where the new event is located, calculate the influence domain of the new event at each future time.

[0042] In this step, this application abandons the traditional fixed-radius division method and adopts a spatiotemporal attenuation factor algorithm to dynamically calculate the influence domain. The algorithm incorporates the following factors: Event type and severity: The initial impact range is determined by the type and severity level in the attribute information of the newly added event. The more severe the event, the larger the initial impact range.

[0043] Current environmental parameters, including rain and snow conditions in frozen regions, low nighttime light levels, and road surface friction coefficient, affect the speed at which congestion spreads and dissipates. Under severe weather conditions, congestion spreads faster and dissipates more slowly, thus affecting a wider area and lasting longer.

[0044] Road network characteristics, including geometric parameters such as curve curvature and gradient, influence the pattern of congestion spreading along roads. On sharp curves and steep slopes, the impact extends further along the road due to limited visibility or increased difficulty in vehicle handling.

[0045] By using the spatiotemporal decay factor algorithm, taking into account the above factors, the influence domain boundary that changes over time is calculated in real time, and the output is the set of influence road segments extending along the road network topology and the effective time window.

[0046] As another example, the influence domain partitioning algorithm can be replaced by an improved Gaussian diffusion model or a traffic wave theory deduction model, that is, dynamically simulating the congestion diffusion trend through three dimensions: the intensity of the event core point, road network connectivity, and traffic flow density. The core logic is still dynamic boundary partitioning based on multiple features.

[0047] In this way, the influence domain is no longer a static geometric circle, but a spatiotemporal region that dynamically adjusts with the evolution of events, changes in the environment, and road network characteristics, providing a precise spatial benchmark for targeted vehicle screening.

[0048] S103. Based on the spatial distance between the current position of each vehicle and the boundary of the influence domain, candidate vehicles are determined. Then, from all candidate vehicles, vehicles whose predicted trajectories will enter the influence domain in future moments are selected and identified as target vehicles affected by the new event.

[0049] This step achieves a two-layer screening of target vehicles: Spatial filtering: Based on the current position in the vehicle real-time status table, calculate the spatial distance between each vehicle and the boundary of the influence domain, and mark vehicles whose distance is less than the preset spatial threshold as spatial candidate vehicles.

[0050] Time-based filtering: For each candidate vehicle in space, extract its historical trajectory sequence (past 30s), input it into a pre-trained LSTM trajectory prediction model, and output the predicted trajectory points within a preset future time period; determine whether the predicted trajectory intersects with the influence domain in space and time (i.e., whether there is a future moment when the vehicle's position falls into the influence domain). If so, identify the vehicle as the target vehicle and record the expected entry time and position.

[0051] As another example, the trajectory prediction algorithm can be replaced by the Transformer time-series prediction model (which is good at capturing long-distance dependencies) or the Kalman filter and heuristic rule combination model (which has low computational cost and is suitable for rapid inference), while the core is still to realize the function of predicting the future driving state of the vehicle.

[0052] In this way, by introducing trajectory prediction, only vehicles that are certain to be affected in the future can be screened out, avoiding misjudging vehicles that only pass through the boundary without entering the core area as control targets, thus achieving precise targeting.

[0053] In severe weather conditions, this application also includes steps to enhance perception of complex environments in seasonally frozen areas to improve the quality of roadside data: The system determines whether the weather is severe based on current environmental parameters. If so, it preprocesses the original images from the roadside sensing devices. This preprocessing includes: using an attention-based rain line separation network to eliminate raindrop interference, and using low-light enhancement technology to improve nighttime sensing accuracy. The preprocessed original images are then input into a multi-scale event recognition model. This model uses a "feature sharing-branch decision" deep learning framework, combined with optical flow to extract motion trajectories and contour matching, to accurately identify roadside events such as traffic accidents and pedestrian intrusions. The identified roadside events are used as part of the roadside data to determine the attribute information of newly added events in S101.

[0054] In addition, the multi-scale event recognition model is compressed into a lightweight model through knowledge distillation technology and deployed on roadside edge nodes to adapt to edge computing power requirements and ensure low-latency response.

[0055] As another example, severe weather image processing can be replaced by traditional morphological filtering and gamma correction (traditional algorithm) or generative adversarial network denoising model (cutting-edge algorithm); model lightweighting can be replaced by model pruning (structured / unstructured) or quantization compression (8-bit / 4-bit quantization).

[0056] In this way, the system can still maintain high-precision perception in complex environments such as rain, snow, and night in the frozen zone, providing a reliable roadside data source for subsequent integration and control.

[0057] S104. Based on the vehicle identifier of each target vehicle, push the corresponding control command to each target vehicle.

[0058] This application breaks away from the "one-size-fits-all" service model, providing customized strategies for three core vehicle categories. Vehicle identification includes identification for private vehicles, commercial vehicles, and emergency vehicles.

[0059] The control instructions pushed to target vehicles with social vehicle identification include: Personalized route optimization and service area recommendations are generated based on real-time traffic conditions, tourism resources, and weather data; and / or blind spot warning instructions are generated based on at least one of the following warning methods: text pop-ups, voice reminders, or audible and visual alarms. The blind spot tiered warning system dynamically adjusts the warning intensity (progressing from light to heavy) based on the relative position of the vehicle and the blind spot risk and the time of collision, avoiding excessive disturbance or insufficient warning.

[0060] The control instructions pushed to target vehicles with the vehicle identification number of commercial vehicles include: Operational guidance instructions based on order hotspot push and optimal route planning generated from cloud-based order heatmaps; and / or safety prompts based on roadside perception data to dynamically avoid curves and blind spots of large vehicles, in order to improve the efficiency and safety of operating vehicles.

[0061] The control commands pushed to target vehicles identified as emergency vehicles include: Based on the priority passage instructions generated by the avoidance guidance assessment model and directed to the vehicles ahead, the model generates precise avoidance instructions based on the emergency vehicle's path and the position of the vehicles ahead; and / or, based on the path guidance instructions generated by the roadside guidance screen linkage strategy for emergency dedicated lane planning and display, the roadside guidance screen is linked to physically delineate a dedicated lane to ensure priority passage for emergency vehicles.

[0062] As another example, the vehicle classification dimension can be replaced by a classification method based on power type (fuel / electric) and usage scenario (commuting / long-distance) to adapt to the management needs of new energy vehicles; the emergency vehicle guidance method can be replaced by vehicle-road cooperative signal priority (green wave), which achieves priority passage by controlling traffic lights.

[0063] In this way, by pushing differentiated instructions, the core needs of different vehicle types can be met, user cooperation and instruction execution rates can be improved, and a breakthrough from "wide-area broadcasting" to "precise targeting" can be achieved.

[0064] In addition, this application provides another example of communication technology, in which the targeted push technology can be replaced by 5G-V2X targeted communication (lower latency) or Beidou short message targeted transmission (covering areas without base stations), while the core is still to meet the need to push instructions only to the target vehicle.

[0065] Compared with the prior art, the embodiments of this application achieve the following technical effects through the above technical solution: 1. Through comprehensive blind spot level warning, accurate perception in complex environments, and emergency vehicle guidance, the system effectively avoids accidents caused by blind spots and severe weather, significantly reducing the traffic accident rate. At the same time, in rainy and snowy weather in frozen areas, the system can quickly identify traffic events and push out warnings, allowing drivers sufficient reaction time and greatly improving the timeliness of hazard response.

[0066] 2. Dynamic targeted control avoids ineffective detours, reduces road network redundancy, and thus improves road network capacity; through dedicated lane planning and avoidance guidance, the passage time of emergency vehicles is significantly shortened; differentiated instructions are precisely adapted to the needs of different vehicles, and compared with indiscriminate broadcasting, the driver's instruction adoption rate and response rate are significantly improved, and the execution rate of control instructions is improved.

[0067] 3. It has strong adaptability to complex environments, completely solving the problem of perception failure in rain, snow and nighttime scenarios in frozen areas. The event recognition accuracy meets the requirements of practical applications. Through semantic collaboration and conflict resolution, it eliminates semantic conflicts of multi-source data, greatly improves data utilization, and provides high-quality data support for subsequent optimization.

[0068] Based on the same inventive concept, this application also provides a traffic targeted control device based on vehicle-road-cloud multi-source information fusion, which corresponds to the traffic targeted control method based on vehicle-road-cloud multi-source information fusion. Since the principle of the device in this application is similar to the traffic targeted control method based on vehicle-road-cloud multi-source information fusion described above in this application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0069] Please see Figure 3 , Figure 3 This is a schematic diagram of the traffic targeted control device based on vehicle-road-cloud multi-source information fusion provided in an embodiment of this application. Figure 3 As shown, the traffic targeted control device 300 based on vehicle-road-cloud multi-source information fusion includes: The vehicle-road-cloud semantic collaborative fusion module 301 is used to determine the attribute information of newly added events based on vehicle-side data, roadside data, and cloud data.

[0070] The influence domain dynamic partitioning module 302 is used to calculate the influence domain of the new event at each future time based on the attribute information, the current environmental parameters of the environment in which the new event occurs, and the road network characteristics of the road segment where the new event is located.

[0071] The dynamic targeting control module 303 is used to determine candidate vehicles based on the spatial distance between the current position of each vehicle and the boundary of the influence domain, and then select from all candidate vehicles all vehicles whose predicted trajectories will enter the influence domain in future moments, and determine them as target vehicles affected by the new event.

[0072] The multi-type vehicle classification and empowerment module 304 is used to push corresponding control instructions to each target vehicle based on the vehicle identifier of each target vehicle.

[0073] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 4 As shown, the electronic device 400 includes a processor 410, a memory 420, and a bus 430.

[0074] The memory 420 stores machine-readable instructions that can be executed by the processor 410. When the electronic device 400 is running, the processor 410 and the memory 420 communicate via the bus 430. When the machine-readable instructions are executed by the processor 410, the steps of the method described above can be performed. For specific implementation details, please refer to the method embodiment, which will not be repeated here.

[0075] This application also provides a computer-readable storage medium storing a computer program. When the computer program is run by a processor, it can execute the steps of the method described above. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0076] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0077] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, 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 shown 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.

[0078] 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.

[0079] In addition, the functional units in the various embodiments of 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.

[0080] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, 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.

[0081] 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 scope of protection 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. Such 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, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A traffic-targeted control method based on vehicle-road-cloud multi-source information fusion, characterized in that, include: Based on vehicle-side data, roadside data, and cloud data, determine the attribute information of newly added events; Based on the attribute information, the current environmental parameters of the environment in which the new event occurs, and the road network characteristics of the road segment where the new event is located, the influence domain of the new event at each future time is calculated. Candidate vehicles are determined based on the spatial distance between the current position of each vehicle and the boundary of the influence domain. Then, from all candidate vehicles, vehicles whose predicted trajectories will enter the influence domain in future moments are selected and identified as target vehicles affected by the new event. Based on the vehicle identifier of each target vehicle, corresponding control instructions are pushed to each target vehicle.

2. The method according to claim 1, characterized in that, The attribute information of newly added events can be determined in the following ways: Acquire the vehicle-side data and roadside data that conform to a unified semantic specification, and upload the vehicle-side data and roadside data to the corresponding edge nodes; Each edge node performs spatiotemporal registration of the received vehicle-side data and roadside data, and performs complementary verification based on the registered vehicle-side data and roadside data to identify mutually corroborating data pairs as candidate events. The cloud platform uses a spatiotemporal correlation analysis model to perform correlation analysis on candidate events reported by multiple edge nodes. By combining the weighted fusion of roadside device credibility weight and vehicle user credit level, conflicts are resolved and fused among multiple data sources pointing to the same real event, so as to obtain the attribute information of the newly added event.

3. The method according to claim 2, characterized in that, The vehicle-side data and the roadside data are spatiotemporally registered using the following method: The vehicle-side data and the roadside data are synchronized in time so that they can be compared on the same time reference. The vehicle-side data and the roadside data after time synchronization are transformed into map coordinates so that the vehicle-side data and the roadside data can be compared in the same spatial coordinate system.

4. The method according to claim 1, characterized in that, Also includes: When the current environmental parameters indicate severe weather, the original image is preprocessed, including eliminating raindrop interference and improving nighttime perception accuracy. The preprocessed original image is input into a multi-scale event recognition model to extract motion trajectories and contour matching in order to identify roadside events under severe weather conditions.

5. The method according to claim 1, characterized in that, The vehicle identification includes identification for civilian vehicles. Among them, the control instructions pushed to target vehicles with social vehicle identification include: Personalized route optimization and service area recommendations generated based on real-time traffic conditions, tourism resources, and weather data; and / or blind spot warning instructions generated based on at least one of the following warning methods: text pop-ups, voice reminders, or audible and visual alarms.

6. The method according to claim 1, characterized in that, The vehicle identification includes identification for commercial vehicles. Among them, the control instructions pushed to target vehicles with the vehicle identification number of commercial vehicles include: Safety prompts based on order hotspot pushes and optimal route planning generated from cloud-based order heatmaps, and / or safety prompts based on dynamic avoidance suggestions for curves and large vehicle blind spots generated from roadside perception data.

7. The method according to claim 1, characterized in that, The vehicle markings include emergency vehicle markings. The control instructions pushed to targeted vehicles identified as emergency vehicles include: Priority passage instructions generated by the avoidance guidance assessment model and directed to other vehicles ahead, and / or route guidance instructions generated and displayed based on the roadside guidance screen linkage strategy for emergency dedicated lane planning.

8. A traffic targeted control device based on vehicle-road-cloud multi-source information fusion, characterized in that, include: The vehicle-road-cloud semantic collaboration and fusion module is used to determine the attribute information of newly added events based on vehicle-side data, roadside data, and cloud data. The influence domain dynamic partitioning module is used to calculate the influence domain of the new event at each future time based on the attribute information, the current environmental parameters of the environment in which the new event occurs, and the road network characteristics of the road segment where the new event is located. The dynamic targeting control module is used to determine candidate vehicles based on the spatial distance between the current position of each vehicle and the boundary of the influence domain, and then filter out all vehicles whose predicted trajectories will enter the influence domain in future moments from all candidate vehicles, and determine them as the target vehicles affected by the new event. The multi-type vehicle classification and empowerment module is used to push corresponding control instructions to each target vehicle based on the vehicle identifier of each target vehicle.

9. An electronic device, characterized in that, include: The 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 steps of the method as described in any one of claims 1 to 7.

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 steps of the method as described in any one of claims 1 to 7.

Images