Traffic adaptive control method and system based on vehicle state perception and intelligent road network

Abstract

The invention relates to the technical field of traffic control, and provides a traffic adaptive control method and system based on vehicle state perception and an intelligent road network. The method comprises the steps that a vehicle state sensing module is installed on a target vehicle, and a vehicle state sensing multi-dimensional data stream is collected; an intelligent road network module is adopted to monitor a road network multi-dimensional data flow; carrying out fusion modeling on the vehicle state perception multi-dimensional data flow and the road network multi-dimensional data flow to generate a dynamic traffic scene model; traffic operation trend parameters are predicted based on a dynamic traffic scene model, and driving dynamic decision and traffic adaptive control are performed on a target vehicle through the parameters, so that the problem that the state of the target vehicle and the road network condition cannot be acquired and accurately analyzed in real time in the existing traffic control technology is solved. The technical problem that traffic flow and traffic adaptive control effects are poor due to the fact that vehicle state sensing and intelligent road network fusion modeling are solved, and the technical effects that accurate prediction and control of traffic flow and vehicle dynamic states are achieved through vehicle state sensing and intelligent road network fusion modeling, so that traffic management is optimized, and driving safety is improved are achieved.

Description

Technical Field

[0001] This application relates to the field of traffic control technology, specifically to a traffic adaptive control method and system based on vehicle status perception and intelligent road networks. Background Technology

[0002] With accelerated urbanization and surging traffic demand, traditional traffic control systems face the dual challenges of insufficient dynamic scene perception and weak multi-source data collaborative modeling capabilities. Existing signal control systems largely rely on fixed timing strategies and local road segment sensor data, making it difficult to capture in real-time the complex coupling relationship between vehicle road contact states, micro-level driving behavior, and macro-level road network operation. This results in control strategies lagging behind sudden changes in traffic flow, exacerbating congestion and safety hazards. Meanwhile, heterogeneous data generated by single vehicle sensors (such as onboard radar and cameras) and roadside intelligent units (such as traffic flow monitors and traffic light controllers) have long been fragmented, lacking a spatiotemporally synchronized fusion mechanism, making it impossible to construct high-fidelity dynamic traffic scene models. Therefore, how to effectively integrate vehicle-state perception data with intelligent road network data to improve the accuracy and intelligence level of adaptive traffic control has become a critical issue urgently needing to be addressed in the current transportation field. Summary of the Invention

[0003] This application provides a traffic adaptive control method and system based on vehicle status perception and intelligent road network, aiming to solve the technical problem in existing traffic control technologies that cannot acquire and accurately analyze the target vehicle status and road network conditions in real time, resulting in poor traffic flow and traffic adaptive control effects.

[0004] The first aspect disclosed in this application provides a traffic adaptive control method based on vehicle status perception and intelligent road network. The method includes: installing a vehicle status perception module on a target vehicle; acquiring a multi-dimensional vehicle status perception data stream through the vehicle status perception module, the multi-dimensional data stream including vehicle-road contact state, vehicle-road radar data, and video data; monitoring and acquiring a multi-dimensional road network data stream using an intelligent road network module, the multi-dimensional data stream including road feature data, traffic flow, and traffic light timing; fusing and modeling the multi-dimensional vehicle status perception data stream and the multi-dimensional road network data stream to generate a dynamic traffic scenario model; predicting traffic operation trend parameters based on the dynamic traffic scenario model; and performing driving dynamic decision-making and traffic adaptive control on the target vehicle using the traffic operation trend parameters.

[0005] Another aspect of this application discloses a traffic adaptive control system based on vehicle status perception and intelligent road network. The system includes: a vehicle status data acquisition unit: a vehicle status perception module installed on the target vehicle, which acquires a multi-dimensional vehicle status perception data stream, including vehicle road surface contact state, vehicle and road radar data, and video data; a road network data acquisition unit: an intelligent road network module monitors and acquires a multi-dimensional road network data stream, including road feature data, traffic flow, and traffic light timing; a fusion modeling unit: a fusion modeling unit that integrates the vehicle status perception multi-dimensional data stream and the road network multi-dimensional data stream to generate a dynamic traffic scene model; and a traffic control unit: a traffic operation trend parameter predicted based on the dynamic traffic scene model, which is used to make driving dynamic decisions and perform traffic adaptive control on the target vehicle.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] The aforementioned traffic adaptive control method based on vehicle state perception and intelligent road networks first installs a vehicle state perception module on the target vehicle to collect real-time data on the vehicle's road contact state, radar data, and video data, forming a multi-dimensional data stream for vehicle state perception. Simultaneously, an intelligent road network module acquires a multi-dimensional data stream of road network information, including road features, traffic flow, and traffic light timing. These two data streams are then fused and modeled to construct a dynamic traffic scenario model. Based on this model, traffic operation trend parameters are predicted, and these parameters are used to make dynamic driving decisions and adaptive control for the target vehicle, achieving intelligent traffic management.

[0008] The above description is merely an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart illustrating a traffic adaptive control method based on vehicle status perception and intelligent road network in one embodiment.

[0011] Figure 2This is an architecture diagram of a traffic adaptive control system based on vehicle status perception and intelligent road network in one embodiment.

[0012] Figure labeling: Vehicle status data acquisition unit 11, road network data acquisition unit 12, fusion modeling unit 13, traffic control unit 14. Detailed Implementation

[0013] This application provides a traffic adaptive control method and system based on vehicle status perception and intelligent road network, which solves the technical problem in existing traffic control technologies that cannot acquire and accurately analyze the status of target vehicles and road network conditions in real time, resulting in poor traffic flow and traffic adaptive control effects.

[0014] 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 a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0015] It should be noted that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to such process, method, product, or device.

[0016] Example 1, as Figure 1 As shown, this application provides a traffic adaptive control method based on vehicle state perception and intelligent road network, the method comprising:

[0017] A vehicle status perception module is installed on the target vehicle, and a multi-dimensional vehicle status perception data stream is collected through the vehicle status perception module. The multi-dimensional vehicle status perception data stream includes vehicle road contact status, vehicle and road radar data, and video data.

[0018] In this embodiment, a vehicle status perception module is installed on the target vehicle. This module collects multi-dimensional vehicle status perception data streams in real time through various sensors and devices. These multi-dimensional vehicle status perception data streams include vehicle road contact state, vehicle-road radar data, and video data. The vehicle road contact state includes tire-ground friction, road surface conditions (such as wet / slippery or dry), and wheel movement. The vehicle-road radar data includes the relative position and distance information between the vehicle and the road obtained through radar technology, which can help monitor the surrounding environment and road features. The video data includes real-time video data of the road and surrounding environment collected using onboard cameras or video surveillance equipment, used to determine traffic conditions, identify obstacles or other road users. This data can provide accurate information support for subsequent driving decisions and traffic control, helping to optimize traffic management and improve driving safety.

[0019] The intelligent road network module is used to monitor and acquire multi-dimensional data streams of the road network, which include road characteristic data, traffic flow, and traffic light timing.

[0020] In one embodiment, the intelligent road network module collects various types of data from the road network in real time through various sensors and monitoring systems, constructing a multi-dimensional road network data stream. This multi-dimensional data stream includes road characteristic data, traffic flow, and traffic light timing. Road characteristic data reflects the basic characteristics of roads, such as structure, length, width, gradient, and number of lanes. Traffic flow data includes the number of vehicles, travel speed, and traffic density on the road segment, providing real-time information about the current traffic situation. Traffic light timing data records the switching cycle and timing scheme of traffic lights, helping to analyze the traffic control effectiveness of the road segment. Integrating this information provides a comprehensive and real-time road network operating status, offering crucial information for traffic management and control.

[0021] The vehicle-state perception multidimensional data stream and the road network multidimensional data stream are fused and modeled to generate a dynamic traffic scene model.

[0022] In one embodiment, after obtaining the multi-dimensional data streams of vehicle-state perception and road network, these data are preprocessed, i.e., cleaned and sorted. After preprocessing, vehicle-state correlation features and road network correlation features are extracted from these standard data, and then spatial correlation is performed on these correlation features to complete the fusion of these two types of features. Subsequently, the fused feature data is used to train a neural network model. Through steps such as forward propagation, loss function, backpropagation, and parameter optimization, the neural network model is gradually optimized until the accuracy on the validation data meets the preset accuracy, thereby generating a dynamic traffic scenario model. This dynamic traffic scenario model can accurately reflect the current traffic environment, dynamically simulate changes in traffic flow and vehicle behavior, thereby providing support for subsequent traffic prediction and intelligent decision-making.

[0023] Furthermore, this application provides the aforementioned dynamic traffic scene model, including:

[0024] The vehicle-state perception multidimensional data stream and the road network multidimensional data stream are cleaned to obtain a standard vehicle-state perception multidimensional data stream and a standard road network multidimensional data stream. The standard vehicle-state perception multidimensional data stream and the standard road network multidimensional data stream are arranged in time sequence to obtain a vehicle-state perception multidimensional sequence data stream and a road network multidimensional sequence data stream. Based on the vehicle-state perception multidimensional sequence data stream and the road network multidimensional sequence data stream, a fusion model is performed to construct a dynamic traffic scenario model.

[0025] Preferably, the raw data streams collected by the vehicle status perception module and the intelligent road network module often suffer from noise, packet loss, redundancy, or inconsistent formats. Therefore, data cleaning is required for the collected vehicle status perception multidimensional data streams and road network multidimensional data streams. The data cleaning process includes imputing missing values ​​(e.g., through time interpolation or historical averages), identifying and removing outliers (e.g., data points exceeding physical thresholds), and standardizing the format and units of various data types (e.g., maximum-minimum normalization) to ensure they conform to unified processing standards. After cleaning, a standard vehicle status perception multidimensional data stream and a standard road network multidimensional data stream with complete structure, low error, and direct applicability for analysis are obtained. Subsequently, to reflect the dynamic characteristics of traffic evolution over time, the cleaned standard vehicle status data and standard road network data are time-series processed according to the time axis, generating time-ordered vehicle status perception multidimensional sequence data streams and road network multidimensional sequence data streams. During this process, timestamp alignment is required to ensure that the vehicle status data and road network data remain synchronized within the same time window, thus supporting subsequent joint analysis. Subsequently, features are extracted from the obtained multi-dimensional sequence data streams of vehicle state perception and road network, respectively. The extracted features are then categorized and summarized to form vehicle state-related feature sets and road network-related feature sets. By spatially fusing these two related feature sets, a dynamic traffic scenario model is trained. This dynamic traffic scenario model can be modeled using time series modeling methods, thereby combining the changing trends of vehicle behavior with road environment characteristics to uncover spatiotemporal correlations, such as the linkage between vehicle speed changes and traffic light states and traffic flow density. This forms a predictable and continuously evolving traffic scenario, providing a data foundation and structural support for subsequent trend prediction and control decisions.

[0026] Furthermore, this application provides the construction of the dynamic traffic scenario model, including:

[0027] Feature extraction is performed on the vehicle-state perception multidimensional sequence data stream and the road network multidimensional sequence data stream to obtain vehicle-state association feature set and road network association feature set; the vehicle-state association feature set and the road network feature set are spatially associated and fused to obtain traffic scene fusion feature set; dynamic scene modeling is performed on the traffic scene fusion feature set according to the traffic prediction target to obtain dynamic traffic scene model.

[0028] Preferably, after completing the time-series processing of vehicle state perception and road network multidimensional data, representative features are extracted from the vehicle state perception multidimensional sequence data stream and the road network multidimensional sequence data stream according to preset feature indicators. For the vehicle state perception multidimensional sequence data stream, the extracted content includes, but is not limited to, vehicle speed change trend, vehicle acceleration and deceleration mode, steering angle, wheel force state, distance and relative speed of obstacles sensed by radar, lane departure or lane change behavior in video images, etc., which form a vehicle state associated feature set. For the road network multidimensional sequence data stream, the extracted content includes, but is not limited to, road structure parameters (such as lane width, slope, road segment type), traffic density changes, traffic light cycle status, traffic event labels (such as congestion, accidents), etc., which form a road network associated feature set. Subsequently, the spatial coordinates of the vehicle within the road network are established based on vehicle positioning information (such as GPS coordinates or inertial navigation system). Combined with the geographic coordinate system of the road network map, the real-time location of the vehicle at a specific road segment or intersection is determined. This allows for spatial matching of the vehicle-state associated feature set with the road network associated feature set, generating a traffic scene fusion feature set. This feature set contains the spatiotemporal correspondence between the vehicle's state and its environment. Next, the fusion feature set is modeled according to preset traffic prediction objectives (such as predicting the traffic density of a road segment in the next 10 minutes, determining whether a vehicle needs to change lanes, and predicting potential traffic conflicts). Neural network models, such as multi-task learning models and spatiotemporal convolutional networks, can be introduced during the modeling process to capture the dynamic dependencies between features. Finally, a dynamic traffic scene model is constructed. This model has real-time update capabilities, continuously adjusting its understanding and prediction of traffic conditions as input data changes, thereby providing more accurate dynamic driving decision support for the vehicle.

[0029] Furthermore, this application provides the aforementioned traffic scene fusion feature set, including:

[0030] Based on the vehicle state association feature set and the road network association feature set, determine the vehicle spatial coordinate set and the road network map coordinate set; based on the vehicle spatial coordinate set and the road network map coordinate set, spatially associate the vehicle state association feature set and the road network feature set to generate a traffic association topology network; according to the traffic association topology network, perform feature fusion on the vehicle state association feature set and the road network feature set to obtain the traffic scene fusion feature set.

[0031] Optionally, after extracting the vehicle status association feature set and the road network association feature set, the real-time location data of the vehicle in the road network is extracted using a vehicle positioning system (such as GPS or high-precision inertial navigation) to construct a vehicle spatial coordinate set. Then, information such as road nodes, road segment numbers, and intersection structures are obtained using existing digital road network maps to establish a road network map coordinate set. Subsequently, based on the spatial positional relationship between the vehicle spatial coordinate set and the road network map coordinate set, the vehicle status association feature set and the road network association feature set are spatially aligned and matched. This spatial association process can be achieved through coordinate matching, map matching algorithms (such as HMM Map Matching), or topological mapping based on road segment IDs, thereby determining which specific road segment, lane, or intersection node the vehicle's status belongs to in the road network at a certain moment, and further determining the relative positional relationship between the vehicle and the surrounding road structure and signal facilities. After completing spatial registration, a traffic association topology network is constructed. This network describes the association structure between vehicles and roads in the form of nodes and edges. Nodes represent spatial points where the vehicle's current position is relative to its position in the road network, while edges describe the vehicle's trajectory on the road segment, lane change possibilities, or interaction relationships with other traffic elements. Subsequently, based on the constructed traffic association topology network, vehicle status features (such as vehicle speed, acceleration, and braking status) are fused with corresponding road network features (such as traffic light status, lane speed limit, and traffic density) through feature splicing to generate a traffic scene fusion feature set with spatiotemporal correlation. This fusion feature set comprehensively reflects the dynamic behavior basis of vehicles in the current traffic environment and the road network response background, providing high-quality input data for subsequent traffic state modeling and prediction.

[0032] Furthermore, this application provides the aforementioned method for obtaining a dynamic traffic scene model, including:

[0033] The traffic prediction objective is decomposed to obtain a set of sub-prediction objectives; a set of sub-objective neural networks is selected based on the set of sub-prediction objectives; the set of sub-objective neural networks is used to predict, train, and model the traffic scene fusion feature set to obtain the dynamic traffic scene model.

[0034] Optionally, to achieve accurate prediction of complex traffic conditions, the overall traffic prediction objective is first broken down into a set of operable sub-prediction objectives. For example, if the overall objective is to improve the driving safety and traffic efficiency of target vehicles, it can be broken down into multiple sub-objectives, such as predicting the traffic flow trend of the road ahead, judging the traffic risk at intersections, estimating vehicle lane-changing needs, and identifying potential traffic conflicts. These sub-prediction objectives focus on different dimensions of traffic scenario evolution, which helps to refine the modeling task. After determining the sub-prediction objectives, a suitable set of sub-objective neural network models is selected for different objectives. Specifically, for time series predictions (such as traffic flow changes), recurrent neural network models such as LSTM or GRU can be used; for spatial distribution feature learning (such as road congestion area identification), convolutional neural networks (CNN) or graph neural networks (GNN) can be used; for comprehensive judgment tasks (such as dangerous behavior identification), attention mechanisms or multi-task learning structures can be combined. Each sub-neural network model is designed and trained for a specific type of feature and output objective. The training process usually includes forward propagation, loss calculation, backpropagation, parameter optimization, and other steps. Subsequently, the fused feature set of historical traffic scenarios and corresponding labels are used as training inputs and fed into the aforementioned sub-target neural network sets for model training. During training, the model continuously optimizes its parameters to fit the feature patterns of specific sub-targets, thereby improving prediction accuracy. After training, each sub-model can output its prediction results. These sub-models are then integrated through a pre-defined integration mechanism (such as weighted aggregation or decision-level fusion) to generate a complete dynamic traffic scenario model. This model can respond synchronously across multiple dimensions and targets, achieving comprehensive prediction and assessment of future traffic conditions, and providing vehicles with high-precision driving assistance and traffic adaptive control strategies.

[0035] Traffic operation trend parameters are predicted based on the dynamic traffic scenario model, and driving dynamic decision-making and traffic adaptive control of the target vehicle are performed based on the traffic operation trend parameters.

[0036] In one embodiment, after constructing a dynamic traffic scenario model, the model is used to predict traffic conditions in the near future, generating a set of traffic trend parameters. These parameters include changes in congestion levels on the road ahead of the target vehicle in the next few seconds, estimated traffic light switching times, movement trends of surrounding vehicles (such as acceleration, deceleration, or lane-changing tendencies), potential traffic conflict risk levels, and recommended lane-changing opportunities. These parameters reflect the evolution of the traffic environment over time and represent a comprehensive quantitative expression of future traffic conditions. After obtaining these traffic trend parameters, dynamic driving decisions for the target vehicle are made based on these parameters, generating a set of driving strategy parameters. Furthermore, considering factors such as traffic rules and safe distances, adaptive adjustments are made to traffic signal control, adaptive speed limits, or cooperative path planning, generating a set of traffic control parameters. Finally, based on the generated driving strategy parameters and traffic control parameters, dynamic driving decisions and adaptive traffic control are implemented for the target vehicle, enabling automated driving assistance functions or prompting the driver to make timely adjustments, thereby improving the intelligence and accuracy of traffic response.

[0037] Furthermore, this application provides the method of performing driving dynamic decision-making and traffic adaptive control on the target vehicle using the traffic operation trend parameters, including:

[0038] Based on the traffic operation trend parameters, driving decision analysis is performed on the target vehicle to determine the target driving strategy parameters; the traffic operation trend parameters are then analyzed for control according to the traffic safety operation objectives to obtain the target traffic control parameters; and driving dynamic decision-making and traffic adaptive control are performed on the target vehicle based on the target driving strategy parameters and the target traffic control parameters.

[0039] Preferably, after acquiring traffic operation trend parameters, the system comprehensively evaluates parameters such as congestion level, traffic light switching time, movement trends of surrounding vehicles, potential traffic conflict risks, and recommended lane changes. Using a built-in vehicle traffic driving strategy library, it selects a matching driving plan to determine the target driving strategy parameters. For example, if the trend parameters indicate impending congestion, it may suggest slowing down or changing lanes in advance; if the traffic light is about to turn green, it may suggest maintaining the current speed to pass through the intersection smoothly. Simultaneously, based on the overall traffic safety operation goals, the system calls upon a built-in traffic control strategy library. This library stores various traffic control strategies that conform to the traffic safety operation goals. Each control strategy corresponds to a set of feasibility condition parameter ranges. These parameter ranges define the typical traffic environment in which the strategy is applicable, such as traffic density, congestion level, traffic light status, probability of traffic conflict, and speed limit level. Subsequently, these conditional parameters are parsed from the traffic operation trend parameters and compared with the feasibility conditional parameter ranges of each strategy in the traffic control strategy library. This comparison process can employ methods such as interval matching and similarity scoring to identify the control strategy that best matches the current traffic trend. Once the optimal control strategy is matched, it is extracted from the traffic control strategy library, and the traffic control parameters contained within this strategy are output as target traffic control parameters. These target traffic control parameters include, but are not limited to, the suggested speed limit for the current road segment, the start and end times of the passage window, recommended lane change / restriction instructions, traffic light response strategies, and lane guidance numbers. These control parameters will guide subsequent traffic cooperative control processes, ensuring that vehicle behavior strictly adheres to road safety requirements and remains consistent with the overall traffic network's dynamic scheduling strategy. It is important to note that the parsing of these target traffic control parameters and subsequent adaptive traffic control only occur when the congestion level reaches the corresponding minimum threshold for traffic control changes. Finally, the target driving strategy parameters are sent to the target vehicle's autonomous driving controller, which directly distributes them to the steering, braking, and acceleration systems, or converts them into prompts on the human-machine interface to assist the human driver in making decisions. Upon triggering a traffic control change command, the parsed target traffic control parameters are also sent to the traffic control system to assist staff in adaptive traffic control, thereby achieving safe, efficient, and intelligent traffic management.

[0040] Furthermore, this application provides the parameters for determining the target driving strategy, including:

[0041] Based on the target vehicle, a vehicle traffic driving strategy library is constructed; based on the vehicle traffic driving strategy library, driving decision analysis is performed on the traffic operation trend parameters to determine the target driving strategy parameters.

[0042] Optionally, to achieve personalized and dynamic driving decisions for the target vehicle, a dedicated vehicle traffic driving strategy library is constructed based on the target vehicle's type (e.g., small car, truck, bus) and operational characteristics (e.g., acceleration, turning radius, braking distance). This library contains optional strategies for the driving behavior of this type of vehicle under different traffic environments. After the strategy library is constructed, parameters such as the target vehicle's current relative distance to the vehicle in front, the predicted time window for reaching the traffic light, the direction of surrounding traffic flow, the road congestion level, and the lane change opportunity window are extracted from traffic trend parameters. These parameters are then compared with the applicable condition range associated with each driving strategy in the vehicle traffic driving strategy library to select the most suitable vehicle traffic driving strategy. The behavioral control parameters defined by this strategy, such as suggested acceleration value, maintaining the current lane, target following distance, and expected lane change time, are extracted and output as target driving strategy parameters to guide the vehicle's actual driving behavior, ensuring that its driving behavior is both consistent with its own characteristics and fully adapted to the current dynamic traffic environment.

[0043] In summary, the embodiments of this application have at least the following technical effects:

[0044] This application embodiment first installs a vehicle status perception module on the target vehicle, and collects a multi-dimensional vehicle status perception data stream through the vehicle status perception module. This multi-dimensional data stream includes vehicle-road contact state, vehicle-road radar data, and video data. Subsequently, an intelligent road network module monitors and acquires a multi-dimensional road network data stream, which includes road feature data, traffic flow, and traffic light timing. Then, the vehicle status perception data stream and the road network data stream are fused and modeled to generate a dynamic traffic scene model. Finally, traffic operation trend parameters are predicted based on the dynamic traffic scene model, and these parameters are used to make driving dynamic decisions and implement adaptive traffic control for the target vehicle. These technical effects collectively solve the technical problem in existing traffic control technologies where the real-time acquisition and accurate analysis of target vehicle status and road network conditions are impossible, leading to poor traffic flow and adaptive traffic control effects. This achieves the technical effect of optimizing traffic management and improving driving safety by using fused modeling of vehicle status perception and intelligent road networks to realize accurate prediction and control of traffic flow and vehicle dynamics.

[0045] Example 2 is based on the same inventive concept as the traffic adaptive control method based on vehicle state perception and intelligent road network in the previous examples, such as... Figure 2As shown, this application provides a traffic adaptive control system based on vehicle status perception and intelligent road network. The system includes: a vehicle status data acquisition unit 11: a vehicle status perception module is installed on the target vehicle, and the vehicle status perception module acquires a multi-dimensional data stream of vehicle status perception, which includes vehicle road surface contact state, vehicle and road radar data, and video data; a road network data acquisition unit 12: an intelligent road network module monitors and acquires a multi-dimensional data stream of road network, which includes road feature data, traffic flow, and traffic light timing; a fusion modeling unit 13: the vehicle status perception multi-dimensional data stream and the road network multi-dimensional data stream are fused and modeled to generate a dynamic traffic scene model; and a traffic control unit 14: traffic operation trend parameters are predicted based on the dynamic traffic scene model, and the target vehicle is driven dynamically and traffic adaptively controlled based on the traffic operation trend parameters.

[0046] Furthermore, the fusion modeling unit 13 is also used to perform the following method:

[0047] The vehicle-state perception multidimensional data stream and the road network multidimensional data stream are cleaned to obtain a standard vehicle-state perception multidimensional data stream and a standard road network multidimensional data stream. The standard vehicle-state perception multidimensional data stream and the standard road network multidimensional data stream are arranged in time sequence to obtain a vehicle-state perception multidimensional sequence data stream and a road network multidimensional sequence data stream. Based on the vehicle-state perception multidimensional sequence data stream and the road network multidimensional sequence data stream, a fusion model is performed to construct a dynamic traffic scenario model.

[0048] Furthermore, the fusion modeling unit 13 is also used to perform the following method:

[0049] Feature extraction is performed on the vehicle-state perception multidimensional sequence data stream and the road network multidimensional sequence data stream to obtain vehicle-state association feature set and road network association feature set; the vehicle-state association feature set and the road network feature set are spatially associated and fused to obtain traffic scene fusion feature set; dynamic scene modeling is performed on the traffic scene fusion feature set according to the traffic prediction target to obtain dynamic traffic scene model.

[0050] Furthermore, the fusion modeling unit 13 is also used to perform the following method:

[0051] Based on the vehicle state association feature set and the road network association feature set, determine the vehicle spatial coordinate set and the road network map coordinate set; based on the vehicle spatial coordinate set and the road network map coordinate set, spatially associate the vehicle state association feature set and the road network feature set to generate a traffic association topology network; according to the traffic association topology network, perform feature fusion on the vehicle state association feature set and the road network feature set to obtain the traffic scene fusion feature set.

[0052] Furthermore, the fusion modeling unit 13 is also used to perform the following method:

[0053] The traffic prediction objective is decomposed to obtain a set of sub-prediction objectives; a set of sub-objective neural networks is selected based on the set of sub-prediction objectives; the set of sub-objective neural networks is used to predict, train, and model the traffic scene fusion feature set to obtain the dynamic traffic scene model.

[0054] Furthermore, the traffic control unit 14 is also used to perform the following methods:

[0055] Based on the traffic operation trend parameters, driving decision analysis is performed on the target vehicle to determine the target driving strategy parameters; the traffic operation trend parameters are then analyzed for control according to the traffic safety operation objectives to obtain the target traffic control parameters; and driving dynamic decision-making and traffic adaptive control are performed on the target vehicle based on the target driving strategy parameters and the target traffic control parameters.

[0056] Furthermore, the traffic control unit 14 is also used to perform the following methods:

[0057] Based on the target vehicle, a vehicle traffic driving strategy library is constructed; based on the vehicle traffic driving strategy library, driving decision analysis is performed on the traffic operation trend parameters to determine the target driving strategy parameters.

[0058] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0059] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0060] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A traffic adaptive control method based on vehicle status perception and intelligent road network, characterized in that, The method includes: A vehicle status perception module is installed on the target vehicle, and a multi-dimensional vehicle status perception data stream is collected through the vehicle status perception module. The multi-dimensional vehicle status perception data stream includes vehicle road contact status, vehicle and road radar data, and video data. The intelligent road network module is used to monitor and acquire multi-dimensional data streams of the road network, which include road characteristic data, traffic flow, and traffic light timing. The vehicle-state perception multidimensional data stream and the road network multidimensional data stream are fused and modeled to generate a dynamic traffic scenario model; Traffic operation trend parameters are predicted based on the dynamic traffic scenario model, and driving dynamic decision-making and traffic adaptive control of the target vehicle are performed based on the traffic operation trend parameters.

2. The traffic adaptive control method based on vehicle status perception and intelligent road network as described in claim 1, characterized in that, The generation of the dynamic traffic scene model includes: Data cleaning is performed on the vehicle state perception multidimensional data stream and the road network multidimensional data stream to obtain standard vehicle state perception multidimensional data stream and standard road network multidimensional data stream; Arrange the standard vehicle state perception multidimensional data stream and the standard road network multidimensional data stream in time sequence to obtain the vehicle state perception multidimensional sequence data stream and the road network multidimensional sequence data stream. A dynamic traffic scenario model is constructed by fusing and modeling the vehicle-state perception multidimensional sequence data stream and the road network multidimensional sequence data stream.

3. The traffic adaptive control method based on vehicle status perception and intelligent road network as described in claim 2, characterized in that, The construction of the dynamic traffic scenario model includes: Feature extraction is performed on the vehicle state perception multidimensional sequence data stream and the road network multidimensional sequence data stream to obtain the vehicle state association feature set and the road network association feature set; The vehicle status association feature set and the road network association feature set are spatially associated and fused to obtain a traffic scene fusion feature set; Dynamic scene modeling is performed on the traffic scene fusion feature set according to the traffic prediction objective to obtain a dynamic traffic scene model.

4. The traffic adaptive control method based on vehicle status perception and intelligent road network as described in claim 3, characterized in that, The obtained traffic scene fusion feature set includes: Based on the vehicle status association feature set and the road network association feature set, determine the vehicle spatial coordinate set and the road network map coordinate set; Based on the vehicle spatial coordinate set and the road network map coordinate set, the vehicle status association feature set and the road network association feature set are spatially associated to generate a traffic association topology network; The vehicle status association feature set and the road network association feature set are fused according to the traffic association topology network to obtain the traffic scene fusion feature set.

5. The traffic adaptive control method based on vehicle status perception and intelligent road network as described in claim 3, characterized in that, The process of obtaining the dynamic traffic scene model includes: The traffic prediction objective is decomposed into a set of sub-prediction objectives; Based on the set of sub-prediction targets, select a set of sub-target neural networks; The dynamic traffic scene model is obtained by using the sub-target neural network set to predict, train, and model the traffic scene fusion feature set.

6. The traffic adaptive control method based on vehicle status perception and intelligent road network as described in claim 1, characterized in that, The step of performing driving dynamic decision-making and traffic adaptive control on the target vehicle using the traffic operation trend parameters includes: Based on the traffic operation trend parameters, a driving decision analysis is performed on the target vehicle to determine the target driving strategy parameters; The traffic operation trend parameters are analyzed and controlled according to the traffic safety operation objectives to obtain the target traffic control parameters. Based on the target driving strategy parameters and the target traffic control parameters, the target vehicle performs driving dynamic decision-making and traffic adaptive control.

7. The traffic adaptive control method based on vehicle status perception and intelligent road network as described in claim 6, characterized in that, The determination of the target driving strategy parameters includes: Based on the target vehicle, construct a vehicle traffic driving strategy library; Based on the vehicle traffic driving strategy library, driving decision analysis is performed on the traffic operation trend parameters to determine the target driving strategy parameters.

8. A traffic adaptive control system based on vehicle status perception and intelligent road network, characterized in that, The system is used to execute the traffic adaptive control method based on vehicle status perception and intelligent road network as described in any one of claims 1-7, and the system includes: Vehicle status data acquisition unit: A vehicle status perception module is installed on the target vehicle. The vehicle status perception module collects and acquires multi-dimensional vehicle status perception data streams, which include vehicle road contact status, vehicle and road radar data, and video data. Road network data acquisition unit: The intelligent road network module is used to monitor and acquire multi-dimensional road network data streams, which include road characteristic data, traffic flow and traffic light timing. Fusion Modeling Unit: Fusion modeling of the vehicle state perception multidimensional data stream and the road network multidimensional data stream to generate a dynamic traffic scene model; Traffic control unit: Based on the dynamic traffic scenario model, traffic operation trend parameters are predicted, and the target vehicle is driven dynamically and traffic adaptively controlled using the traffic operation trend parameters.

Images