A vehicle lane change influence prediction method and related device

Abstract

The embodiment of the application provides a vehicle lane-changing influence prediction method and related equipment, and belongs to the technical field of traffic control. The method comprises the following steps: acquiring traffic environment data around a lane-changing vehicle, wherein the traffic environment data comprises traffic flow data and vehicle-level data; predicting, according to the vehicle-level data, the influence of the lane-changing vehicle on other vehicles when the lane-changing vehicle changes to a target lane by using an improved car following model to obtain a vehicle influence estimation index; predicting, according to the traffic flow data, the influence of the lane-changing vehicle on traffic flow when the lane-changing vehicle changes to the target lane by using a traffic flow model to obtain a traffic flow influence estimation index; and performing lane-changing benefit calculation according to the vehicle influence estimation index and the traffic flow influence estimation index to obtain a benefit index of the lane-changing vehicle changing to the target lane. The embodiment of the application can improve the accuracy of vehicle lane-changing influence prediction and provide an accurate reference for the lane-changing behavior of the lane-changing vehicle.

Description

Technical Field

[0001] This application relates to the field of traffic control technology, and in particular to a method and related equipment for predicting the impact of vehicle lane changes. Background Technology

[0002] Research on vehicle lane-changing behavior is crucial for improving traffic efficiency and ensuring traffic safety. Vehicle lane-changing decisions are influenced by various factors, including the vehicle's own condition, information about surrounding vehicles, and traffic flow conditions. With the increasing complexity of the current traffic environment and the continuous growth in the number of vehicles, traffic flow models and car-following models have limitations in describing driver behavior and inter-vehicle interactions. On the one hand, when assessing the impact of lane-changing behavior on traffic, these technologies often only focus on the single aspect of the lane-changing vehicle's impact on surrounding vehicles, making it difficult to comprehensively and accurately predict the overall impact of lane-changing vehicles on traffic.

[0003] In summary, the technical problems existing in the relevant technologies need to be improved. Summary of the Invention

[0004] The main objective of this application is to propose a method and related equipment for predicting the impact of vehicle lane changes, which can improve the accuracy of predicting the impact of vehicle lane changes and provide accurate reference for the lane-changing behavior of vehicles.

[0005] To achieve the above objectives, one aspect of this application proposes a method for predicting the impact of vehicle lane changing, the method comprising the following steps:

[0006] Acquire traffic environment data around the lane-changing vehicle, including traffic flow data and vehicle-level data;

[0007] Based on the vehicle-level data, the impact of the lane-changing vehicle on other vehicles when it changes to the target lane is predicted by the improved car-following model, and a vehicle impact prediction index is obtained.

[0008] Based on the traffic flow data, the impact of the lane-changing vehicle changing to the target lane on the traffic flow is predicted by the traffic flow model, and a traffic flow impact prediction index is obtained.

[0009] The lane-changing benefit is calculated based on the vehicle impact prediction index and the traffic flow impact prediction index to obtain the benefit index of the lane-changing vehicle changing to the target lane.

[0010] In some embodiments, acquiring traffic environment data around the lane-changing vehicle includes:

[0011] The vehicle-level data is obtained by collecting data on itself and surrounding vehicles through the onboard sensors of the lane-changing vehicle.

[0012] Traffic flow data is obtained by collecting traffic data within a certain range centered on the lane-changing vehicle through roadside units.

[0013] In some embodiments, the car-following model includes a first car-following module and a second car-following module, and the improved car-following model is obtained through the following steps:

[0014] The fixed safety distance parameter in the original first car-following module is modified to an adaptive safety distance parameter, and the second driving trajectory parameter is modified to a parameter that varies with the second driving trajectory and the third driving trajectory, resulting in the improved first car-following module;

[0015] The fixed time response parameters regarding the first driving trajectory in the original second car-following module are modified to adaptive time response parameters to obtain the improved second car-following module;

[0016] The improved first car-following module and the improved second car-following module are combined to obtain the improved car-following model;

[0017] Wherein, the first driving trajectory is the longitudinal driving trajectory of the following vehicle in the target lane, the second driving trajectory is the longitudinal driving trajectory of the initial preceding vehicle of the following vehicle in the target lane, and the third driving trajectory is the longitudinal driving trajectory of the lane-changing vehicle.

[0018] In some embodiments, the step of predicting the impact of the lane-changing vehicle changing to the target lane on other vehicles using an improved car-following model based on the vehicle-level data, and obtaining a vehicle impact prediction index, includes:

[0019] Based on the acceleration changes of the following vehicle in the vehicle-level data and the distance between different vehicles, the improved first car-following module predicts the trajectory of the following vehicle to obtain the first predicted trajectory of the following vehicle.

[0020] Based on the real-time traffic density of the traffic flow data, the improved second car-following module is used to predict the trajectory of the following vehicle, and the second predicted trajectory of the following vehicle is obtained.

[0021] The vehicle impact prediction index is obtained by calculating based on the first and second predicted trajectories.

[0022] In some embodiments, the step of predicting the trajectory of the following vehicle using the improved first car-following module based on the acceleration changes of the following vehicle in the vehicle-level data and the distance between different vehicles, to obtain the first predicted trajectory of the following vehicle, includes:

[0023] Based on the acceleration changes of the following vehicle in the vehicle-level data, a driving style analysis is performed to obtain the driving style coefficient of the following vehicle.

[0024] Based on the driving style coefficient and the real-time traffic density of the traffic flow data, a safe distance analysis is performed to determine the distance value for the adaptive safe distance parameter;

[0025] Attention is allocated based on the longitudinal distance and longitudinal speed difference between the following vehicle and the initial leading vehicle, and a first attention coefficient of the second driving trajectory in the changing parameters is determined.

[0026] Attention is allocated based on the lateral distance, longitudinal distance, and longitudinal speed difference between the following vehicle and the lane-changing vehicle, and a second attention coefficient of the third driving trajectory is determined among the changing parameters.

[0027] The first attention coefficient and the second attention coefficient are normalized to obtain the initial preceding vehicle attention coefficient of the second driving trajectory and the lane-changing vehicle attention coefficient of the third driving trajectory.

[0028] Based on the distance value, the initial attention coefficient of the preceding vehicle, and the attention coefficient of the lane-changing vehicle, the improved first car-following module predicts the trajectory of the following vehicle to obtain the first predicted trajectory.

[0029] In some embodiments, the step of predicting the trajectory of the following vehicle using the improved second car-following module based on the real-time traffic density of the traffic flow data to obtain the second predicted trajectory of the following vehicle includes:

[0030] Response time is calculated based on the real-time traffic density to determine the time value of the adaptive time response parameter;

[0031] Based on the time value, the improved second car-following module predicts the trajectory of the following vehicle to obtain the second predicted trajectory.

[0032] In some embodiments, the step of predicting the impact of lane-changing vehicles changing to the target lane on traffic flow based on the traffic flow data using a traffic flow model, and obtaining a traffic flow impact prediction index, includes:

[0033] The traffic flow model is used to predict the density change of the lane-changing vehicles when they change lanes to the target lane, thereby obtaining the change in traffic density of the target lane.

[0034] Based on the current traffic density and the change in the target lane in the traffic flow data, the density is calculated to obtain the predicted traffic density after the lane change;

[0035] Traffic flow is calculated based on the current traffic density and the predicted traffic density to obtain the traffic flow before and after the lane change.

[0036] The traffic flow impact prediction index is obtained by calculating the traffic flow volume before and after the lane change.

[0037] In some embodiments, the step of calculating the lane-changing benefit based on the vehicle impact prediction index and the traffic flow impact prediction index to obtain the benefit index of the lane-changing vehicle changing to the target lane includes:

[0038] Obtain lane change safety indicators and lane change driving efficiency;

[0039] The benefit index is obtained by weighted summation of the vehicle impact prediction index, the traffic flow impact prediction index, the lane change safety index, and the lane change driving efficiency.

[0040] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0041] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the methods described above.

[0042] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer program product, including a computer program that, when executed by a processor, implements the aforementioned method.

[0043] The embodiments of this application include at least the following beneficial effects: This application provides a method and related equipment for predicting the impact of lane changing. This method acquires traffic environment data around the lane-changing vehicle, including traffic flow data and vehicle-level data. Based on the vehicle-level data, an improved car-following model is used to predict the impact of the lane-changing vehicle on other vehicles when it changes to the target lane, obtaining a vehicle impact prediction index. Based on the traffic flow data, a traffic flow model is used to predict the impact of the lane-changing vehicle on the traffic flow when it changes to the target lane, obtaining a traffic flow impact prediction index. The lane-changing benefit is calculated based on the vehicle impact prediction index and the traffic flow impact prediction index, obtaining a benefit index for the lane-changing vehicle changing to the target lane. This application, by separately predicting the impact of lane-changing vehicles on surrounding vehicles and the overall traffic flow, and then comprehensively evaluating the benefit of lane changing based on the obtained vehicle impact prediction index and traffic flow impact prediction index, achieves comprehensive prediction of lane-changing impact at both the micro and macro levels, improves the accuracy of lane-changing impact prediction, and provides an accurate reference for the lane-changing behavior of lane-changing vehicles. Attached Figure Description

[0044] Figure 1 This is a flowchart of the vehicle lane-changing impact prediction method provided in the embodiments of this application;

[0045] Figure 2 This is a schematic diagram of a camera and millimeter-wave radar detecting vehicles ahead, provided in an embodiment of this application.

[0046] Figure 3 This is a schematic diagram of a vehicle lane-changing scenario provided in an embodiment of this application;

[0047] Figure 4 This is a schematic diagram of the position parameters of the improved car-following model provided in the embodiments of this application;

[0048] Figure 5 This is a complete implementation flowchart of the vehicle lane change impact prediction method provided in the embodiments of this application;

[0049] Figure 6 This is a schematic diagram of the speed change curve of the vehicle behind in the target lane when the vehicle changes lanes, provided in an embodiment of this application.

[0050] Figure 7 This is a schematic diagram showing the comparison between the average speed of vehicles behind in the target lane and the standard deviation when a vehicle changes lanes, as provided in an embodiment of this application.

[0051] Figure 8 This is a box plot of lane-changing utility values ​​provided in an embodiment of this application;

[0052] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0054] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit the scope of this application.

[0055] Research on vehicle lane-changing behavior is crucial for improving traffic efficiency and ensuring traffic safety. Vehicle lane-changing decisions are influenced by various factors, including the vehicle's own condition, information about surrounding vehicles, and traffic flow conditions. With the increasing complexity of the current traffic environment and the continuous growth in the number of vehicles, traditional traffic flow models and car-following models have limitations in describing driver behavior and inter-vehicle interactions. On the one hand, existing models struggle to accurately capture the behavioral differences of drivers under different traffic densities and driving styles, failing to adapt to complex and ever-changing traffic scenarios. On the other hand, when assessing the impact of lane-changing behavior on traffic flow, existing methods often only focus on one aspect of the impact of lane-changing vehicles on surrounding vehicles or traffic flow, lacking a comprehensive framework that fully considers the impact of lane-changing vehicles on surrounding vehicles and traffic flow, making it difficult to comprehensively and accurately assess the impact of lane-changing vehicles on surrounding vehicles and overall traffic flow.

[0056] In related technologies, the classic Newell model is representative of traditional car-following models. This model posits that car-following behavior is primarily determined by response time and safe distance. During calculation, response time and safe distance are typically set to fixed values ​​that do not change with traffic conditions. Regarding data acquisition, it mainly relies on onboard sensors to obtain information such as vehicle speed and acceleration, lacking comprehensive collection and utilization of traffic environment data.

[0057] Because it uses fixed response time and safety distance parameters, traditional Newell models cannot adapt to changes in traffic density and driving style. In real-world traffic, changes in traffic density significantly affect drivers' response speed and safety distance choices. For example, in high-density traffic scenarios, drivers become more alert due to reduced distances between surrounding vehicles, resulting in shorter response times—a dynamic change that traditional models fail to reflect. Furthermore, fixed safety distance settings fail to account for differences in driving styles; conservative and aggressive drivers choose different safety distances under the same traffic conditions, a distinction that traditional models cannot differentiate, reducing model accuracy. In addition, relying solely on data collected from onboard sensors fails to acquire information such as traffic density and inter-vehicle traffic flow, limiting the model's ability to analyze the relationship between vehicle behavior and the overall traffic environment and making it difficult to comprehensively assess vehicle behavior under the influence of traffic flow.

[0058] In related technologies, vehicle lane-changing models often focus on a single perspective when analyzing lane-changing behavior. Some models only consider local factors, focusing on the interaction between individual vehicles, such as the distance and speed difference between the lane-changing vehicle and surrounding vehicles, and determining the feasibility of lane changing by setting thresholds and rules. For example, a lane-changing operation is allowed when the distance between the lane-changing vehicle and the vehicle in front in the target lane is greater than a certain distance, while maintaining a certain safe distance from the vehicle behind. Such models can describe the behavioral logic of individual vehicles at the moment of lane changing to a certain extent, mainly focusing on the changes in distance and speed adjustment between the lane-changing vehicle and directly related surrounding vehicles, and assessing the safety of lane changing and its impact on local traffic by calculating the changes in these vehicle operating parameters. At the overall level, although some traffic flow models consider information such as traffic flow and density, they are independent of lane-changing models and do not incorporate the impact of lane-changing behavior on traffic flow into a comprehensive analytical framework.

[0059] Due to a lack of comprehensive consideration of complex traffic environments, assessments of lane-changing impacts focus on interactions between individual vehicles while neglecting the overall effect on traffic flow. Lane-changing behavior not only affects surrounding vehicles but can also lead to a redistribution of traffic flow density, thus impacting the efficiency of the entire road segment. For example, a single lane change can trigger a chain reaction, causing fluctuations in traffic flow. Existing methods do not comprehensively consider multiple local and overall factors, making it difficult to make optimal decisions in complex traffic scenarios. This may result in lane-changing behavior increasing traffic congestion and reducing traffic efficiency.

[0060] In view of this, this application provides a method and related equipment for predicting the impact of lane changing on vehicles. This method predicts the impact of lane changing on surrounding vehicles and the overall traffic flow, and then comprehensively evaluates the benefits of lane changing based on the obtained vehicle impact prediction index and traffic flow impact prediction index. This achieves comprehensive prediction of the impact of lane changing at both the micro and macro levels, improves the accuracy of impact prediction, and provides an accurate reference for the lane changing behavior of lane changing vehicles.

[0061] This application provides a method and related equipment for predicting the impact of vehicle lane changes, relating to the field of traffic control technology. The method for predicting the impact of vehicle lane changes provided in this application can be applied to a terminal, a server, or software running on a terminal or server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or vehicle terminal, but is not limited to these. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application implementing a method for predicting the impact of vehicle lane changes, but is not limited to the above forms.

[0062] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0063] Figure 1 This is an optional flowchart of a vehicle lane-changing impact prediction method provided in an embodiment of this application. Figure 1 The method may include, but is not limited to, steps S101 to S104.

[0064] Step S101: Obtain traffic environment data around the lane-changing vehicle. The traffic environment data includes traffic flow data and vehicle-level data.

[0065] Step S102: Based on vehicle-level data, predict the impact of lane-changing vehicles changing to the target lane on other vehicles using the improved car-following model to obtain vehicle impact prediction index.

[0066] Step S103: Based on traffic flow data, predict the impact of lane-changing vehicles changing to the target lane on traffic flow using a traffic flow model to obtain traffic flow impact prediction indicators.

[0067] Step S104: Calculate the lane-changing benefits based on the vehicle impact prediction index and the traffic flow impact prediction index to obtain the benefit index of lane-changing vehicles changing to the target lane.

[0068] Steps S101 to S106 as shown in the embodiments of this application predict the impact of lane-changing vehicles on surrounding vehicles and the overall traffic flow, and then comprehensively evaluate the benefits of lane changing based on the obtained vehicle impact prediction index and traffic flow impact prediction index. This achieves comprehensive prediction of the impact of lane changing at both the micro and macro levels, improves the accuracy of impact prediction, and provides accurate reference for the lane-changing behavior of lane-changing vehicles. It helps lane-changing vehicles make lane-changing decisions that ensure individual driving safety and efficiency while also taking into account the overall traffic flow operation efficiency.

[0069] In step S101 of some embodiments, the vehicle-level data includes at least the contours, accelerations, speeds, and attitudes of the lane-changing vehicle and other vehicles within a certain range of the lane-changing vehicle, as well as semantic information such as road traffic signs and lane markings. The traffic flow data includes at least traffic density, time-averaged speed, spatial-averaged speed, free-flow speed, and congestion density. Traffic environment data can provide a comprehensive data foundation for subsequent lane-changing impact analysis.

[0070] In some embodiments, step S101 may include, but is not limited to, steps S110 to S120:

[0071] Step S110: Collect data of itself and surrounding vehicles through the on-board sensors of the lane-changing vehicle to obtain vehicle-level data;

[0072] Step S120: Traffic flow data is obtained by collecting traffic data within a certain range centered on lane-changing vehicles through roadside units.

[0073] In step S110 of some embodiments, the vehicle-mounted sensors include a high-precision positioning module, an inertial measurement unit, a wheel speed sensor, a millimeter-wave radar, a lidar, and a vision camera, etc.

[0074] A high-precision positioning module, inertial measurement unit, and wheel speed sensors collect real-time data on the position, speed, and acceleration of the following vehicle. The onboard positioning system, utilizing satellite positioning technology, determines the vehicle's location in geospatial space, providing fundamental positional data for subsequent following and lane-changing decisions. The inertial measurement unit monitors the vehicle's acceleration and angular velocity in real time, accurately sensing changes in the vehicle's motion state. The wheel speed sensors indirectly obtain the vehicle's speed by measuring the rotational speed of the wheels. These three sensors work together to collect the vehicle's motion information from different dimensions, improving the accuracy and real-time nature of vehicle motion data and providing core data support for vehicle behavior analysis.

[0075] Millimeter-wave radar utilizes the principle of electromagnetic wave transmission and reception in the millimeter-wave band to accurately measure the distance and relative speed between a vehicle and surrounding objects. LiDAR, by measuring the time delay of reflected light, constructs a three-dimensional point cloud model of the surrounding environment, enabling precise acquisition of the target vehicle's outline and size data. Cameras provide visual information about the vehicle's appearance; combined with radar, they can more accurately identify vehicle types, such as passenger cars, heavy trucks, and buses. This information helps assess a vehicle's handling performance, driving stability, and potential driving risks. For example, heavy trucks, due to their larger size and weight, require greater safety distances and longer response times during following and lane changes; accurate vehicle type identification helps the system plan more rationally for following and lane-changing strategies. Fusion of data from radar and vision sensors fully leverages the advantages of both, overcoming the limitations of a single sensor. For instance, millimeter-wave radar can still operate stably in adverse weather conditions such as rain or fog, providing supplementary data to vision sensors.

[0076] In step S120 of some embodiments, the roadside unit adaptively determines the spatial boundary for collecting section traffic flow data. Roadside equipment data includes rear vehicle data and section traffic flow data, specifically collected via roadside cameras, millimeter-wave radar, lidar, and geomagnetic sensors, measuring parameters such as traffic density, average speed, free-flow speed, and congestion density of the target lane. The data update frequency is 1Hz. Traffic density reflects the distribution of vehicles per unit length of lane and is an important indicator for measuring traffic congestion; average speed reflects the average speed of vehicles over a certain period; free-flow speed represents the speed of vehicles without traffic interference; and congestion density describes the vehicle density when the road reaches its congestion limit. Furthermore, the roadside unit also obtains historical traffic flow statistics (such as peak hours and off-peak hours) from the traffic management center. This historical data covers traffic information under different time periods, weather conditions, and road conditions, providing rich data resources for model training and validation. By combining real-time collected data with historical data and utilizing data mining and machine learning techniques, we can analyze traffic flow patterns more deeply, predict future traffic conditions, and provide more forward-looking information support for subsequent vehicle decisions.

[0077] For example, this embodiment first determines the spatiotemporal boundaries of the interval data collection. The collection range is defined as the spatial interval [LCP-L] centered on the lane change point (LCP). backward ,LCP+L forward ], where: the length L of the rear region backward Covers the preliminary calculated impact range R of the lane change disturbance propagating backward. influence The front length L forward Covering the preliminary calculated local traffic area F for vehicle driving forward focus influence First, regarding L... backward and L forward Take experience values ​​respectively Each distance is 300 meters, then the rear impact range R is calculated. influence And compare it with empirical values, as shown in Equations (1) and (2).

[0078] (1);

[0079] (2);

[0080] The propagation range of the disturbance is determined by the speed of the traffic shock wave and the duration of the disturbance. The preliminary calculation of the rear impact range is shown in equation (3).

[0081] (3);

[0082] in, The duration of the disturbance. Maximum wave velocity. Duration of disturbance. An empirical value of 8 seconds was adopted, covering the time for lane changing and speed adjustment while following other vehicles. The maximum wave speed was initially derived using the Green-Shields linear model. As shown in equation (4).

[0083] (4);

[0084] in, The current traffic flow in the target lane at the LCP. The current density of the target lane at LCP. The target lane congestion density is determined using historical data.

[0085] In some embodiments, multi-sensor data is fused and structured data is extracted. A time-space alignment algorithm is used to fuse vehicle-mounted and roadside data. For example, a method combining timestamps and interpolation algorithms ensures that the vehicle distance detected by radar and the information acquired by the roadside camera are consistent in the time dimension. After time synchronization, the data also needs to be spatially fused. Since the observation perspectives and coordinate systems of vehicle-mounted sensors and roadside equipment are different, a coordinate transformation algorithm is needed to unify all data into a global traffic coordinate system. Taking vehicle-mounted LiDAR and roadside cameras as an example, the LiDAR establishes a polar coordinate system centered on itself, while the coordinates of the camera image data are related to the imaging plane. Through a coordinate transformation algorithm, the polar coordinate data of the LiDAR is converted into Cartesian coordinates, and the image coordinates are converted into global traffic coordinates based on the camera's intrinsic and extrinsic parameters, as well as its installation position and attitude information. In this process, considering the influence of sensor installation errors and environmental factors on coordinate transformation, a calibration model based on the least squares method is used to optimize the coordinate transformation parameters. By collecting calibration points at multiple known locations, measuring the observation data of these points from different perspectives using sensors, and then using the least squares method to solve for the coordinate transformation parameters, the coordinate error after transformation is minimized, thereby improving the accuracy of spatial fusion.

[0086] Raw data requires preprocessing to eliminate noise, redundant information, and scene interference. This includes data cleaning and noise reduction: based on statistical thresholding methods (such as the 3σ criterion) and physical constraints (such as vehicle speed range), outlier data exceeding reasonable ranges is filtered out. Kalman filtering and an adaptive sliding window algorithm are used to smooth acceleration and distance data, reducing instantaneous noise interference. In vehicle acceleration data processing, Kalman filtering predicts the current acceleration based on the previous moment's acceleration and corrects it using the current measurement value, resulting in smooth and accurate acceleration data. The adaptive sliding window algorithm dynamically adjusts the window size according to the data's changing characteristics, performing local smoothing. For distance data, when vehicles are traveling in areas with complex traffic conditions and frequent changes in distance, the adaptive sliding window algorithm automatically reduces the window size to more accurately track real-time changes in distance; while when distance is relatively stable, the window is appropriately enlarged to improve data processing efficiency and smoothing effect. By comprehensively considering the characteristics and advantages of each sensor, multi-source data is integrated into accurate and comprehensive vehicle and environmental information, providing a solid data foundation for subsequent driving decisions.

[0087] For example, multi-source data fusion is implemented in the CARLA simulation environment as follows:

[0088] (1) Sensor data preprocessing and target list generation. The raw data output by the vehicle-mounted vision sensor and radar sensor needs to be converted into a unified target list. The vision detection module outputs the first... Frame target list As shown in equation (5).

[0089] (5);

[0090] in, For frame timestamps, , For the first The total number of visually detected targets per frame. For the first One visual detection target, For the first The parameters of each visual detection bounding box are defined as follows: , To detect the pixel coordinates of the top-left corner of the bounding box in the image coordinate system, The width and height of the detection box are in pixels. For the first The category confidence of each visually detected target. For the first Category labels for visually detected targets.

[0091] The radar sensing module outputs the first... Frame target list As shown in equation (6).

[0092] (6);

[0093] in, , For the first The total number of targets detected by the frame radar. For the first One radar target, For the first The radar detects the three-dimensional coordinates of the target in the vehicle's coordinate system. For the first The radial velocity of a radar target; the sign indicates whether it is moving away from or near the radar sensor. For the first The average reflection intensity of a radar target point cloud is a dimensionless relative value.

[0094] (2) Data Association. The Hungarian algorithm is used to solve the optimal allocation problem between visual and radar targets. A cost matrix is ​​constructed. , dimension Its elements Indicates radar target With visual target The associated cost is calculated using the formula shown in equation (7).

[0095] (7);

[0096] in, The function will target radar The 3D coordinates of the vehicle's coordinate system are transformed into a 2D image coordinate system using known camera intrinsic and extrinsic parameters through a projection model. For visual inspection boxes The coordinates of the center point are obtained. By solving for the optimal allocation, the set of matched pairs (Matches), the unmatched visual targets (Unmatched_Visual), and the unmatched radar targets (Unmatched_Radar) are obtained. This is the L2 norm operator, used to calculate the Euclidean distance (in pixels) between two two-dimensional points.

[0097] (3) Conflict detection. Existing conflict: all unmatched radar targets. All of these trigger an existence conflict alarm, as shown in equation (8).

[0098] (8);

[0099] in, For the first The frame existence collision list, whose elements are all radar targets for which no visual target match was found. .

[0100] Geometric attribute conflict: for each matching pair Calculate the distance deviation in its three-dimensional space, as shown in equation (9).

[0101] (9);

[0102] in, To estimate the 3D position of the vehicle in the coordinate system obtained through the visual inspection box, the estimation method may include using the midpoint of the bottom edge of the inspection box, assuming a standard target height and combining it with the ground plane model, and solving it through inverse perspective transformation. The 3D coordinates of the target as measured by radar. To estimate the Euclidean distance between the visual location and the radar-measured location, a dynamic distance threshold is set. This threshold can be dynamically adjusted based on factors such as target type, relative speed, and sensor historical confidence level. A typical empirical value is 1.0 meter to 2.0 meters. If the geometric attribute conflict exists, it is determined that there is a conflict, as shown in equation (10).

[0103] (10);

[0104] Indicates the first A list of geometric attribute conflicts for a frame, whose elements are all positions whose deviations exceed a threshold. Matching target pair .

[0105] (4) Confidence assessment. The reliability of each sensor target is assessed based on the comparison with the true value. The visual confidence is calculated based on the intersection over union (IoU), as shown in Equation (11).

[0106] (11);

[0107] In order to match the visual target The corresponding ground truth detection box parameters. IoU is the intersection-union ratio calculation function, which is the ratio of the intersection area to the union area of ​​two detection boxes. The value ranges from [0,1]. The larger the value, the more accurate the visual detection. For the first The confidence level of a visual target, with a value range of [0,1].

[0108] The radar confidence level is calculated based on the Euclidean distance between its measured location and the true value, as shown in Equation (12).

[0109] (12);

[0110] in, The attenuation coefficient controls the rate at which confidence decreases with increasing distance; it is typically set based on sensor accuracy, for example... . This is the Euclidean distance between the radar target and the corresponding real-value target's 3D coordinates. For the first Confidence level of each radar target.

[0111] (5) Adaptive correction strategy.

[0112] For the existence of conflicting objectives If its confidence level If the value exceeds a certain threshold, such as 0.7 or 0.9, the radar target is accepted and a warning is issued. For targets with conflicting geometric attributes... The final position is calculated using a confidence-based weighted fusion method, as shown in equation (13).

[0113] (13);

[0114] in, This formula estimates the 3D position of the target after fusion. It uses a weighted average to fuse radar measurements. and visual estimation of position The weights are their respective confidence levels. and .

[0115] The targets in the final output list are visualized using different colors: green indicates collision-free targets, blue indicates visually missed targets supplemented by radar, and red indicates targets with severe collisions or those identified as false targets. Thus, through the above steps, the fusion of multi-source data, collision handling, and adaptive correction are completed, providing accurate and robust environmental perception information for subsequent vehicle decisions. The detection results are as follows: Figure 2 As shown.

[0116] In step S102 of some embodiments, traditional car-following models (such as the Newell model) use fixed response time and safety distance parameters, failing to consider the impact of traffic density changes on driver alertness and reaction characteristics, and also ignoring the differentiated requirements for safety distance selection based on different driving styles (aggressive / conservative). The improved car-following model is more human-like and can more accurately simulate vehicle car-following behavior in various traffic scenarios. This not only helps in the in-depth analysis of local traffic operation mechanisms but also provides a reference for the formulation of intelligent connected vehicle following strategies and for improving the driving safety and stability of intelligent connected vehicles in complex traffic environments.

[0117] In some embodiments, the car-following model includes a first car-following module and a second car-following module, and the improved car-following model can be obtained by including, but not limited to, steps S210 to S230.

[0118] Step S210: Modify the fixed safety distance parameter in the original first car-following module to an adaptive safety distance parameter, and modify the second driving trajectory parameter to a parameter that varies with the second driving trajectory and the third driving trajectory, to obtain the improved first car-following module;

[0119] Step S220: Modify the fixed time response parameters of the original second car-following module regarding the first driving trajectory to adaptive time response parameters to obtain the improved second car-following module;

[0120] Step S230: Combine the improved first car-following module and the improved second car-following module to obtain the improved car-following model;

[0121] The first driving trajectory is the longitudinal driving trajectory of the following vehicle in the target lane, the second driving trajectory is the longitudinal driving trajectory of the initial preceding vehicle in the target lane, and the third driving trajectory is the longitudinal driving trajectory of the lane-changing vehicle.

[0122] In some embodiments, the car-following model is exemplified by the Newell model, with the traditional Newell model shown in equations (14) and (15). The first car-following module is shown in equation (14), and the second car-following module is shown in equation (15).

[0123] (14);

[0124] (15);

[0125] Please refer to Figure 3 and Figure 4 , For the initial leading vehicle The longitudinal driving trajectory (second driving trajectory). To follow the vehicle The longitudinal driving trajectory (first driving trajectory). To follow the vehicle longitudinal driving speed, This represents the minimum distance between the front ends of the vehicles. To follow the vehicle With lag time Follow the longitudinal trajectory of the vehicle in front. The fixed safety distance parameter is... The fixed-time response parameter is .

[0126] To address the limitations of the traditional Newell car-following model, which suffers from fixed parameters and poor adaptability, this embodiment proposes a dynamic parameter adjustment mechanism, as shown in equations (16) and (17). Equation (17) represents the improved second car-following module.

[0127] (16);

[0128] (17);

[0129] in, For adaptive response time parameters, This is an adaptive safety distance parameter.

[0130] Furthermore, a variation parameter for the second and third driving trajectories is introduced into equation (16), as shown in equation (18). Equation (18) is the improved first car-following module.

[0131] (18);

[0132] Among them, the changing parameters are . This is the third driving trajectory (the longitudinal driving trajectory of vehicles changing lanes). and These are the corresponding attention coefficients.

[0133] The improved car-following model includes equations (17) and (18).

[0134] In some embodiments, step S102 may include, but is not limited to, steps S201 to S203:

[0135] Step S201: Based on the acceleration changes of the following vehicle in the vehicle-level data and the distance between different vehicles, the improved first car-following module is used to predict the trajectory of the following vehicle to obtain the first predicted trajectory of the following vehicle.

[0136] Step S202: Based on the real-time traffic density of the traffic flow data, the improved second car-following module is used to predict the trajectory of the following vehicle to obtain the second predicted trajectory of the following vehicle.

[0137] Step S203: Calculate the vehicle impact prediction index based on the first and second predicted trajectories.

[0138] In steps S201 and S203 of some embodiments, the adaptive safety distance parameter in equation (18) is adjusted according to the vehicle's acceleration changes. This has an impact; by obtaining the acceleration changes of the following vehicle, the adaptive safety distance parameters can be determined. The specific values, in equation (18), the change parameters of the second and third driving trajectories include the attention parameters for the second and third driving trajectories. The attention parameters are changed in real time for different driving trajectories according to the different distances between vehicles, thereby determining the corresponding change parameters. Based on the above parameters, the improved first following module can predict the driving impact of lane-changing vehicles on the following vehicles, that is, the values ​​in equation (18). The real-time traffic density affects the driver's response time. When the traffic density is high, the driver will be more alert, thus the adaptive response time parameter will be smaller. Therefore, the improved first car-following module can predict the second predicted trajectory of the following vehicle based on the real-time traffic density, i.e., Equation (17). The vehicle impact prediction index is calculated by combining equations (17) and (18). As shown in equations (19) and (20).

[0139] (19);

[0140] Taking the impact of speed changes as an example, the influence of lane-changing vehicles on following vehicles in the target lane can be defined.

[0141] (20);

[0142] In some embodiments, step S201 may include, but is not limited to, steps S211 to S216:

[0143] Step S211: Perform driving style analysis based on the acceleration changes of the following vehicle in the vehicle-level data to obtain the driving style coefficient of the following vehicle.

[0144] Step S212: Perform a safety distance analysis based on the driving style coefficient and the real-time traffic density of traffic flow data to determine the distance value for the adaptive safety distance parameter;

[0145] Step S213: Based on the longitudinal distance and longitudinal speed difference between the following vehicle and the initial preceding vehicle, attention is allocated to determine the first attention coefficient of the second driving trajectory in the changing parameters;

[0146] Step S214: Based on the lateral distance, longitudinal distance, and longitudinal speed difference between the following vehicle and the lane-changing vehicle, attention is allocated to determine the second attention coefficient of the third driving trajectory among the changing parameters;

[0147] Step S215: Normalize the first attention coefficient and the second attention coefficient to obtain the initial preceding vehicle attention coefficient of the second driving trajectory and the lane-changing vehicle attention coefficient of the third driving trajectory.

[0148] Step S216: Based on the distance value, the initial attention coefficient of the preceding vehicle, and the attention coefficient of the lane-changing vehicle, the improved first car-following module is used to predict the trajectory of the following vehicle to obtain the first predicted trajectory.

[0149] In steps S211 and S212 of some embodiments, the adaptive safety distance parameter In terms of design, considering the dynamic influence of driving style differences and traffic density on the selection of safe distance, a driving style-density coupled dynamic safe distance model is proposed to adaptively adjust the safe distance parameters. The expression is shown in equation (21).

[0150] (twenty one);

[0151] Equation (21) introduces a nonlinear adjustment effect of density on safety distance through the Sigmoid function. Density sensitivity coefficient for safe distance, This is a critical density parameter used to delineate the phase transition region of traffic conditions. When the actual density... much smaller During periods of low-density traffic, a safe distance should be maintained. Unchanged; when near At the same time, the safe distance adaptively decreases to a reasonable range, balancing efficiency and safety, and avoiding a decrease in traffic efficiency due to overly conservative spacing settings. and This can represent the maximum and minimum values ​​of the preset safe distance range. Furthermore, different vehicle models have different dimensions; larger vehicles require a larger safe distance to ensure driving safety, thus a vehicle model correction coefficient for safe distance is introduced. Its value varies depending on the vehicle model.

[0152] Furthermore, driving style coefficient As a weighting factor, the density term dynamically adjusts its influence on safe distance. Conservative drivers will choose a larger safe distance at the same density to compensate for risk, while aggressive drivers tend to maintain a distance closer to the baseline. The spacing.

[0153] Driving Style Coefficient This method quantifies the aggressiveness of driver actions and is calculated based on the time-domain fluctuations in vehicle acceleration. The specific process involves: within a sliding time window... Inside, vehicle acceleration sequences are collected in real time. Calculate the acceleration variance within this window. and normalize it to the preset maximum squared acceleration value. The driving style coefficient is obtained through equation (22).

[0154] (twenty two);

[0155] in, This represents the driving style sensitivity coefficient. The formula indicates that when a driver frequently performs rapid acceleration or emergency braking (i.e., when the acceleration variance is large), A decreasing value indicates a more aggressive driving style; conversely, a gradual change in acceleration indicates a more aggressive driving style. It reflects a conservative driving style.

[0156] In steps S213 and S215 of some embodiments, a multi-vehicle target mechanism considering the influence of lane-changing vehicles is established to simulate the attention allocation of following vehicles to the initial preceding vehicle and the lane-changing vehicle during vehicle following behavior. For the attention to the lane-changing vehicle, lateral distance is used as the key factor. An initial attention weight is determined by setting a lateral distance threshold, and a decay function is used to reflect the change in attention as the distance decreases. The weight is adjusted in stages, taking into account factors such as the rate of change of lateral distance, longitudinal distance, and speed difference, to comprehensively measure the influence of the kinematic parameters between the lane-changing vehicle and the following vehicle on the following vehicle. For the attention to the initial preceding vehicle, adjustments are made based on the speed difference between the following vehicle and the initial preceding vehicle. Finally, the two attention coefficients are normalized.

[0157] Specifically, please refer to Figure 3 For the initial attention coefficient of the vehicle in front (first attention coefficient) .

[0158] Preliminary weighting based on vertical distance: setting a vertical distance threshold. The longitudinal distance influencing factor of attention When the longitudinal distance At that time, the initial weight ;when hour, For example, in scenarios where urban roads frequently start and stop, The vehicle is relatively large, and when following closely, it pays more attention to the vehicle in front initially.

[0159] Considering the adjustment of the speed difference: Let the longitudinal speed difference between the following vehicle and the initial preceding vehicle be... Set speed difference threshold And speed difference influence factor .when hour, ;when hour, In expressway scenarios, vehicles travel at higher speeds and are more sensitive to speed differences. It is a relatively large value.

[0160] Attention coefficient for lane-changing vehicles (secondary attention coefficient) .

[0161] Preliminary weighting based on lateral distance: setting a lateral distance threshold When the lateral distance between the following vehicle and the vehicle changing lanes. At that time, the initial weight ;when hour, ,in It is a parameter that controls the decay rate and can be adjusted according to actual conditions, such as in scenarios with high traffic volume and small vehicle spacing. The value is relatively large, and the following vehicle's attention to the lane-changing vehicle increases more rapidly as the lateral distance decreases.

[0162] Considering adjustments for the rate of change of lateral distance: Calculate the rate of change of lateral distance. .like This means that when a vehicle changes lanes, it moves closer to the following vehicle, and an adjustment factor is introduced. Adjusted weights ;like ,but . This reflects the driver's sensitivity to the approaching speed of vehicles changing lanes, especially in scenarios with a more aggressive driving style. The value is relatively small, meaning that the driver's reaction to the approach of a vehicle changing lanes is relatively weak.

[0163] Combined with vertical distance adjustment: Set a vertical distance threshold. and longitudinal distance influence factor When the longitudinal distance hour, ;when hour, . The size of the longitudinal distance determines the degree to which it affects the attention coefficient. In traffic congestion and when the longitudinal distance between vehicles is small, When the value is relatively large, the vertical distance has a stronger effect on adjusting the attention coefficient.

[0164] Considering the adjustment of speed difference: the longitudinal speed difference between the following vehicle and the lane-changing vehicle is Set speed difference threshold And speed difference influence factor .when hour, ;when hour, . This reflects the driver's awareness of speed differences; in highway scenarios, vehicle speeds are relatively high. This makes vehicles pay more attention to vehicles changing lanes with a large speed difference.

[0165] Normalization process, calculation of normalization factor Normalized attention coefficients for the initial preceding vehicle (initial preceding vehicle attention coefficients) The normalized attention coefficient for lane-changing vehicles (lane-changing vehicle attention coefficient) ,thereby .

[0166] In the Newell model, the relevant part of the original car-following distance calculation is: After incorporating the attention coefficient, the following distance becomes... This refers to the changing parameters. In real-world scenarios, if a lane-changing vehicle is relatively close laterally and gradually approaches the following vehicle, while simultaneously exhibiting a significant speed difference between them, This will increase, thus increasing the following distance relative to the changing lane vehicle; if the initial longitudinal distance to the vehicle in front is close and the speed difference is large, This will increase the vehicle's attention to the initial vehicle in front, allowing it to adjust its following distance more appropriately and improving the model's accuracy in simulating vehicle following behavior.

[0167] In step S216 of some embodiments, after determining the specific values ​​of the above parameters based on the data obtained in step S101, that is, determining the specific values ​​of each parameter of equation (18), the first predicted trajectory of equation (18) is determined. It can be done Characterization.

[0168] In some embodiments, step S202 may include, but is not limited to, steps S221 to S226:

[0169] Step S221: Calculate the response time based on the real-time traffic density and determine the time value of the adaptive time response parameter;

[0170] Step S222: Based on the time value, the improved second car-following module is used to predict the trajectory of the following vehicle to obtain the second predicted trajectory.

[0171] In step S221 of some embodiments, the adaptive time response parameter We can obtain it through equation (23):

[0172] (twenty three);

[0173] Among them, the reference response time parameter This serves as a baseline reference value for drivers' reaction capabilities under standard traffic conditions. Subsequently, a traffic density sensitivity term is introduced, which is expressed as an exponential function. The response time is dynamically adjusted. This indicates the real-time traffic density of the current lane. This is the density sensitivity coefficient for response time, used to quantify the degree to which density modulates response time. (Exponential term) The mechanism of action is as follows: when lane density increases, drivers become more alert due to the reduced distance between surrounding vehicles, improving the response efficiency of the perception-decision-execution link and thus reducing response time. It exhibits exponential decay characteristics.

[0174] Furthermore, equation (23) introduces a speed difference sensitive term, which is weighted by a linear weighting factor. The response time is corrected a second time. To determine the speed difference between the following vehicle and the target vehicle in front, The sensitivity coefficient for the speed difference in response time. This represents the free-flow speed on the road. This correction term reflects the driver's dynamic response to fluctuations in the speed difference between their vehicle and the vehicle in front: when the speed difference increases, drivers tend to increase alertness and shorten their reaction time to quickly adjust their driving behavior to cope with the change, thereby reducing the risk of rear-end collisions. Furthermore, considering the impact of vehicle type on response time, different vehicle types, due to differences in vehicle size, handling performance, driver visibility, and other factors, will result in variations in driver reaction ability. Therefore, a vehicle type correction coefficient for response time is introduced. Its value varies depending on the vehicle model.

[0175] The aforementioned correction factors are coupled through a product to form the overall response time. The model's parameter updates depend on the input traffic density. With speed difference It can dynamically adapt to the behavioral characteristics of drivers in different traffic scenarios.

[0176] Similarly, equation (18) It can also be obtained through the steps described above.

[0177] In step S222 of some embodiments, the second predicted trajectory of equation (17) It can be done Characterization.

[0178] In some embodiments, step S103 may include, but is not limited to, steps S301 to S304:

[0179] Step S301: Predict the density change of lane-changing vehicles when they change lanes to the target lane using a traffic flow model, and obtain the change in traffic density of the target lane.

[0180] Step S302: Calculate the density based on the current traffic density and change of the target lane in the traffic flow data to obtain the predicted traffic density after the lane change.

[0181] Step S303: Calculate the traffic flow based on the current traffic density and the predicted traffic density to obtain the traffic flow before and after the lane change.

[0182] Step S304: Calculate the traffic flow impact prediction index based on the traffic flow volume before and after the lane change.

[0183] In steps S301 to S304 of some embodiments, the impact of lane-changing vehicles on the traffic flow of the target lane is quantified based on a traffic flow model. The LWR (Lighthill, Whitham and Richards) model is based on the principle of traffic flow conservation, treating traffic flow as a continuous medium flow, and presenting the macroscopic characteristics of traffic flow through the relationship between traffic density, speed and volume. Specifically, it is shown in equation (24).

[0184] (twenty four);

[0185] in, Traffic flow speed, For traffic density, For observation time, The observation distance is given by equation (24). Equation (24) shows that the number of vehicles flowing into and out of a certain area per unit time is conserved. In practical applications, in order for the LWR model to accurately reflect the target lane conditions, its parameters need to be finely calibrated. Using a large amount of multi-source traffic data collected in the second step, covering traffic density, speed and flow data of different time periods and road segments, the least squares method is used for parameter fitting. According to the Green Shields model, under normal traffic flow density conditions, it is assumed that the speed-density relationship exhibits linear characteristics, as shown in equation (25).

[0186] (25);

[0187] in, Traffic flow speed, For free flow velocity, Let's consider the congestion density. When including lane-changing vehicles in the LWR model analysis, their impact on the traffic density and speed of the target lane must be fully considered. Lane-changing vehicles entering the target lane alter the local vehicle distribution, leading to changes in traffic density. Assume that before the lane-changing vehicles enter, the traffic density of the target lane is... After vehicles changing lanes enter, the change in traffic density in the local area is: The new traffic density The relationship is shown in equation (26).

[0188] (26);

[0189] Traffic flow before lane change Traffic flow after lane change Specifically, as shown in equations (27) and (28).

[0190] (27);

[0191] (28);

[0192] Traffic flow impact prediction indicators As shown in equation (29).

[0193] (29);

[0194] This indicator reflects the relative change in traffic flow caused by a single lane change, thus allowing assessment of the impact of lane-changing vehicles on traffic flow in the target lane. Similarly, changes in the current lane are also assessed.

[0195] In some embodiments, step S104 may include, but is not limited to, steps S401 to S402:

[0196] Step S401: Obtain lane change safety indicators and lane change driving efficiency;

[0197] Step S402: The vehicle impact prediction index, traffic flow impact prediction index, lane change safety index, and lane change driving efficiency are weighted and summed to obtain the benefit index.

[0198] In steps S401 to S402 of some embodiments, a lane-changing behavior model is established by combining the impact of lane changing on following vehicles in the target lane and on traffic flow. In traffic scenarios, the decision of lane-changing vehicles not only affects their own driving efficiency but also influences surrounding vehicles and the overall traffic flow. From the perspective of the impact on surrounding vehicles, lane-changing vehicles have a close interactive relationship with them. When a lane-changing vehicle is preparing to enter the target lane, if the distance between the lane-changing vehicle and the following vehicle in the target lane is too close or the speed difference is too large, it may cause a significant impact on the following vehicle, interfering with the vehicle's trajectory and speed. From the perspective of the impact on traffic flow, lane-changing behavior affects the traffic flow state of the target lane, including the speed, density, and volume of traffic flow. A large number of lane-changing behaviors may lead to or exacerbate traffic congestion, reducing the overall capacity of the road. Let the lane-changing vehicle be... During the decision-making cycle Inside, vehicles changing lanes There is a set of behaviors These correspond to constant speed lane changing, acceleration lane changing, deceleration lane changing, and straight driving, respectively. Define the utility function. This indicates that lane-changing vehicles are in the decision-making cycle. Take action The benefit at that time, i.e. the benefit index, the utility function comprehensively considers the influence of surrounding vehicles, traffic flow, safety, efficiency and other factors, as shown in equation (30).

[0199] (30);

[0200] in, These represent the influence weights for each indicator, with values ​​assigned according to different strategy preferences; vehicle impact prediction indicator. The extended Newell model established reflects the immediate impact of lane-changing vehicles on following vehicles; traffic flow impact prediction indicators. Quantified based on traffic flow models (such as the LWR model), this reflects the impact of lane-changing behavior on traffic flow in the target lane. Lane-changing safety indicators The measurement is based on the risk of collision between vehicles and the dynamic stability assessment of vehicles changing lanes. Lane change efficiency. It is mainly measured by the degree of improvement in the driving environment of vehicles after changing lanes.

[0201] Based on the outputs of traffic condition and vehicle impact models and traffic flow impact models, safe and efficient lane-changing operations are achieved while minimizing the negative impact on local traffic flow. Lane-changing operations are completed in a smoother, safer manner with less impact on traffic flow.

[0202] In some embodiments, please refer to Figure 5 The process involves seven steps: First, onboard sensors perceive surrounding vehicle data. Through collaborative sensing between onboard sensors and roadside equipment, comprehensive dynamic information about vehicles and the traffic environment is acquired. Second, roadside units adaptively determine the spatial boundaries for traffic flow data collection within a given section. Third, multi-sensor data fusion extracts structured data. A time-space alignment algorithm is used to fuse onboard and roadside data. Fourth, a dynamic time-space Newell model is extended by combining traffic density and driving style, dynamically coupling and extending the response time and safety distance parameters in the traditional Newell car-following model in multiple dimensions. Fifth, a multi-target following mechanism considering the influence of lane-changing vehicles is established to simulate the attention allocation of following vehicles to the initial preceding vehicle and lane-changing vehicles during car-following behavior. Sixth, the impact of lane-changing vehicles on following vehicles in the target lane is quantified based on the extended Newell model. The comprehensive impact of lane-changing vehicles entering the target lane on the motion state of following vehicles is analyzed based on the extended Newell model. Seventh, the impact of lane-changing vehicles on the traffic flow in the target lane is quantified based on the traffic flow model. The eighth step is to establish a lane-changing behavior impact model by combining the impact of lane changing on vehicles following in the target lane and the impact on traffic flow.

[0203] In some embodiments, please refer to Figure 6 , Figure 7 and Figure 8 To verify the effectiveness of the proposed extended Newell model and dual attention method, a lane-changing scenario was constructed to compare the proposed method and the original Newell model in a simulation. The simulation software Carla was used as the test platform, and the following settings were configured: Figure 3 The lane-changing scenario shown is tested, where there is a vehicle in front and a vehicle behind in the target lane, and the vehicle changing lanes enters the target lane. The longitudinal speeds of the vehicle in front and the vehicle changing lanes in the target lane are set to 30 km / h, and the initial speed of the vehicle behind in the target lane is 25 km / h. Figure 6 This is a graph showing the speed change of vehicles following in the target lane when a vehicle changes lanes. Figure 7 For speed-related statistical comparison, the original mean speed was 31.74 km / h with a standard deviation of 7.39 km / h, while the planned mean speed was 31.91 km / h with a standard deviation of 2.40 km / h. This represents a 0.17 km / h increase in speed while reducing the standard deviation by 67.5%. It can be seen that the original Newell model only switched the car-following target after the lane-changing vehicle crossed the lane line, and did not adjust according to changes in distance and speed difference, resulting in significant fluctuations in the original speed. The model proposed in this embodiment focuses on the vehicle when it begins to change lanes, and during the lane-changing process, it simultaneously utilizes the motion information of the vehicle in front in the current lane and the changing vehicle, comprehensively considering multiple factors such as lateral distance, longitudinal distance, and speed difference to dynamically allocate attention weights and adjust car-following parameters, achieving smooth car-following switching while maintaining a high average speed and a low standard deviation. Finally, simulations were conducted under different traffic flow densities, and the lane-changing utility function was used for evaluation. The experiment used a three-lane highway scenario and set up three typical traffic densities: low density (20 veh / km), medium density (30 veh / km) and high density (45 veh / km) environments to simulate lane-changing behavior under different levels of congestion on real roads. 100 independent experiments were conducted in each traffic density scenario. Figure 8 Box plots are shown for the lane-changing utility values ​​of the two methods. It can be seen that the average utility value improved by 16.4% in the lane-changing evaluation of the model in this embodiment, verifying the effectiveness of the proposed method.

[0204] The embodiments of this application include at least the following beneficial effects:

[0205] In terms of environmental perception and data acquisition, a dynamic data acquisition mechanism that considers spatiotemporal boundaries is introduced to improve the relevance of environmental perception and the completeness of decision-making information. Traditional methods often use fixed ranges or empirical values ​​for data acquisition, which is difficult to adapt to the dynamic changes in the impact range of lane-changing behavior under different traffic conditions, easily leading to incomplete or redundant information acquisition. This embodiment is based on traffic wave theory, and dynamically determines the optimal data acquisition interval centered on the estimated lane-changing point by predicting the propagation characteristics of lane-changing disturbances in traffic flow and the forward attention range of vehicles following. On the one hand, this reduces the waste of resources or information loss under different traffic densities with fixed acquisition ranges, improving the working efficiency and data effectiveness of the perception system; on the other hand, it provides a comprehensive and structured spatiotemporal data foundation for subsequent lane-changing decisions and impact assessments, enhancing the adaptability to complex dynamic traffic environments and the reliability of decision-making.

[0206] In terms of car-following behavior modeling, the improved Newell car-following model, by introducing traffic density sensitivity, adaptive driving style adjustment mechanisms, and multi-objective following mechanisms, can more realistically reflect the driver's behavioral characteristics under different traffic environments and driving styles. This model can dynamically adjust the driver's response time based on real-time changes in traffic density and reasonably adjust the safety distance according to differences in driving style, making the model more human-like and able to more accurately simulate the car-following behavior of vehicles in various traffic scenarios. This not only helps to deeply analyze local traffic operation mechanisms but also provides a reference for the formulation of intelligent connected vehicle following strategies and improving the driving safety and stability of intelligent connected vehicles in complex traffic environments.

[0207] Regarding the construction of lane-change impact models, this embodiment establishes a comprehensive assessment method for lane-change impact that integrates microscopic vehicle interaction effects and macroscopic traffic flow effects. At the microscopic level, it utilizes a multi-follower target mechanism to accurately quantify the dynamic interference of lane changes on following vehicles (such as sudden acceleration changes and decreased comfort). At the macroscopic level, it assesses the potential impact of lane changes on lane capacity, traffic stability, and overall efficiency based on traffic flow theory. This improves upon the disconnect between microscopic and macroscopic analysis in existing methods, enabling autonomous vehicles to make lane-change decisions that both enhance their own safety and reduce negative impacts on traffic flow. This increases their anthropomorphism, makes them more acceptable to drivers, and improves the cooperation of drivers in surrounding vehicles.

[0208] When the training samples of the model involved in this embodiment are sufficiently abundant, the obtained model parameters and decision-making mechanisms are highly representative. The established model can be used to analyze traffic flow, thereby providing a reference for strengthening traffic control and guidance, alleviating traffic congestion, and improving road traffic efficiency.

[0209] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0210] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0211] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0212] The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0213] The memory 902 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 using the methods described in the embodiments of this application.

[0214] The input / output interface 903 is used to implement information input and output;

[0215] The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0216] Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904);

[0217] The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0218] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0219] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0220] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0221] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0222] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0223] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0224] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0225] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0226] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0227] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0228] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0229] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0230] The units described above 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.

[0231] Furthermore, 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. The integrated unit can be implemented in hardware or as a software functional unit.

[0232] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part 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 multiple 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 of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0233] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for predicting the impact of vehicle lane changing, characterized in that, The method includes the following steps: Acquire traffic environment data around the lane-changing vehicle, including traffic flow data and vehicle-level data; Based on the vehicle-level data, the impact of the lane-changing vehicle on other vehicles when it changes to the target lane is predicted by the improved car-following model, and a vehicle impact prediction index is obtained. Based on the traffic flow data, the impact of the lane-changing vehicle changing to the target lane on the traffic flow is predicted by the traffic flow model, and a traffic flow impact prediction index is obtained. The lane-changing benefit is calculated based on the vehicle impact prediction index and the traffic flow impact prediction index to obtain the benefit index of the lane-changing vehicle changing to the target lane. The car-following model includes a first car-following module and a second car-following module. The improved car-following model is obtained through the following steps: The fixed safety distance parameter in the original first car-following module is modified to an adaptive safety distance parameter, and the second driving trajectory parameter is modified to a parameter that varies with the second driving trajectory and the third driving trajectory, resulting in the improved first car-following module; The fixed time response parameters regarding the first driving trajectory in the original second car-following module are modified to adaptive time response parameters to obtain the improved second car-following module; The improved first car-following module and the improved second car-following module are combined to obtain the improved car-following model; Wherein, the first driving trajectory is the longitudinal driving trajectory of the following vehicle in the target lane, the second driving trajectory is the longitudinal driving trajectory of the initial preceding vehicle of the following vehicle in the target lane, and the third driving trajectory is the longitudinal driving trajectory of the lane-changing vehicle. The step involves using the vehicle-level data and an improved car-following model to predict the impact of a lane-changing vehicle transitioning to the target lane on other vehicles, resulting in a vehicle impact prediction index, including: Based on the acceleration changes of the following vehicle in the vehicle-level data and the distance between different vehicles, the improved first car-following module predicts the trajectory of the following vehicle to obtain the first predicted trajectory of the following vehicle. Based on the real-time traffic density of the traffic flow data, the improved second car-following module is used to predict the trajectory of the following vehicle, and the second predicted trajectory of the following vehicle is obtained. The vehicle impact prediction index is obtained by calculating based on the first and second predicted trajectories.

2. The method according to claim 1, characterized in that, The acquisition of traffic environment data around lane-changing vehicles includes: The vehicle-level data is obtained by collecting data on itself and surrounding vehicles through the onboard sensors of the lane-changing vehicle. Traffic flow data is obtained by collecting traffic data within a certain range centered on the lane-changing vehicle through roadside units.

3. The method according to claim 1, characterized in that, The step of predicting the trajectory of the following vehicle based on the acceleration changes of the following vehicle in the vehicle-level data and the distance between different vehicles, using the improved first car-following module, to obtain the first predicted trajectory of the following vehicle, includes: Based on the acceleration changes of the following vehicle in the vehicle-level data, a driving style analysis is performed to obtain the driving style coefficient of the following vehicle. Based on the driving style coefficient and the real-time traffic density of the traffic flow data, a safe distance analysis is performed to determine the distance value for the adaptive safe distance parameter; Attention is allocated based on the longitudinal distance and longitudinal speed difference between the following vehicle and the initial leading vehicle, and a first attention coefficient of the second driving trajectory in the changing parameters is determined. Attention is allocated based on the lateral distance, longitudinal distance, and longitudinal speed difference between the following vehicle and the lane-changing vehicle, and a second attention coefficient of the third driving trajectory is determined among the changing parameters. The first attention coefficient and the second attention coefficient are normalized to obtain the initial preceding vehicle attention coefficient of the second driving trajectory and the lane-changing vehicle attention coefficient of the third driving trajectory. Based on the distance value, the initial attention coefficient of the preceding vehicle, and the attention coefficient of the lane-changing vehicle, the improved first car-following module predicts the trajectory of the following vehicle to obtain the first predicted trajectory.

4. The method according to claim 1, characterized in that, The step of predicting the trajectory of the following vehicle using the improved second car-following module based on the real-time traffic density of the traffic flow data to obtain the second predicted trajectory of the following vehicle includes: Response time is calculated based on the real-time traffic density to determine the time value of the adaptive time response parameter; Based on the time value, the improved second car-following module predicts the trajectory of the following vehicle to obtain the second predicted trajectory.

5. The method according to claim 1, characterized in that, The step involves predicting the impact of lane-changing vehicles transitioning to the target lane on traffic flow based on the traffic flow data using a traffic flow model, thereby obtaining traffic flow impact prediction indicators, including: The traffic flow model is used to predict the density change of the lane-changing vehicles when they change lanes to the target lane, thereby obtaining the change in traffic density of the target lane. Based on the current traffic density and the change in the target lane in the traffic flow data, the density is calculated to obtain the predicted traffic density after the lane change; Traffic flow is calculated based on the current traffic density and the predicted traffic density to obtain the traffic flow before and after the lane change. The traffic flow impact prediction index is obtained by calculating the traffic flow volume before and after the lane change.

6. The method according to claim 1, characterized in that, The step of calculating the lane-changing benefit based on the vehicle impact prediction index and the traffic flow impact prediction index to obtain the benefit index of the lane-changing vehicle changing to the target lane includes: Obtain lane change safety indicators and lane change driving efficiency; The benefit index is obtained by weighted summation of the vehicle impact prediction index, the traffic flow impact prediction index, the lane change safety index, and the lane change driving efficiency.

7. An electronic device, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the method as described in any one of claims 1 to 6.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

Images