Intelligent network connected vehicle dynamic grouping method for expressway weaving area

Abstract

The application discloses a kind of fast road interweaving area intelligent network connection vehicle dynamic grouping method, belong to intelligent transportation and vehicle networking collaborative control technical field, comprising:A. based on the dynamic grouping rule of longitudinal position;B. based on the dynamic grouping update period adjustment of communication link quality;C. group cooperation control strategy;D. group cooperation control strategy;E. real world fast road interweaving area grouping method calibration, the advantages of the application are: by real-time dynamic grouping, the global cooperative control problem of complex intelligent network connection vehicle in fast road interweaving area is decomposed into multiple local sub-problems, simplify the complex vehicle interweaving problem, thereby reduce the complexity of problem solving.For target vehicle with lane-changing demand, suitable gap is created, and smooth and efficient merging and exiting of vehicles is realized.Time and space resources are fully utilized, and the collaboration capability of CAV is exerted, thereby improving the safety and traffic efficiency of fast road interweaving area.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and vehicle-to-everything (V2X) collaborative control technology; specifically, it relates to a method for dynamic grouping of intelligent connected vehicles in expressway weaving zones. Background Technology

[0002] With the rapid growth of urban traffic demand, urban expressways, as high-grade urban roads connecting various urban areas, directly impact the overall traffic flow of the urban transportation system. In recent years, the development of autonomous driving technology and the Internet of Vehicles (IoV) has provided new solutions for the intelligent management of urban expressways. Connected and Autonomous Vehicles (CAVs), through real-time communication between vehicles and between vehicles and roads, can alleviate traffic congestion and reduce accident rates to some extent. However, in urban expressways, weaving zones are one of the core bottlenecks restricting road capacity and safety. In typical urban expressway weaving zones, entrance and exit ramps are close together and often adjacent. All vehicles entering and exiting the expressway need to complete at least one lateral lane change within a limited space, while vehicles on multi-lane main roads also frequently change lanes to maintain high speeds. This results in a high degree of overlap between lateral lane changes and longitudinal acceleration / deceleration behaviors within the weaving zone, forming a complex traffic flow structure with multiple objectives and conflict points. Therefore, how to achieve coordinated control of multiple CAVs in the weaving zone to ensure traffic safety and efficiency is a key research focus and challenge in the field of intelligent transportation.

[0003] Currently, research on group-based cooperative control of intelligent connected vehicles often simplifies the research scenario to a merging zone model with a single entrance or exit, or equates the main road traffic flow to a single-lane flow model. For example, the existing literature "Group-based Cooperative Control of Merging Zones on Intelligent Connected Highways" addresses traffic congestion in merging zones on highways, balancing traffic efficiency and model solution complexity. It proposes a group-based cooperative merging control method for intelligent connected vehicles, based on K-means clustering to group all ungrouped vehicles on the main road and ramps within the detection area; secondly, it determines the passage order between groups according to the first-in-first-out (FIFO) rule. Another existing literature, "Research on Cooperative Control of Ramp Merging for Intelligent Connected Vehicles," conducts research on multi-vehicle cooperative planning and control, multi-vehicle queue merging control, and vehicle lane-changing control in ramp areas. It divides the ramp merging problem into control problems under normal traffic flow and peak traffic flow, proposing multi-vehicle cooperative control schemes for both scenarios, improving vehicle traffic efficiency and reducing energy consumption. The existing paper, "APlatoon-Based Hierarchical Merging Control for On-Ramp Vehicles Under Connected Environment," proposes a queue-based hierarchical merging control algorithm for inbound ramp vehicles to achieve automatic merging control in connected traffic environments. The proposed algorithm optimizes the merging strategy of inbound ramp vehicles to smooth their merging trajectory without frequent deceleration or stopping at the ramp end, and minimizes the interference of the merging zone on mainline traffic. While this type of method has some theoretical value, it neglects the characteristics of multi-objective, multi-conflict point, and multi-vehicle interactions in actual multi-lane weaving zones, especially failing to fully consider the impact of frequent lane changes by mainline vehicles on merging and exiting vehicles, resulting in limited effectiveness of the model in practical applications.

[0004] Chinese patent CN120014879A discloses a "Control Method for Intelligent Connected Vehicles in Expressway Weaving Zones with Mixed Driving by Humans and Machines". This invention constructs a lane-changing intention prediction model for expressway weaving zones based on CNN, LSTM, and attention mechanisms. It effectively and accurately predicts the lane-changing intentions of intelligent connected vehicles in expressway weaving zones and controls the vehicles to execute corresponding lane-changing strategies. However, it fails to truly consider the actual situation of multiple lanes and multiple vehicles in expressway weaving zones. Summary of the Invention

[0005] In view of the above problems, the purpose of this invention is to provide a dynamic grouping method for intelligent connected vehicles (CAVs) in expressway weaving zones. This method decomposes the complex global cooperative control problem of CAVs in expressway weaving zones into multiple local sub-problems through real-time dynamic grouping, simplifying the complex vehicle weaving problem and thus reducing the complexity of the solution. Compared with existing models, the significant advantage of this model is that it allows information exchange between vehicles and creates suitable gaps for target vehicles with lane-changing needs through precise acceleration or deceleration operations, thereby achieving smooth and efficient merging and exiting of vehicles. It fully utilizes time and space resources and leverages the cooperative capabilities of CAVs, thereby improving the safety and traffic efficiency in expressway weaving zones and overcoming the shortcomings of the existing technologies.

[0006] This invention provides a method for dynamic grouping of intelligent connected vehicles in expressway weaving zones, comprising the following steps:

[0007] A. Dynamic grouping rules based on longitudinal position: Vehicles are sorted according to their longitudinal position in the weaving zone, and combined with the expected safety distance and group size constraints, the vehicles are dynamically divided into several cooperative control groups, and the grouping results are dynamically updated at a fixed update cycle.

[0008] B. Dynamic packet update cycle adjustment based on communication link quality: The update cycle of vehicle packets is adaptively adjusted by real-time monitoring of vehicle network communication indicators;

[0009] C. Intra-group cooperative control strategy: Within the same vehicle group, by identifying lane-changing needs and establishing safety constraints and optimization objective functions, the distributed cooperative control method is used to coordinate the longitudinal movement of adjacent vehicles, creating safe gaps for lane-changing vehicles, and achieving smooth and safe merging and splitting;

[0010] D. Inter-group collaborative control strategy: For the interaction between different vehicle groups, a control barrier function is introduced to constrain the safe distance between groups, and it is embedded into the quadratic programming solution framework to realize safe collaborative trajectory planning between multiple vehicle groups.

[0011] E. Calibration of Grouping Method for Weaving Zones on Real-World Expressways: Real traffic data is collected using drone aerial video, vehicle motion information is extracted through target detection and trajectory tracking algorithms, and parameters of the dynamic grouping model are calibrated and verified based on this.

[0012] As a preferred embodiment of the present invention, step A further includes the following step:

[0013] A1. Construction of the coordinate system for the weaving area: The coordinate system is constructed with the starting point of the expressway weaving area as the origin of the coordinate system;

[0014] A2. Vehicle coordinate settings: Set the coordinates of each vehicle in the interlacing area coordinate system to... ,in These are the vehicle's longitudinal position, velocity, and acceleration. These are the vehicle's lateral position, velocity, and acceleration;

[0015] A3. The current set of vehicles participating in grouping within the weaving zone is: The total number of its vehicles is All vehicles are sorted according to their longitudinal coordinates, with the sorting direction from largest to smallest longitudinal coordinate, to reflect the relative order of the vehicles in the direction of travel.

[0016] A4. Assign the vehicle with the largest vertical coordinate to Group 1, and define the maximum number of vehicles within each group as... ;

[0017] A5. A method for calculating safe distance based on intelligent driver model and vehicle elliptical model;

[0018] A5.1. Desired Gap Based on Intelligent Driver Model The definition is as follows:

[0019]

[0020] in: For a safe time interval, For vehicles At any moment speed, For at any time vehicles Speed ​​difference with the vehicle in front It is the first Maximum comfort acceleration of the vehicle, It is the first The vehicle's maximum comfort deceleration;

[0021] A5.2. Use an elliptical vehicle model for calibration. Its expression is as follows:

[0022]

[0023] in: : Vehicle length, Vehicle width, Driver type parameter, Bicycle speed, Speed ​​of the vehicle in front;

[0024] A6. For vehicle 2, if the longitudinal distance between vehicle 1 and vehicle 2 is less than... And the number of vehicles in group 1 is less than If so, then vehicle 2 is part of group 1; otherwise, vehicle 2 belongs to group 2.

[0025] A7. For vehicle 3, if the longitudinal distance between vehicle 2 and vehicle 3 is less than... ， And the number of vehicles in group 2 is less than If vehicle 3 is grouped with vehicle 2, then vehicle 3 is grouped together; otherwise, vehicle 3 is grouped into the next group.

[0026] A8. Regarding vehicles Repeat steps A5-A7 until all vehicles have been traversed once, at which point the grouping process ends.

[0027] A9. Repeat steps A3–A8, updating the vehicle grouping results every Tupdate time.

[0028] As a preferred embodiment of the present invention, step B further includes the following step:

[0029] B1. Communication link quality detection, detecting the strength of the received signal RSSI;

[0030] B2. Obtaining the Packet Loss Rate (PLR) metric;

[0031] B3. Obtaining the Topological Change Rate (TCR) index;

[0032] B4. Determination of communication link degradation;

[0033] B5. Dynamic update cycle calculation: To balance real-time performance and computational burden, the group update cycle Tupdate is adaptively calculated using the following formula;

[0034]

[0035] in: : This refers to a dynamic update cycle. : These are adaptive adjustment coefficients, used to control the sensitivity of RSSI, PLR, and TCR to period adjustment, respectively;

[0036] B6. Feedback and Closed-Loop Processing: After completing the vehicle grouping update, the real-time communication index acquisition module dynamically monitors the vehicle-to-everything (V2X) communication quality within the weaving zone. Based on the changes in indicators before and after the update, a closed-loop optimization strategy is executed. To evaluate the effectiveness of the grouping update, the relative rate of change of communication indicators is defined. ;

[0037] B7. Parameter adaptive optimization: If the communication indicators do not improve after group updates, the sensitivity coefficient is dynamically adjusted. The updated adaptive adjustment parameters are as follows:

[0038]

[0039]

[0040]

[0041] in: : This is the adaptive step size, with a range of values. , : for the first The absolute change value of the indicator, ， : These are the updated adaptive adjustment coefficients, used to control the sensitivity of RSSI, PLR, and TCR to periodic adjustments, respectively.

[0042] As a preferred embodiment of the present invention, step B4 further includes the following step:

[0043] B4.1. The formula for the dynamic adaptive period Tupdate is as follows:

[0044]

[0045] in: : For dynamic adaptive period, and These are the minimum and maximum update periods, respectively. , 、 These are the thresholds for triggering packet updates of received signal strength, packet loss rate, and topology change rate, respectively, to determine link degradation. 、 、 These are the thresholds for triggering packet updates of received signal strength, packet loss rate, and topology change rate, respectively, to determine link stability. For adaptive adjustment coefficients;

[0046] B4.2. When communication links deteriorate or traffic conditions are highly unstable, or or This will immediately trigger group updates ahead of schedule.

[0047] B4.3. When the communication link is good and the traffic flow is stable, and and Extend the group update cycle to make it close to ;

[0048] B4.4. Transitional State Handling: When communication quality is between deteriorating and good, or traffic conditions are between highly unstable and stable, the vehicle packet update cycle is... and Continuous adjustment within the range, achieving a smooth transition through adaptive adjustment.

[0049] As a preferred embodiment of the present invention, step C further includes the following step:

[0050] C1. Lane change requirement identification: Determine the conflicts between vehicles in the group that have lane change requirements and those that do not, as well as the conflicts between vehicles that have lane change requirements. For vehicles that do not have lane change requirements, their lateral movement remains 0. For vehicles that have lane change requirements, determine whether the vehicle should change lanes by detecting the vehicle coordinate difference.

[0051] C1.1. Vehicles At any moment Position coordinates Represented as:

[0052]

[0053] C1.2. Lateral offset calculation: Determine whether the vehicle needs to change lanes by measuring the difference between the vehicle's lateral position and the center line of the target lane.

[0054]

[0055] in: : for the first The lateral offset of the vehicle. : represents the lateral coordinate of the target lane's centerline. : for the first The vehicle's current lateral position;

[0056] C2. Intra-group safety constraints: For safe driving, vehicles in the same group must meet the following constraints:

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063]

[0064] in, For vehicles At any moment speed, For vehicles At any moment Velocity in the lateral position, Representative vehicle At any moment Acceleration in the longitudinal position, Representative in vehicle At any moment Acceleration in the lateral position, Representative vehicle At any moment Impact intensity in the longitudinal direction, Representative vehicle At any moment Impact intensity in the lateral position, This represents the maximum impact force in the vertical direction. This represents the maximum impact force in the lateral direction. Representative vehicle With vehicles Longitudinal position difference between , Represents the width of the vehicle;

[0065] C3. Intra-group vehicle coordination control to ensure lane-changing vehicles In groups To ensure a safe and smooth lane change, the system needs to identify nearby vehicle groups. It also coordinates the longitudinal movement of vehicles within the group, proactively creating the necessary safety gaps for lane-changing vehicles, thereby reducing the risk of lane-changing conflicts and improving the operating efficiency of the weaving area.

[0066] As a preferred embodiment of the present invention, step C3 further includes the following step:

[0067] C3.1. Control Objective 1: Actively accelerate or decelerate, adjust and change lanes for vehicles. The longitudinal spacing;

[0068]

[0069] in: : is the desired safe distance;

[0070] C3.2. Control Objective 2: While meeting traffic safety constraints, reduce energy consumption and unnecessary speed fluctuations;

[0071]

[0072] in: : This is the longitudinal acceleration control input;

[0073] C3.3. Control Objective 3: To create the necessary safe clearance for lane changing and ensure that there is no risk of collision when lane-changing vehicles are executing lateral trajectories;

[0074]

[0075]

[0076]

[0077] in: The longitudinal distance between a vehicle changing lanes and the adjacent vehicle in front. The longitudinal distance between a vehicle changing lanes and its adjacent vehicle behind it. Safe distance;

[0078] C3.4. The three sub-objectives are weighted and combined to form a comprehensive optimization problem for intra-group vehicle cooperative control:

[0079]

[0080] in: 、 、 : These are the objective functions for longitudinal spacing, energy consumption, and safety clearance, respectively. : These are weighting coefficients, corresponding to longitudinal spacing, energy consumption, and safety clearance, respectively. : This is the comprehensive optimization function that integrates the three sub-objectives.

[0081] As a preferred embodiment of the present invention, step D further includes the following step:

[0082] D1. Definition of safe distance between adjacent vehicle groups: For group 1, the constraints are satisfied; for group 2, trajectory planning must consider the safe distance between the trajectories of group 1 and group 2; for group 2... Group should be with the first To maintain a safe distance and achieve coordinated trajectory planning across the entire weaving zone, it is necessary to consider the safe distance between the first and last vehicles in adjacent groups, which is defined as follows: That is, the first The last car in the group, That is, the first The first car in the group, if and only if ,and, activation;

[0083] D2. Control Barrier Function Modeling: To ensure safety, a control barrier function (CBF) constraint is introduced to describe the safety distance.

[0084] To avoid collisions and unsafe path overlaps, the safe distance between vehicle groups is defined by the following formula:

[0085]

[0086]

[0087] in, For vehicles and The actual distance, For the first Group and No. safe distance between groups Let this be the barrier function, and substitute it into the control barrier function CBF constraint:

[0088]

[0089] in: : Buffer coefficient, : is the derivative of the barrier function, if This indicates that you are approaching an unsafe area. This indicates that they are moving away from an unsafe area. Barrier function right The partial derivatives, for The time derivative;

[0090] D3. Embed the barrier function CBF constraint into the QP solver. After the barrier function CBF constraint is linearized, it is directly embedded into the QP solver for solving. The barrier function CBF constraint provides proof of the non-emptiness of the feasible region, which is used to ensure that the optimization problem always has a solution and avoid conflict situations.

[0091] As a preferred embodiment of the present invention, step E further includes the following step:

[0092] E1. Drone video recording;

[0093] E2. Image preprocessing;

[0094] E3. YOLOv8 and Deep-SORT-based detection and vehicle tracking algorithms: YOLOv8 and Deep-SORT are used to detect and track vehicle targets in videos to achieve continuous tracking of vehicle trajectories.

[0095] E4. Lane line recognition and lane marking of the vehicle: Use the marking tool to recognize lane lines;

[0096] E5. Vehicle speed and acceleration acquisition: After obtaining the continuous trajectory of the vehicle in the real coordinate system, the longitudinal and lateral velocities of the vehicle are calculated based on the displacement changes between adjacent video frames, and the vehicle's acceleration is further obtained.

[0097] E6. Calibrate the vehicle dynamic grouping model. Using the real-world interleaving zone data obtained in steps E1-E5, calibrate the vehicle dynamic grouping model.

[0098] As a preferred embodiment of the present invention, step E2 further includes the following step:

[0099] E2.1. Remove rotation and offset by using After Effects software to preprocess the video data;

[0100] E2.2. Image cropping: After the video stabilization is completed, AfterEffects should be used to crop out irrelevant areas, keeping only the interlaced areas to speed up the extraction process.

[0101] The beneficial effects of this invention are as follows:

[0102] 1. This invention addresses the issue that existing research often focuses on idealized traffic models with single-ramp merging or single lanes. This solution, for the first time, uses urban expressway weaving zones as the core research scenario, comprehensively considering the multi-objective, multi-conflict point, and multi-vehicle interactions of vehicles on multi-lane main roads within the weaving zone. It introduces realistic geometric modeling, establishing a two-dimensional coordinate system based on the starting point of the weaving zone, defining longitudinal and lateral positions to ensure the model has higher practical application value. This solves the problem that traditional single-objective, single-entry merging control methods cannot cope with the realities of complex urban expressway weaving zones.

[0103] 2. This invention dynamically determines grouping relationships based on longitudinal distance and desired gaps, and combines this with an elliptical vehicle geometry model to describe potential risk areas, making grouping boundaries more consistent with actual traffic flow dynamics. It decomposes the complex global collaborative control problem in weaving areas into multiple local sub-problems, significantly reducing computational complexity and communication requirements. The number of vehicles N within a group is controllable, ensuring manageable and real-time grouping size. Unlike traditional fixed grouping or K-means clustering, this method can respond in real-time to changes in vehicle position and speed, achieving continuous dynamic adjustment of groupings.

[0104] 3. Under the background of vehicle-to-everything (V2X) cooperative control, this invention incorporates communication link quality indicators (RSSI, PLR, TCR) into the optimization of the packet update cycle. Based on these three indicators, the packet update time (Tupdate) is dynamically calculated, triggering packet updates in advance when communication deteriorates or traffic becomes highly dynamic. When traffic is stable, the cycle is extended to reduce computational and communication overhead. This ensures that packets always maintain real-time performance and reliability, solving the problem that traditional fixed update cycles cannot adapt to fluctuations in the V2X network.

[0105] 4. This invention introduces a Control Barrier Function (CBF) between groups, using formal mathematical methods to ensure that different groups maintain a safe distance in both the longitudinal and lateral directions. The CBF constraints are linearized and embedded into the QP solver, ensuring that the global trajectory optimization problem always has a solution and avoiding vehicle collisions. Compared to traditional hard constraints, CBF provides a proof of the non-emptiness of the feasible region, solving the problem of optimization infeasibility caused by inter-group conflicts, achieving global safety assurance across groups, and making the coordinated control of the entire interleaving zone more stable. Attached Figure Description

[0106] Other objects and results of the invention will become more apparent and readily understood with reference to the following description taken in conjunction with the accompanying drawings. In the drawings:

[0107] Figure 1 This is a technical roadmap of an embodiment of the present invention;

[0108] Figure 2 A schematic diagram of the coordinate system grouping for the expressway weaving area constructed according to an embodiment of the present invention;

[0109] Figure 3 This is a schematic diagram of vehicle grouping based on real-world calibration in an embodiment of the present invention. Detailed Implementation

[0110] See Figure 1-3 The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0111] This invention provides a method for dynamic grouping of intelligent connected vehicles in expressway weaving zones, comprising the following steps:

[0112] A. Dynamic grouping rules based on longitudinal position: Vehicles are sorted according to their longitudinal position in the weaving zone, and combined with the expected safety distance and group size constraints, the vehicles are dynamically divided into several cooperative control groups, and the grouping results are dynamically updated at a fixed update cycle.

[0113] A1. Establishment of the coordinate system for the weaving area: The coordinate system is constructed with the starting point of the expressway weaving area as the origin, and points along the driving direction as... The axis, perpendicular to the lane direction is The axis, the interlacing zone length, refers to the distance between the triangular end at the entrance of the interlacing zone (width 0.6m) and the triangular end at the exit (width 3.7m). The specific coordinate system diagram is shown below. Figure 2 As shown;

[0114] A2. Vehicle coordinate settings: Set the coordinates of each vehicle in the interlacing area coordinate system to... ,in These are the vehicle's longitudinal position, velocity, and acceleration. These are the vehicle's lateral position, velocity, and acceleration;

[0115] A3. The current set of vehicles participating in grouping within the weaving zone is: The total number of its vehicles is All vehicles are sorted according to their longitudinal coordinates, with the sorting direction from largest to smallest longitudinal coordinate, to reflect the relative order of the vehicles in the direction of travel.

[0116] A4. Assign the vehicle with the largest vertical coordinate to Group 1, and define the maximum number of vehicles within each group as... To ensure that the number of vehicles in each group is within a reasonable range;

[0117] A5. A method for calculating safe distance based on intelligent driver model (IDM) and vehicle elliptical model;

[0118] A5.1. Desired Gap Based on Intelligent Driver Model The definition is as follows:

[0119]

[0120] in: For a safe time interval, For vehicles At any moment speed, For at any time vehicles Speed ​​difference with the vehicle in front It is the first Maximum comfort acceleration of the vehicle, It is the first The vehicle's maximum comfort deceleration, 、 ;

[0121] A5.2. Use an elliptical vehicle model for calibration. Based on the vehicle's geometric space model, the vehicle elliptical model fully considers vehicle speed and driver type factors during the modeling process. Its major axis dynamically adjusts with changes in driving speed and driver type parameters. Compared to the vehicle rectangular model, vehicle circular model, and vehicle dynamic circular model, the vehicle elliptical model can more accurately reflect the potential risk characteristics of the vehicle under different driving scenarios. Its expression is shown below:

[0122]

[0123] in: : Length of the vehicle (m) Vehicle width (m) Driver type parameter, Vehicle speed (km / h) Speed ​​of the vehicle in front (km / h);

[0124] A6. For vehicle 2, if the longitudinal distance between vehicle 1 and vehicle 2 is less than... And the number of vehicles in group 1 is less than If so, then vehicle 2 is part of group 1; otherwise, vehicle 2 belongs to group 2.

[0125] A7. For vehicle 3, if the longitudinal distance between vehicle 2 and vehicle 3 is less than... ， And the number of vehicles in group 2 is less than If vehicle 3 is grouped with vehicle 2, then vehicle 3 is grouped together; otherwise, vehicle 3 is grouped into the next group.

[0126] A8. Regarding vehicles Repeat steps A5-A7 until all vehicles have been traversed once, at which point the grouping process ends.

[0127] A9. Repeat steps A3–A8, updating the vehicle grouping results every Tupdate (dynamic update cycle). The updated diagram of the expressway weaving zone grouping is shown below. Figure 3 As shown.

[0128] B. Dynamic packet update cycle adjustment based on communication link quality: The update cycle of vehicle packets is adaptively adjusted by real-time monitoring of vehicle network communication indicators (RSSI, PLR, TCR).

[0129] B1. Communication link quality detection, detecting the strength of the received signal RSSI.

[0130] RSSI (Received Signal Strength Indicator) is a key indicator for measuring the quality of wireless links between vehicles, usually expressed in dBm. Its value is closely related to the distance between vehicles, obstacles, and the degree of interference. The higher the RSSI, the stronger the communication signal and the more stable the link; conversely, the lower the RSSI, the worse the link quality.

[0131] B2. Packet Loss Ratio (PLR) metric acquisition, the specific calculation formula is as follows:

[0132]

[0133] in: Packet loss rate, value range , The number of data packets lost within a unit of time period. The number of data packets sent within a unit of time period;

[0134] B3. The Topology Change Rate (TCR) metric is obtained, and the specific calculation formula is as follows:

[0135]

[0136] in: The number of vehicles in the weaving zone at the current moment. The number of vehicles in the weaving zone at the previous moment. The time interval between two adjacent detection times;

[0137] B4. Determination of communication link degradation

[0138] B4.1. In interleaved environments, as RSSI decreases, the signal-to-noise ratio drops, making vehicles more susceptible to data packet loss due to interference or collisions, leading to an increase in PLR. A high PLR signifies unreliable transmission of cooperative control information in vehicle-to-everything (V2X) networks, manifesting as information delays or even data loss.

[0139] Information delays can disrupt the consistency and real-time performance of convoy trajectory planning in cooperative driving, thereby reducing the safety and operational efficiency of vehicle groups merging and diverging in weaving zones. The dynamic adaptive period Tupdate formula is as follows:

[0140]

[0141] in: : For dynamic adaptive period, and These are the minimum and maximum update periods, respectively. , 、 These are the thresholds for triggering packet updates of received signal strength, packet loss rate, and topology change rate, respectively, to determine link degradation. 、 、 These are the thresholds for triggering packet updates of received signal strength, packet loss rate, and topology change rate, respectively, to determine link stability. For adaptive adjustment coefficients;

[0142] B4.2. When communication links deteriorate or traffic conditions are highly unstable, or or Immediately trigger group updates in advance, and through group adjustments, ensure that collaborative control information can maintain real-time performance and consistency even under high-risk conditions;

[0143] B4.3. When the communication link is good and the traffic flow is stable, and and The system extends the group update cycle to make it close to While maintaining security, it reduces unnecessary communication burden and computing overhead;

[0144] B4.4. Transitional State Handling: When communication quality is between deteriorating and good, or traffic conditions are between highly unstable and stable, the vehicle packet update cycle is... and Continuous adjustment within the range, and smooth transition through adaptive adjustment, avoids frequent switching that could lead to unstable control strategies.

[0145] B5. Dynamic update cycle calculation: To balance real-time performance and computational burden, the group update cycle Tupdate is adaptively calculated using the following formula;

[0146]

[0147] in: : This refers to a dynamic update cycle. : These are adaptive adjustment coefficients, used to control the sensitivity of RSSI, PLR, and TCR to period adjustment, respectively;

[0148] B6. Feedback and Closed-Loop Processing: After completing the vehicle grouping update, the system dynamically monitors the vehicle-to-everything (V2X) communication quality within the weaving zone through a real-time communication indicator acquisition module. Based on the changes in indicators before and after the update, a closed-loop optimization strategy is executed. To evaluate the effectiveness of the grouping update, the relative rate of change of communication indicators is defined. The calculation formula is as follows:

[0149]

[0150] in: : for the first The value of the communication indicator, , : This refers to the updated indicator value. The indicator values ​​before the update. Group update before and after The relative rate of change of each indicator;

[0151] B7. Parameter adaptive optimization: If communication metrics do not improve after group updates, the system dynamically adjusts the sensitivity coefficient. The updated adaptive adjustment parameters are as follows:

[0152]

[0153]

[0154]

[0155] in: : This is the adaptive step size, with a range of values. , : for the first The absolute change value of the indicator, ， : These are the updated adaptive adjustment coefficients, used to control the sensitivity of RSSI, PLR, and TCR to periodic adjustments, respectively.

[0156] C. Intra-group Cooperative Control Strategy: Within the same vehicle group, by identifying lane-changing needs and establishing safety constraints and optimization objective functions, a distributed cooperative control method is used to coordinate the longitudinal movement of adjacent vehicles, creating safe gaps for lane-changing vehicles and achieving smooth and safe merging and splitting; the distributed cooperative control method is calculated as follows:

[0157]

[0158] in: For vehicles The control input, , vehicles and the position of the vehicle in front of it, , vehicles and the speed of the car in front, For the desired vehicle spacing, To control the gain coefficient;

[0159] C1. Lane change requirement identification: Determine the conflicts between vehicles with lane change requirements and those without, as well as the conflicts between vehicles with lane change requirements. For vehicles without lane change requirements, their lateral movement remains 0. For vehicles with lane change requirements, their lateral trajectory is often a smooth curve. The vehicle coordinate difference is detected to determine whether the vehicle needs to change lanes.

[0160] C1.1. Vehicles At any moment Position coordinates Represented as:

[0161]

[0162] C1.2. Lateral offset calculation: Determine whether the vehicle needs to change lanes by measuring the difference between the vehicle's lateral position and the center line of the target lane.

[0163]

[0164] in: : for the first The lateral offset of the vehicle. : represents the lateral coordinate of the target lane's centerline. : for the first The vehicle's current lateral position;

[0165] C2. Intra-group safety constraints: For safe driving, vehicles in the same group must meet the following constraints:

[0166]

[0167]

[0168]

[0169]

[0170]

[0171]

[0172]

[0173] in, For vehicles At any moment speed, For vehicles At any moment Velocity in the lateral position, Representative vehicle At any moment Acceleration in the longitudinal position, Representative in vehicle At any moment Acceleration in the lateral position, Representative vehicle At any moment Impact intensity in the longitudinal direction, Representative vehicle At any moment Impact intensity in the lateral position, This represents the maximum impact force in the vertical direction. This represents the maximum impact force in the lateral direction. Representative vehicle With vehicles Longitudinal position difference between , Represents the width of the vehicle;

[0174] C3. Intra-group vehicle coordination control to ensure lane-changing vehicles In groups To ensure a safe and smooth lane-changing operation, the system needs to identify nearby vehicle groups. It also coordinates the longitudinal movement of vehicles within the group, proactively creating the necessary safety gaps for lane-changing vehicles, thereby reducing the risk of lane-changing conflicts and improving the operating efficiency of the weaving area.

[0175] For each neighboring vehicle The control objectives are as follows: D3.1-D3.3;

[0176] C3.1. Control Objective 1: Actively accelerate or decelerate, adjust and change lanes for vehicles. The longitudinal spacing;

[0177]

[0178] in: : is the desired safe distance;

[0179] C3.2. Control Objective 2: While meeting traffic safety constraints, minimize energy consumption and unnecessary speed fluctuations;

[0180]

[0181] in: : This is the longitudinal acceleration control input;

[0182] C3.3. Control Objective 3: To create the necessary safe clearance for lane changing and ensure that there is no risk of collision when lane-changing vehicles are executing lateral trajectories;

[0183]

[0184]

[0185]

[0186] in: The longitudinal distance between a vehicle changing lanes and the adjacent vehicle in front. The longitudinal distance between a vehicle changing lanes and its adjacent vehicle behind it. Safe distance;

[0187] C3.4. The three sub-objectives are weighted and combined to form a comprehensive optimization problem for intra-group vehicle cooperative control:

[0188]

[0189] in: 、 、 : These are the objective functions for longitudinal spacing, energy consumption, and safety clearance, respectively. : These are weighting coefficients, corresponding to longitudinal spacing, energy consumption, and safety clearance, respectively. : This is the comprehensive optimization function that integrates the three sub-objectives.

[0190] D. Inter-group cooperative control strategy: To address the interaction between different vehicle groups, a control barrier function (CBF) is introduced to constrain the safe distance between groups, and it is embedded into the quadratic programming (QP) solution framework to achieve safe cooperative trajectory planning between multiple vehicle groups;

[0191] D1. Definition of safe distance between adjacent vehicle groups: For group 1, the constraints need to be met. For group 2, trajectory planning must consider the safe distance between the trajectories of group 1 and group 2. Similarly, safe distances must also be maintained between different groups. Group should be with the first To maintain a safe distance and achieve coordinated trajectory planning across the entire weaving zone, it is necessary to consider the safe distance between the first and last vehicles in adjacent groups, which is defined as follows: That is, the first The last car in the group, That is, the first The first car in the group, if and only if ,and, activation;

[0192] D2. Control Barrier Function Modeling: To ensure safety, the Control Barrier Function (CBF) constraint is introduced to describe the safety distance. The control barrier function is a mathematical tool used for safety constraint modeling and verification in control systems. It can formally ensure that the system always stays within a predefined safety set during operation.

[0193] To avoid collisions and unsafe path overlaps, the safe distance between vehicle groups is defined by the following formula:

[0194]

[0195]

[0196] in, For vehicles and The actual distance, For the first Group and No. safe distance between groups Let this be the barrier function, and substitute it into the control barrier function CBF constraint:

[0197]

[0198] in: : Buffer coefficient, : is the derivative of the barrier function, if This indicates that the system is approaching an unsafe area. This indicates that the system is moving away from an unsafe area. Barrier function right The partial derivatives, for The time derivative;

[0199] D3. Embed the barrier function CBF constraint into the QP solver. After the barrier function CBF constraint is linearized, it is directly embedded into the QP solver for solving. Use an appropriate QP solver (OSQP, qpOASES, Gurobi) to solve the QP problem to ensure that at any time, compared with traditional hard constraints, the barrier function CBF constraint provides proof of the non-emptiness of the feasible region, which is used to ensure that the optimization problem always has a solution and avoids conflict situations.

[0200] E. Calibration of Grouping Method for Weaving Zones on Real-World Expressways: Real traffic data is collected using drone aerial video, vehicle motion information is extracted through target detection and trajectory tracking algorithms, and parameters of the dynamic grouping model are calibrated and verified based on this.

[0201] E1. Drone video shooting: Data was collected on a certain area's expressway during a clear, windless day (midday). The drone video image resolution was 3840×2160, the frame rate was 30FPS, and the shooting height was 250m. It was able to completely cover the minimum height of the weaving area. The weaving area was 375m long, with four lanes in one direction, and the design speed was 80km / h.

[0202] E2. Image preprocessing

[0203] E2.1. Removing rotation and offset: When drones shoot at high altitudes, they are affected by wind direction and airflow, causing the video footage to rotate and shift. As a result, the video inevitably has slight shaking. This change in the video footage will change the pixel coordinates of the vehicle, causing it to deviate from its actual position. In order to stabilize the video and extract the vehicle trajectory, After Effects software is used to preprocess the video data.

[0204] E2.2. Image cropping: Since the interlacing area was located on an elevated road, vehicles from other unrelated roads were included in the video. These vehicles were irrelevant to the study but would seriously affect the trajectory extraction speed of YOLO. After stabilizing the video, AfterEffects was used to crop out the irrelevant areas, leaving only the interlacing area to speed up the extraction process.

[0205] E3. A vehicle detection and tracking algorithm based on YOLOv8 (object detection model) and Deep-SORT (classic multi-object tracking algorithm). The algorithm detects and tracks vehicle targets in the video using YOLOv8 and Deep-SORT. After vehicle detection, the detection results need to be combined with the multi-object tracking (MOT) algorithm to achieve continuous tracking of vehicle trajectories.

[0206] E4. Lane line recognition and vehicle lane marking: The vehicle trajectory file output by YOLOv8 only contains four columns of data: timestamp, vehicle ID, X-direction position, and Y-direction position. It is necessary to obtain the lane of any vehicle in each frame. For this purpose, the labeling tool (Labelme) is used to recognize the lane lines.

[0207] E5. Vehicle speed and acceleration acquisition: After obtaining the continuous trajectory of the vehicle in the real coordinate system, the longitudinal and lateral velocities of the vehicle are calculated based on the displacement changes between adjacent video frames, and the vehicle's acceleration is further obtained.

[0208] E6. Calibrate the vehicle dynamic grouping model. Using the real-world interleaving region data obtained in steps E1-E5, calibrate the vehicle dynamic grouping model. The drone aerial video is 45 minutes long, yielding 908 grouping results. Four representative consecutive grouping updates are selected. Figure 3 .

[0209] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for dynamic grouping of intelligent connected vehicles in a weaving zone on a fast road, characterized in that, Includes the following steps: A. Dynamic grouping rules based on longitudinal position: Vehicles are sorted according to their longitudinal position in the weaving zone, and combined with the expected safety distance and group size constraints, the vehicles are dynamically divided into several cooperative control groups, and the grouping results are dynamically updated at a fixed update cycle. B. Dynamic packet update cycle adjustment based on communication link quality: The update cycle of vehicle packets is adaptively adjusted by real-time monitoring of vehicle network communication indicators; C. Intra-group cooperative control strategy: Within the same vehicle group, by identifying lane-changing needs and establishing safety constraints and optimization objective functions, the distributed cooperative control method is used to coordinate the longitudinal movement of adjacent vehicles, creating safe gaps for lane-changing vehicles, and achieving smooth and safe merging and splitting; D. Inter-group collaborative control strategy: For the interaction between different vehicle groups, a control barrier function is introduced to constrain the safe distance between groups, and it is embedded into the quadratic programming solution framework to realize safe collaborative trajectory planning between multiple vehicle groups. E. Calibration of Grouping Method for Weaving Zones on Real-World Expressways: Real traffic data is collected using drone aerial video, vehicle motion information is extracted through target detection and trajectory tracking algorithms, and parameters of the dynamic grouping model are calibrated and verified based on this.

2. The method for dynamic grouping of intelligent connected vehicles in a weaving zone on a fast road according to claim 1, characterized in that, Step A also includes the following steps: A1. Construction of the coordinate system for the weaving area: The coordinate system is constructed with the starting point of the expressway weaving area as the origin of the coordinate system; A2. Vehicle coordinate settings: Set the coordinates of each vehicle in the interlacing area coordinate system to... ,in These are the vehicle's longitudinal position, velocity, and acceleration. These are the vehicle's lateral position, velocity, and acceleration; A3. The current set of vehicles participating in grouping within the weaving zone is: The total number of its vehicles is All vehicles are sorted according to their longitudinal coordinates, with the sorting direction from largest to smallest longitudinal coordinate, to reflect the relative order of the vehicles in the direction of travel. A4. Assign the vehicle with the largest vertical coordinate to Group 1, and define the maximum number of vehicles within each group as... ; A5. A method for calculating safe distance based on intelligent driver model and vehicle elliptical model; A5.

1. Desired Gap Based on Intelligent Driver Model The definition is as follows: ； in: For a safe time interval, For vehicles At any moment speed, For at any time vehicles Speed ​​difference with the vehicle in front It is the first Maximum comfort acceleration of the vehicle, It is the first The vehicle's maximum comfort deceleration; A5.

2. Use an elliptical vehicle model for calibration. Its expression is as follows: ； in: : Vehicle length, Vehicle width, Driver type parameter, Bicycle speed, Speed ​​of the vehicle in front; A6. For vehicle 2, if the longitudinal distance between vehicle 1 and vehicle 2 is less than... And the number of vehicles in group 1 is less than If so, then vehicle 2 is part of group 1; otherwise, vehicle 2 belongs to group 2. A7. For vehicle 3, if the longitudinal distance between vehicle 2 and vehicle 3 is less than... ， And the number of vehicles in group 2 is less than If vehicle 3 is grouped with vehicle 2, then vehicle 3 is grouped together; otherwise, vehicle 3 is grouped into the next group. A8. Regarding vehicles Repeat steps A5-A7 until all vehicles have been traversed once, at which point the grouping process ends. A9. Repeat steps A3–A8, updating the vehicle grouping results every Tupdate time.

3. The method for dynamic grouping of intelligent connected vehicles in a weaving zone on a fast road according to claim 1, characterized in that, Step B also includes the following steps: B1. Communication link quality detection, detecting the strength of the received signal RSSI; B2. Obtaining the Packet Loss Rate (PLR) metric; B3. Obtaining the Topological Change Rate (TCR) index; B4. Determination of communication link degradation; B5. Dynamic update cycle calculation: To balance real-time performance and computational burden, the group update cycle Tupdate is adaptively calculated using the following formula; ； in: : This refers to a dynamic update cycle. : These are adaptive adjustment coefficients, used to control the sensitivity of RSSI, PLR, and TCR to period adjustment, respectively; B6. Feedback and Closed-Loop Processing: After completing the vehicle grouping update, the real-time communication index acquisition module dynamically monitors the vehicle-to-everything (V2X) communication quality within the weaving zone. Based on the changes in indicators before and after the update, a closed-loop optimization strategy is executed. To evaluate the effectiveness of the grouping update, the relative rate of change of communication indicators is defined. ; B7. Parameter adaptive optimization: If the communication indicators do not improve after group updates, the sensitivity coefficient is dynamically adjusted. The updated adaptive adjustment parameters are as follows: ； ； ； in: : This is the adaptive step size, with a range of values. , : for the first The absolute change value of the indicator, ， : These are the updated adaptive adjustment coefficients, used to control the sensitivity of RSSI, PLR, and TCR to periodic adjustments, respectively.

4. The method for dynamic grouping of intelligent connected vehicles in a weaving zone on a fast road according to claim 1, characterized in that, Step B4 also includes the following steps: B4.

1. The formula for the dynamic adaptive period Tupdate is as follows: ； in: : For dynamic adaptive period, and These are the minimum and maximum update periods, respectively. , 、 These are the thresholds for triggering packet updates of received signal strength, packet loss rate, and topology change rate, respectively, to determine link degradation. 、 、 These are the thresholds for triggering packet updates of received signal strength, packet loss rate, and topology change rate, respectively, to determine link stability. For adaptive adjustment coefficients; B4.

2. When communication links deteriorate or traffic conditions are highly unstable, or or This will immediately trigger group updates ahead of schedule; B4.

3. When the communication link is good and the traffic flow is stable, and and Extend the group update cycle to make it close to ; B4.

4. Transitional State Handling: When communication quality is between deteriorating and good, or traffic conditions are between highly unstable and stable, the vehicle packet update cycle is... and Continuous adjustment within the range, achieving a smooth transition through adaptive adjustment.

5. The method for dynamic grouping of intelligent connected vehicles in a weaving zone on a fast road according to claim 1, characterized in that, Step C also includes the following steps: C1. Lane change requirement identification: Determine the conflicts between vehicles in the group that have lane change requirements and those that do not, as well as the conflicts between vehicles that have lane change requirements. For vehicles that do not have lane change requirements, their lateral movement remains 0. For vehicles that have lane change requirements, determine whether the vehicle should change lanes by detecting the vehicle coordinate difference. C1.

1. Vehicles At any moment Position coordinates Represented as: ； C1.

2. Lateral offset calculation: Determine whether the vehicle needs to change lanes by measuring the difference between the vehicle's lateral position and the center line of the target lane. ； in: : for the first The lateral offset of the vehicle. : represents the lateral coordinate of the target lane's centerline. : for the first The vehicle's current lateral position; C2. Intra-group safety constraints: For safe driving, vehicles in the same group must meet the following constraints: ； ； ； ； ； ； ； in, For vehicles At any moment speed, For vehicles At any moment Velocity in the lateral position, Representative vehicle At any moment Acceleration in the longitudinal position, Representative in vehicle At any moment Acceleration in the lateral position, Representative vehicle At any moment Impact intensity in the longitudinal direction, Representative vehicle At any moment Impact intensity in the lateral position, This represents the maximum impact force in the vertical direction. This represents the maximum impact force in the lateral direction. Representative vehicle With vehicles Longitudinal position difference between , Represents the width of the vehicle; C3. Intra-group vehicle coordination control to ensure lane-changing vehicles In groups To ensure a safe and smooth lane change, the system needs to identify nearby vehicle groups. It also coordinates the longitudinal movement of vehicles within the group, proactively creating the necessary safety gaps for lane-changing vehicles, thereby reducing the risk of lane-changing conflicts and improving the operating efficiency of the weaving area.

6. The method for dynamic grouping of intelligent connected vehicles in a weaving zone on a fast road according to claim 1, characterized in that, Step C3 also includes the following steps: C3.

1. Control Objective 1: Actively accelerate or decelerate, adjust and change lanes for vehicles. The longitudinal spacing; ； in: : is the desired safe distance; C3.

2. Control Objective 2: While meeting traffic safety constraints, reduce energy consumption and unnecessary speed fluctuations; ； in: : This is the longitudinal acceleration control input; C3.

3. Control Objective 3: To create the necessary safe clearance for lane changing and ensure that there is no risk of collision when lane-changing vehicles are executing lateral trajectories; ； ； ； in: The longitudinal distance between a vehicle changing lanes and the adjacent vehicle in front. The longitudinal distance between a vehicle changing lanes and its adjacent vehicle behind it. Safe distance; C3.

4. The three sub-objectives are weighted and combined to form a comprehensive optimization problem for intra-group vehicle cooperative control: ； in: 、 、 : These are the objective functions for longitudinal spacing, energy consumption, and safety clearance, respectively. : These are weighting coefficients, corresponding to longitudinal spacing, energy consumption, and safety clearance, respectively. : This is the comprehensive optimization function that integrates the three sub-objectives.

7. The method for dynamic grouping of intelligent connected vehicles in a weaving zone on a fast road according to claim 1, characterized in that, Step D also includes the following steps: D1. Definition of safe distance between adjacent vehicle groups: For group 1, the constraints are satisfied; for group 2, trajectory planning must consider the safe distance between the trajectories of group 1 and group 2; for group 2... Group should be with the first To maintain a safe distance and achieve coordinated trajectory planning across the entire weaving zone, it is necessary to consider the safe distance between the first and last vehicles in adjacent groups, which is defined as follows: That is, the first The last car in the group, That is, the first The first car in the group, if and only if ,and, activation; D2. Control Barrier Function Modeling: To ensure safety, a control barrier function (CBF) constraint is introduced to describe the safety distance. To avoid collisions and unsafe path overlaps, the safe distance between vehicle groups is defined by the following formula: ； ； in, For vehicles and The actual distance, For the first Group and No. safe distance between groups Let this be the barrier function, and substitute it into the control barrier function CBF constraint: ； in: : Buffer coefficient, : is the derivative of the barrier function, if This indicates that you are approaching an unsafe area. This indicates that they are moving away from an unsafe area. Barrier function right The partial derivatives, for The time derivative; D3. Embed the barrier function CBF constraint into the QP solver. After the barrier function CBF constraint is linearized, it is directly embedded into the QP solver for solving. The barrier function CBF constraint provides proof of the non-emptiness of the feasible region, which is used to ensure that the optimization problem always has a solution and avoid conflict situations.

8. The method for dynamic grouping of intelligent connected vehicles in a weaving zone on a fast road according to claim 1, characterized in that, Step E also includes the following steps: E1. Drone video shooting, E2. Image preprocessing E3. YOLOv8 and Deep-SORT-based detection and vehicle tracking algorithms: YOLOv8 and Deep-SORT are used to detect and track vehicle targets in videos to achieve continuous tracking of vehicle trajectories. E4. Lane line recognition and lane marking of the vehicle: Use the marking tool to recognize lane lines; E5. Vehicle speed and acceleration acquisition: After obtaining the continuous trajectory of the vehicle in the real coordinate system, the longitudinal and lateral velocities of the vehicle are calculated based on the displacement changes between adjacent video frames, and the vehicle's acceleration is further obtained. E6. Calibrate the vehicle dynamic grouping model. Using the real-world interleaving zone data obtained in steps E1-E5, calibrate the vehicle dynamic grouping model.

9. A method for dynamic grouping of intelligent connected vehicles in a weaving zone on a fast road according to claim 1, characterized in that, Step E2 also includes the following steps: E2.

1. Remove rotation and offset by using After Effects software to preprocess the video data; E2.

2. Image cropping: After the video stabilization is completed, AfterEffects should be used to crop out irrelevant areas, keeping only the interlaced areas to speed up the extraction process.

Images