A multi-queue vehicle bidirectional lane-changing cooperative control method and system

Abstract

The application discloses a kind of multi-queue vehicle bidirectional lane-changing cooperative control method and system, the method includes: determining lane-changing mode based on the lane-changing request information of target vehicle, and constructing bidirectional time-varying communication topology based on lane-changing mode;Based on the driving state information of target vehicle at current time, update original vehicle dynamics model, obtain target vehicle dynamics model;Based on bidirectional time-varying communication topology and target vehicle dynamics model, construct bidirectional lane-changing distributed model predictive control framework;The weight parameter in bidirectional lane-changing distributed model predictive control framework is updated and handled using weight online adaptive algorithm based on gradient tracking, and the optimal control sequence of each vehicle node is obtained;According to the optimal control sequence of each vehicle node, control corresponding vehicle to execute longitudinal and lateral cooperative motion, complete the bidirectional lane-changing operation of multi-queue vehicle, effectively improve the safety margin of bidirectional lane-changing, shorten the lane-changing time.

Description

Technical Field

[0001] This invention relates to the field of intelligent connected vehicles and autonomous driving technology, and relates to, but is not limited to, a method and system for bidirectional lane-changing cooperative control of multi-queue vehicles. Background Technology

[0002] With the rapid development of intelligent connected vehicles and autonomous driving technologies, autonomous vehicle platooning control has become a core research direction for improving road traffic efficiency and driving safety due to its significant advantages such as shortening following distance, improving road capacity, reducing fuel consumption, and minimizing human error. During platooning operations, vehicles often perform lane-changing operations due to urgent needs such as navigation, obstacle avoidance, and merging / exiting. The stability of lane-changing control directly determines the overall safety and traffic efficiency of platooning operations.

[0003] Existing autonomous vehicle platooning lane change control technologies are mostly based on the Distributed Model Predictive Control (DMPC) framework. This framework has become the mainstream technical solution for autonomous vehicle platooning lane change control due to its inherent advantage in handling multi-constraint and multi-objective optimization problems. However, existing research mainly focuses on single lane-changing scenarios with one-way entry or exit. In more complex road sections such as highway main ramp connection areas and service area entrances and exits, vehicle entry and exit behaviors often occur simultaneously, forming a spatiotemporally coupled bidirectional lane-changing scenario. In this case, existing technologies have the following shortcomings: First, they cannot dynamically adapt to the communication topology changes caused by the addition or removal of vehicle nodes, and lack a unified bidirectional time-varying topology modeling framework; Second, existing DMPC methods mostly use fixed weight matrices, which cannot adapt to the dynamic changes in the priority of various performance indicators (such as tracking accuracy, driving comfort, and coordination) at different stages of lane changing; if the weight matrix is ​​simply used as the optimization variable, it may degenerate into a zero matrix due to the use of only semidefinite constraints, causing the corresponding performance indicators to completely fail; Third, existing distributed optimization algorithms (such as ADMM) have slow convergence speed and require the maintenance of dual variables, resulting in insufficient real-time control; Fourth, when the trajectories of entering and exiting vehicles overlap in time and space, there is a lack of targeted cross-interference safety constraints. The aforementioned issues mean that existing technologies cannot meet the stable control requirements for multiple queues of vehicles simultaneously performing bidirectional lane-changing behaviors such as lane entry and exit. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method and system for coordinated control of bidirectional lane changing for multi-queue vehicles, aiming to overcome the deficiencies of the prior art, improve the adaptability of communication topology, and meet the control requirements when multiple queue vehicles simultaneously perform bidirectional lane changing behaviors of cutting in and cutting out.

[0005] The specific technical solutions of this invention are as follows: In a first aspect, embodiments of the present invention provide a multi-queue vehicle bidirectional lane-changing cooperative control method, comprising: The system acquires real-time driving status information and lane-change request information of target vehicles in multiple queues; determines the lane-change mode based on the lane-change request information of the target vehicles, and constructs a bidirectional time-varying communication topology based on the lane-change mode; updates the original vehicle dynamics model based on the driving status information of the target vehicles at the current moment to obtain the target vehicle dynamics model; constructs a bidirectional lane-change distributed model predictive control framework based on the bidirectional time-varying communication topology and the target vehicle dynamics model; updates the weight parameters in the bidirectional lane-change distributed model predictive control framework using a gradient tracking-based online adaptive weight algorithm to obtain the optimal control sequence for each vehicle node; and controls the corresponding vehicle to perform longitudinal and lateral coordinated motion according to the optimal control sequence of each vehicle node to complete the bidirectional lane-change operation of multiple queues of vehicles.

[0006] Secondly, embodiments of the present invention provide a multi-queue vehicle bidirectional lane-changing cooperative control system, including an information acquisition module, a topology construction module, a model update module, a framework establishment module, an optimal control sequence acquisition module, and a lane-changing operation control module, wherein: The information acquisition module is used to acquire in real time the driving status information and lane change request information of the target vehicle in multiple queues of vehicles; The topology construction module is used to determine the lane-changing mode based on the lane-changing request information of the target vehicle, and to construct a two-way time-varying communication topology based on the lane-changing mode. The model update module is used to update the original vehicle dynamics model based on the driving state information of the target vehicle at the current moment, and obtain the target vehicle dynamics model; The framework establishment module is used to construct a bidirectional lane-changing distributed model predictive control framework based on the bidirectional time-varying communication topology and the target vehicle dynamics model. The optimal control sequence acquisition module is used to update the weight parameters in the bidirectional lane-changing distributed model predictive control framework using a gradient tracking-based online adaptive weight algorithm to obtain the optimal control sequence for each vehicle node. The lane-changing operation control module is used to control the corresponding vehicles to perform longitudinal and lateral coordinated movements according to the optimal control sequence of each vehicle node, so as to complete the bidirectional lane-changing operation of multiple queues of vehicles.

[0007] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this embodiment of the invention, firstly, the driving status information and lane-change request information of the target vehicle in multiple queues are acquired in real time, and the lane-change mode is determined based on the lane-change request information to dynamically construct a bidirectional time-varying communication topology. This topology evolution rules are uniformly described under normal mode, cut-in mode, cut-out mode, and bidirectional simultaneous operation mode (i.e., cut-in and cut-out bidirectional mode). A bypass communication link is constructed to ensure uninterrupted information flow during topology reconstruction, supporting simultaneous cut-in and cut-out operations. Then, a target vehicle dynamics model containing Coriolis coupling terms is constructed based on the current longitudinal and lateral state information, significantly improving the model prediction accuracy under high-speed lane-change conditions. Next, a bidirectional lane-change distributed model prediction control framework is constructed, introducing cross-interference safety constraints to achieve spatiotemporal isolation of the trajectories of the cutting-in and cutting-out vehicles, effectively reducing the collision risk in bidirectional lane-change scenarios. Specifically, a model based on… - An online adaptive algorithm for gradient tracking weights in positive definite cone projection, by introducing - Positive definite constraints set a lower bound for the weight matrix, fundamentally avoiding the problem of weights degenerating into zero matrices, allowing the priority of various performance indicators (absolute convergence, fuel economy, driving comfort, and relative convergence) to be dynamically adjusted as the lane-changing process progresses; compared to the traditional ADMM algorithm, the gradient tracking algorithm has a linear convergence speed and does not require maintaining dual variables, requiring only one round of neighbor communication per iteration, significantly improving control real-time performance; finally, it generates the optimal control sequence for each vehicle node, driving the vehicle to complete the bidirectional lane-changing operation, effectively improving the safety margin of bidirectional lane changing, shortening the lane-changing time, and ensuring the recursive feasibility and closed-loop stability under dynamic topology changes. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein: Figure 1 This is a schematic diagram of a scenario for the multi-queue vehicle bidirectional lane-changing cooperative control method provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the multi-queue vehicle bidirectional lane-changing cooperative control method provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of a bidirectional lane-switching example under the LT communication topology provided in an embodiment of the present invention; Figure 4 This is a speed curve diagram of the bidirectional lane-switching process under the LT communication topology provided in an embodiment of the present invention; Figure 5Torque curve diagram of bidirectional lane-changing process under LT communication topology provided in the embodiment of the present invention; Figure 6 An acceleration curve diagram of the bidirectional lane-switching process under the LT communication topology provided in an embodiment of the present invention; Figure 7 A three-dimensional spatiotemporal trajectory visualization panoramic view of a two-way lane-changing process provided in an embodiment of the present invention; Figure 8 This is a diagram illustrating the adaptive evolution of the weights of following vehicle nodes in the LT topology provided in an embodiment of the present invention. Figure 9 This is a system block diagram of a multi-queue vehicle bidirectional lane-changing cooperative control system provided in an embodiment of the present invention. Detailed Implementation

[0009] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0010] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0011] It should be noted that the terms "first, second, and third" used in the embodiments of the present invention are only used to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first, second, and third" can be interchanged in a specific order or sequence where permitted, so that the embodiments of the present invention described herein can be implemented in an order other than that illustrated or described herein.

[0012] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which these embodiments of the invention pertain. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0013] Figure 1This is a schematic diagram illustrating a scenario of a multi-queue vehicle bidirectional lane-changing cooperative control method provided in an embodiment of the present invention. The schematic diagram shows... There are several queues, each containing i+1 vehicles. The first vehicle in a queue is called the lead vehicle, the last vehicle is called the tail vehicle, and the vehicles between the lead and tail vehicles are called follow vehicles. A follow vehicle is called an "entering vehicle" when it needs to join the target queue, and an "exiting vehicle" when it needs to leave its current queue. In this embodiment, the lead and tail vehicles do not participate in the entry and exit actions. It should be noted that the number of vehicles in each queue can be the same or different, depending on the actual situation; this embodiment does not impose a limitation. Figure 1 As shown, This indicates the distance between the last car in one queue and the lead car in the other queue. This indicates the distance between two adjacent following vehicles in the same queue.

[0014] Figure 2 This is a flowchart illustrating a multi-queue vehicle bidirectional lane-changing cooperative control method provided in an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes at least the following steps: Step S110: Obtain the driving status information and lane change request information of the target vehicle in the multi-queue vehicles in real time.

[0015] Among them, the target vehicle refers to the following vehicle in the queue that has a lane change requirement, including vehicles entering the target queue from adjacent lanes or adjacent queues, and vehicles exiting the current queue to enter adjacent lanes or adjacent queues.

[0016] In this embodiment, the driving status information refers to the target vehicle's basic status information and queue association information. The basic status information includes the vehicle identifier and the vehicle's own driving status information, including but not limited to the vehicle's mass, wheelbase, moment of inertia, position, speed, acceleration, driving torque, heading angle, yaw angle, and yaw rate. The queue association information refers to the queue information and vehicle information associated with the target vehicle, including but not limited to the queue identifier of the queue to which the target vehicle belongs, the target vehicle's position information within the queue (e.g., the 3rd vehicle in queue A), and the driving status information of vehicles associated with the target vehicle's entry and exit points. In this embodiment, vehicles associated with the target vehicle's entry and exit points include, but are not limited to, the lead vehicle in the queue to which the target vehicle belongs, and vehicles adjacent to the target vehicle's entry and / or exit positions.

[0017] Lane change request information refers to the information sent by the target vehicle requesting a lane change, including but not limited to the target vehicle identifier, lane change mode, target queue identifier, and target location. The target vehicle can only enter the lane change execution phase after receiving lane change consent responses from all associated vehicles.

[0018] Furthermore, in this embodiment, the following vehicle needs to track the speed of the lead vehicle and maintain a desired distance from the vehicle in front, specifically: (1.1); In the formula, Indicates time, Indicates speed, Indicates location, Indicates the first The th queue Two following vehicles, among which Indicates the lead car. Indicates following the vehicle. Indicates in Time of the first The first queue The speed of the following vehicle Indicates in Time of the first The speed of the lead car in each queue Indicates in Time of the first The first queue The position of the following vehicle Indicates in Time of the first The first queue The position of the following vehicle This represents the distance between two adjacent following vehicles in the same queue. The two equations above indicate that the speed tracking error and the spacing error of the following vehicles asymptotically converge to zero, respectively.

[0019] In this embodiment, the target vehicle needs to smoothly complete the lateral lane change and merge into the target queue when performing the cut-in operation, and finally the yaw angle returns to zero. The corresponding expression is: (1.2); In the formula, This indicates the moment when the target vehicle completes the cut-in operation. This indicates that the target vehicle was in [the area] when performing the cut-in operation. The horizontal position coordinates at time [time]. This represents the lateral coordinate of the centerline of the target lane. This indicates that the target vehicle was in [the following situation] when performing the cut-in operation. The yaw angle at any given moment. The target lane refers to the lane the target vehicle needs to enter when performing the cut-in operation, i.e., the lane where the target queue is located.

[0020] In this embodiment, the target vehicle needs to smoothly complete the lateral lane change and merge into the traffic flow of the outer lane when performing the lane-cutting operation: (1.3); In the formula, This indicates the moment when the target vehicle completes the cut-out operation. This indicates that the target vehicle was in [the following situation] when performing the cut-out operation. The horizontal position coordinates at time [time]. This indicates the lateral coordinates of the center line of the outer lane. This indicates that the target vehicle was in [the following situation] when performing the cut-out operation. The yaw angle at any given moment. The outer lane refers to the lane the target vehicle needs to enter when performing the cut-out operation.

[0021] Furthermore, in this embodiment, the distance between any two adjacent vehicles must be greater than the minimum safe distance, specifically: (1.4); In the formula, express Time of the first Car and the The distance between vehicles This indicates the minimum safe distance. Among them, .

[0022] Step S120: Determine the lane-changing mode based on the lane-changing request information of the target vehicle, and construct a two-way time-varying communication topology based on the lane-changing mode.

[0023] In this embodiment, the lane change request information includes lane entry request information and lane exit request information. When the lane change request information is a lane entry request information, the corresponding lane change mode is the lane entry mode; when the lane change request information is a lane exit request information, the corresponding lane change mode is the lane exit mode; when both lane entry and lane exit request information appear simultaneously, the corresponding lane change mode is the bidirectional lane entry and lane exit mode.

[0024] Understandably, a target vehicle cannot simultaneously issue both a lane-change entry request and a lane-change exit request. Therefore, the bidirectional lane-change entry / exit mode in this embodiment targets two different target vehicles; that is, while one target vehicle is issuing a lane-change entry request, another target vehicle is also issuing a lane-change exit request. The target vehicles include the entry vehicle and the exit vehicle. The entry vehicle refers to the target vehicle executing the entry mode, and the exit vehicle refers to the target vehicle executing the exit mode.

[0025] In the communication topology, the topology node corresponding to the vehicle cutting in is designated as the incoming vehicle node, and the topology node corresponding to the vehicle cutting out is designated as the outgoing vehicle node. The topology node corresponding to the lead vehicle is designated as the lead vehicle node, and the topology node corresponding to the following vehicle is designated as the following vehicle node. Bidirectional time-varying communication topology refers to the communication topology structure that dynamically evolves over time in a bidirectional lane-changing scenario. Here, "bidirectional" in the bidirectional lane-changing scenario represents both the target vehicle cutting out of the queue and the target vehicle cutting into the queue.

[0026] Specifically, unlike static topology, the topology in a two-way lane-changing scenario evolves dynamically over time. It is necessary to describe the topological features under the following modes: normal mode (i.e., normal driving without cut-in and / or cut-out operations), cut-in mode, cut-out mode, and two-way cut-in / cut-out mode (i.e., cut-in and cut-out operations occur simultaneously). When the lane-changing mode is cut-in mode, a communication link addition operation is performed based on the cutting-in vehicle node to update the communication edge set. When the lane-changing mode is cut-out mode, a communication link deletion operation is performed based on the cutting-out vehicle node to update the communication edge set. When the lane-changing mode is a two-way cut-in / cut-out mode, a mixed communication link operation is performed based on both the cutting-in and cutting-out vehicle nodes to update the communication edge set. This mixed operation includes both communication link addition and deletion operations. Finally, by combining the lane-changing mode, the corresponding vehicle nodes for each lane-changing mode, and the updated communication edge set, a two-way time-varying communication topology under different lane-changing modes is generated.

[0027] Furthermore, the communication link addition operation includes: adding a communication link between the cutting vehicle node and the target gap preceding vehicle node and a communication link between the cutting vehicle node and the target gap following vehicle node in the communication edge set; wherein, the target gap preceding vehicle node is the preceding neighbor vehicle node of the cutting vehicle's planned insertion position, and the target gap following vehicle node is the following neighbor vehicle node of the cutting vehicle's planned insertion position.

[0028] The communication link deletion operation includes: deleting the communication link between the cutting-out vehicle node and the original preceding vehicle node and the communication link between the cutting-out vehicle node and the original following vehicle node in the communication edge set, and establishing a bypass communication link between the original preceding vehicle node and the original following vehicle node; wherein, the original preceding vehicle node is the vehicle node that followed the cutting-out vehicle before the cutting-out vehicle performed the cutting-out operation, and the original following vehicle node is the vehicle node that followed the cutting-out vehicle before the cutting-out vehicle performed the cutting-out operation.

[0029] In this embodiment, the target gap refers to the longitudinal space at the planned insertion position of the cutting vehicle, that is, the driving gap between the front vehicle node and the rear vehicle node of the target gap.

[0030] Understandably, in this embodiment, there is a one-to-one mapping relationship between physical vehicles and vehicle nodes in the communication topology; both refer to the same object. To facilitate the distinction between scenarios, "vehicle" is used as a designation when describing physical technical content such as vehicle dynamics models and safety constraints; while "vehicle node" is used as a designation when describing logical technical content such as communication topology and bidirectional lane-changing distributed model predictive control framework.

[0031] For ease of explanation, this embodiment uses [the following] in the communication topology and bidirectional lane-changing distributed model predictive control framework. This indicates the entry point to the vehicle node. This indicates that the vehicle node has been cut off. Indicates the node of the vehicle preceding the target clearance. Represents the target clearance rear vehicle node; used in vehicle dynamics models and safety constraint sets. Indicates the vehicle to be cut in. This indicates that the vehicle has been switched out. Indicates the target gap ahead vehicle. The target gap is indicated by the following vehicle.

[0032] Furthermore, this embodiment defines a global communication topology. ,in, express The set of vehicle nodes at any given time. express The set of edges at any given time (i.e., the set of communication links).

[0033] (1) Normal mode topology When there are no cut-in / cut-out operations, the topology remains stable and can be represented as the union of the topology within the queue and the topology between queues: (2.1); In the formula, the queue topology Indicates the first Information flow within each queue This represents the information flow between the lead vehicles in each queue.

[0034] In the communication topology, for the first The first queue The set of vehicles whose information can be obtained by a following vehicle is called the neighbor vehicle node set, defined as: .in, Let be the set of neighboring vehicle nodes within the queue, representing the number of nodes that can be routed to the 1st vehicle. Vehicle nodes within the queue that sends information to vehicles following each other; (like )or (like ) indicates the first The first queue Whether the following vehicle communicates directly with the corresponding lead vehicle. Correspondingly, define the first... The set of information sending objects for the following vehicles, i.e., the receiving vehicle. The set of vehicle nodes for following vehicle information is: .in, Indicates the first Vehicle nodes.

[0035] (2) Cut-in mode topological evolution When the vehicle cuts in The moment begins to cut into the first Team position When this occurs, communication links need to be established in the topology between the entering vehicle node and the preceding vehicle node in the target gap, and between the entering vehicle node and the following vehicle node in the target gap. The edge set update rule for the topology is as follows: (2.2); In the formula, express Time of the first The set of edges of each queue (i.e., the set of communication links). Indicates the start time of the vehicle's entry. This indicates the end time of the vehicle's entry.

[0036] The above formula represents the cut-in vehicle node during the cut-in operation of the target vehicle. Receive target gap preceding vehicle node Information, and to the target gap rear vehicle node Send the message. At this point, the set of neighboring vehicle nodes of the incoming vehicle node is: After the cut-in operation is completed ( The communication topology was restored to the normal mode structure, but the number of queue members increased by one.

[0037] (3) Cut-out pattern topological evolution The topological evolution of the cut-out vehicle is the reverse of the cut-in operation. When the first... The team's first The car is When performing a cut-out operation, a bypass connection needs to be established to maintain the connectivity of the information flow and prevent information transmission from being interrupted due to the departure of the cut-out vehicle. The topology edge set update rule is as follows: (2.3); In the formula, express Time of the first The set of edges of each queue (i.e., the set of communication links). This indicates the start time of the vehicle being cut off. This indicates the end time of the vehicle's cut-out.

[0038] The above formula indicates that when the target vehicle performs a cut-out operation, the cut-out vehicle node is deleted respectively. With the original preceding node and cutting out vehicle nodes With the original rear vehicle node The original connections between them are maintained, and the original leading node is re-established. and the original rear vehicle node This involves bypass connections between vehicles. This ensures that the information flow of other vehicles in the queue is not affected while the target vehicle is performing a cut-out operation.

[0039] It is worth noting that when performing a cut-out operation, the vehicle node is cut out. Temporary communication with the original neighboring vehicle nodes must still be maintained for security monitoring until the switched-out vehicle has completely left the original queue: Once the cut-out operation is complete and the cut-out vehicle node is completely removed from the original queue topology ( ( ), cut off the vehicle to become an independent vehicle or merge into the traffic flow of the outer lane.

[0040] (4) Cut-in and cut-out bidirectional topology evolution When both cutting in and cutting out occur simultaneously, topological evolution needs to consider the effects of both operations. The edge set in this case can be represented as: (2.4); In the formula, express The newly added edge (communication link) is executed at any time during the cut-in operation. express The bypass side (bypass communication link) that performs the cut-out operation at all times. express Edges (communication links) that need to be deleted at any time.

[0041] Step S130: Update the original vehicle dynamics model based on the target vehicle's driving status information at the current moment to obtain the target vehicle dynamics model.

[0042] The original vehicle dynamics model refers to the vehicle dynamics model constructed for the target vehicle at the initial moment; the target vehicle dynamics model refers to the vehicle dynamics model obtained by updating the information in the original vehicle dynamics model based on the driving state information of the target vehicle at the current moment.

[0043] Specifically, based on the target vehicle's current driving state information, the current longitudinal state information and the current lateral state information are determined. These are then used as initial state vectors input into the original vehicle dynamics model. A Coriolis coupling term is calculated based on the current lateral velocity and current yaw rate; this term characterizes the influence of lateral motion on longitudinal velocity. Next, the target vehicle's characteristic parameters are obtained, and combined with the Coriolis coupling term, a state-space equation containing longitudinal and lateral coupling is constructed to generate the target vehicle dynamics model. Here, the current longitudinal state information refers to the target vehicle's longitudinal state information at the current moment, including but not limited to position, velocity, acceleration, and driving torque; the current lateral state information refers to the target vehicle's lateral state information at the current moment, including but not limited to lateral position, heading angle, lateral velocity, and yaw rate.

[0044] The vehicle dynamics model in this embodiment includes an in-cutting vehicle dynamics model and an out-cutting vehicle dynamics model.

[0045] Furthermore, in this embodiment, both the entering and exiting vehicles employ an 8-state dynamic model with longitudinal and lateral coupling, including four longitudinal states (longitudinal position, longitudinal velocity, longitudinal acceleration, and driving torque) and four lateral states (lateral position, heading angle, lateral velocity, and yaw rate). They have the same structure, the main difference being: the entering vehicle's longitudinal velocity tracks the cruising speed of the target queue, and its lateral displacement direction points towards the lane where the target queue is located; the exiting vehicle's longitudinal velocity tracks the traffic flow speed of the outer lane, and its lateral displacement direction moves away from the lane where the original queue was located. The comparison between the two is shown in Table 1. Table 1 ; The following example, using a vehicle being cut out, illustrates the dynamics modeling process. The modeling process for a vehicle being cut in is similar; only the subscripts in the state vector, reference velocity, and control input need to be changed. Replace with " The other forms are the same, and will not be repeated here.

[0046] Define the driving status information of the vehicle to be switched out as follows: Among them, vertical state information Includes the position, velocity, acceleration, and driving torque of the cut-out vehicle, and its lateral state. This includes the lateral position, heading angle, lateral velocity, and yaw rate of the cut-out vehicle. This is the matrix transpose symbol.

[0047] The longitudinal dynamic equations of the cut-out vehicle are the same as those of a regular vehicle, and are described by the following model: (3.1); In the formula, This represents the acceleration response time constant of the vehicle being cut off. This represents the time constant of the drive system when the vehicle is switched off. and These represent the desired acceleration and desired driving torque of the cutting-out vehicle, respectively. The key difference from the cutting-in vehicle is that the longitudinal velocity of the cutting-out vehicle is the traffic flow velocity of the outer lane. The longitudinal speed of the merging vehicle is the target queue speed. This means that the merging vehicle may need to accelerate or decelerate during the lane change process to adapt to the traffic conditions in the outer lane.

[0048] Lateral dynamics employs a two-degree-of-freedom vehicle model, which can effectively describe the lateral motion characteristics of a vehicle within the typical highway speed range (approximately 60-120 km / h). (3.2); In the formula, and These represent the lateral stiffness of the front and rear wheels, respectively. and These represent the distances from the front axle and rear axle to the vehicle's center of gravity, respectively. This indicates the mass of the vehicle from which the cut-out vehicle was located. This represents the moment of inertia of the vehicle about its vertical axis. This indicates the input for controlling the front wheel steering angle of the vehicle. This indicates the lateral position of the vehicle being cut out. This indicates the heading angle of the vehicle being cut off. This indicates the lateral velocity of the vehicle cutting out. This indicates the yaw rate of the vehicle being cut off. This indicates the longitudinal speed at which the vehicle is cut out.

[0049] The longitudinal and lateral motions influence each other through the Coriolis effect. When the shearing vehicle has lateral velocity and yaw rate, it will affect the longitudinal velocity. This Coriolis coupling term can be expressed as: ; In the formula, This represents the increase in longitudinal velocity caused by lateral motion. This indicates the lateral velocity of the vehicle cutting out. This indicates the yaw rate of the vehicle being cut off. Indicates the sampling period.

[0050] To facilitate digital controller design, this embodiment uses a sampling period. Discretizing the continuous dynamics model yields: (3.3); In the formula, Indicates the current moment. and These represent the state vectors of the vehicle at the current time and the next time, respectively. This represents the control input vector of the vehicle at the current moment. and These represent the expected acceleration and front wheel steering angle of the vehicle at the current moment, respectively. and These represent the state transition matrix and control input matrix of the discretized vehicle, respectively.

[0051] For the other following vehicles in the queue of the vehicle that was cut off, the output vector is defined as a combination of position and velocity: (3.4); In the formula, Indicates the first The th queue A following vehicle The output vector at time step 1 and They represent the first The th queue A following vehicle The longitudinal position and longitudinal velocity at any given moment.

[0052] The output equation can be expressed as: (3.5); In the formula, Indicates the first The th queue A following vehicle The vertical state vector at time t. Represents the output matrix, used to analyze the vertical state. The position and velocity components are extracted from them.

[0053] Step S140: Based on the bidirectional time-varying communication topology and the target vehicle dynamics model, construct a bidirectional lane-changing distributed model predictive control framework.

[0054] Specifically, the process iterates through each topology node in the bidirectional time-varying communication topology, sequentially identifying each topology node as the current vehicle node and obtaining the driving status information of the vehicle corresponding to the current vehicle node. Based on the bidirectional time-varying communication topology, all neighboring vehicle nodes of the current vehicle node are taken as target vehicle nodes, and the driving status information of the vehicles corresponding to the target vehicle nodes is obtained. The driving status information of the vehicles corresponding to the current vehicle node and the driving status information of the vehicles corresponding to the target vehicle nodes are input into the target vehicle dynamics model to obtain the prediction model of the current vehicle node. Finally, the optimization cost function of the current vehicle node is constructed based on the prediction model, and the safety constraint set of the current vehicle node is established. Combining the optimization cost functions and safety constraint sets of all current vehicle nodes, a bidirectional lane-changing DMPC is constructed.

[0055] Furthermore, this embodiment defines a lane-changing mode. The specific meanings of each lane-changing mode are as follows: : In normal mode, there is no cut-in / cut-out operation, and the queue maintains a stable formation while moving. : Cut-in mode, only the cut-in operation is performed; : Cutout mode, only the cutout operation is performed; : Cut-in and cut-out bidirectional mode, where the cut-in and cut-out operations are performed simultaneously.

[0056] To quantify the completion rate of vehicle entry, for the vehicle entry node Define the cut-in progress variable: (4.1); In the formula, Indicates the start-up progress variable. Indicates the node where the vehicle is connected. The horizontal position coordinates at time [time]. This indicates the starting time of the vehicle connection node. The horizontal position coordinates, This indicates the lateral coordinates of the centerline of the target lane.

[0057] In this embodiment, the range of values ​​for the cut-in progress variable is: ,when This indicates that the target vehicle has not yet begun the cut-in operation. The time indicates that the target vehicle has completed the cut-in operation. The progress variable can be used to smoothly switch control strategies and adjust weights.

[0058] Similarly, for the cut-out vehicle node Define the cut-out progress variable: (4.2); In the formula, This represents the cut-out progress variable. Indicates the vehicle node being cut out. The horizontal position coordinates at time [time]. This indicates the moment when the vehicle node is switched out. The horizontal position coordinates, This indicates the lateral coordinates of the center line of the outer lane.

[0059] when When the time is right, it indicates that the target vehicle has completed the cut-out operation and has merged into the outer lane.

[0060] Furthermore, the lane-changing mode is determined by the current state of the set of vehicle nodes entering / leaving the lane, and the switching logic is as follows: (4.3); The lane-changing mode's switching action is event-triggered; it is triggered when a new lane-in / lane-out request occurs or an existing operation completes. In different modes, the set of terminal constraints for the optimization problem changes accordingly to adapt to different control objectives. (4.4); In the formula, Indicates lane change mode The terminal constraint set below, , , , These represent the terminal constraint sets in normal mode, cut-in mode, cut-out mode, and cut-in / cut-out bidirectional mode, respectively.

[0061] Furthermore, vehicle safety is the primary objective of platooning coordinated control. Safety constraints in two-way lane-changing scenarios are more complex than those in one-way lane-changing scenarios, requiring consideration of multiple potential collision risks.

[0062] Specifically, the set of security constraints set in this embodiment includes: 1) Set of safety constraints within the queue Sufficient safe distance must be maintained between adjacent vehicles in the queue to cope with situations such as emergency braking: (4.5); In the formula, This indicates the minimum safe distance between adjacent vehicles in the queue.

[0063] 2) Enter the safety constraint set Distance constraints between the vehicle cutting in and the vehicle ahead of the target, and between the vehicle behind the target and the vehicle cutting in, are crucial for ensuring safe cutting in. (Definition) Cut into the gap between the vehicle and the target vehicle The distance constraint between them is: (4.6); After cutting into the gap between the vehicle and the target, the vehicle... The distance constraint between them is: (4.7); In the formula, and These represent the vehicle lengths of the vehicle ahead and the vehicle cutting in at the target gap, respectively. and These represent the minimum safe distances between the vehicle cutting in and the vehicle in front of the target gap, and the vehicle behind the target gap, respectively.

[0064] 3) Cut out the safety constraint set Vehicles cutting out must maintain a safe distance from vehicles already in the outer lane. Definition The distance constraint between the vehicle cutting out and the nearest preceding vehicle in the outer lane is: (4.8); The distance constraint between the vehicle cutting out and the nearest following vehicle in the outer lane is: (4.9); In the formula, and These represent the vehicles in the outer lane that are closest to the insertion point of the cutting-out vehicle, respectively. The relevant distance constraints are used to ensure that the cutting-out vehicle safely integrates into the traffic flow of the outer lane.

[0065] 4) Cross-interference safety constraint set When both a cutting-in vehicle and a cutting-out vehicle appear simultaneously, their cutting-in trajectories may intersect in space and time (e.g., their cutting-in and cutting-out positions are adjacent), requiring additional cross-trajectory safety constraints. (4.10); In the formula, and These represent the vehicles entering and exiting the switch, respectively. The vertical position of time. and These represent the vehicles entering and exiting the switch, respectively. The horizontal position of the moment. This indicates the minimum safe distance for cross-interference. This represents the two-dimensional Euclidean norm.

[0066] The constraint is in the form of two-dimensional Euclidean distance, taking into account both longitudinal and lateral positional differences.

[0067] 5) Unified set of security constraints All the above security constraints can be uniformly represented as a set of security constraint conditions: (4.11); In the formula, express A unified set of security constraints at any given time. Represents the set of safety constraints within the queue. , , They represent The set of security constraints for each moment, including the set of security constraints for each moment, the set of security constraints for each moment, and the set of security constraints for each moment of cross-interference.

[0068] Furthermore, the bidirectional lane-changing DMPC established in this embodiment is as follows: (4.12); In the formula, express The total cost function of all vehicle nodes at time 1. Indicates the total number of queues. Indicates the first The number of vehicles following in each queue and They represent the first The optimized cost function for the lead vehicle node and follower vehicle nodes of each queue. and Let represent the optimization cost functions for the inbound and outbound vehicle nodes, respectively. and They represent The set of entering vehicle nodes and the set of exiting vehicle nodes at any given time.

[0069] The first term in the above formula is the optimized cost function for each queue's lead vehicle node and follow vehicle node, the second term is the optimized cost function for all incoming vehicle nodes, and the third term is the optimized cost function for all outgoing vehicle nodes.

[0070] Step S150: The weight parameters in the bidirectional lane-changing distributed model predictive control framework are updated using a gradient tracking-based online adaptive weight algorithm to obtain the optimal control sequence for each vehicle node.

[0071] In this embodiment - Positive definite constraint conditions are: ; In the formula, This indicates a pre-defined positive definite constant. , , , They represent the first The first queue The absolute convergence weight matrix, fuel economy weight matrix, driving comfort weight matrix, and relative convergence weight matrix of each following vehicle node.

[0072] By setting - Positive definite constraint condition: Sets a positive definite constant for the weight matrix. The lower bound is determined to prevent the weight matrix from degenerating into a zero matrix during the optimization process, which would cause the corresponding performance indicators to be completely ignored.

[0073] Specifically, in - Initialize the weight matrix of each current vehicle node in the bidirectional lane-changing distributed model predictive control framework within the positive definite cone to obtain the current weight matrix.

[0074] By combining the current weight matrix and the optimization cost function, safety constraint set, and target vehicle dynamics model in the bidirectional lane-changing distributed model predictive control framework, a temporary optimal control sequence is generated. Based on the temporary optimal control sequence, the local gradient of the optimization cost function with respect to the current weight matrix is ​​calculated, and the gradient estimate iterative update step is performed based on the local gradient.

[0075] Furthermore, this embodiment also sets a stability constraint, which is specifically as follows: ; Constraints require comfort weight Not less than the relative convergence weight of all sending objects The sum of these values ​​is a sufficient condition to guarantee the stability of the bidirectional lane-changing distributed model predictive control framework. In addition, special cases need to be considered: ; When following the car When not communicating directly with the lead vehicle ( It cannot obtain the expected output. Therefore, the weights of absolute convergence It should be set to zero.

[0076] ; When following the car When there are no neighboring vehicles in the same team, the relative convergence weights are... Set it to zero.

[0077] To facilitate the design of the bidirectional lane-changing distributed model predictive control framework, this embodiment introduces auxiliary variables. Establish consistency constraints: ; ; The constraints are decomposed into two parts: consistency equality constraints and auxiliary variables. - Positive definite constraints facilitate subsequent execution of gradient estimation iteration update steps based on local gradients.

[0078] The gradient estimate iterative update step specifically includes the following steps: Based on the bidirectional time-varying communication topology in the bidirectional lane-changing distributed model predictive control framework, a dual-randomness mixing matrix is ​​generated. Based on this matrix, a weighted average is calculated for the weight matrix estimates of each current vehicle node and its corresponding target vehicle node to obtain a fusion matrix. The fusion matrix is ​​then updated using gradient descent along the negative direction of the current local gradient, and the updated fusion matrix is ​​further refined. - Positive definite cone projection correction yields an estimated value of the correction matrix.

[0079] Furthermore, - Positive definite cone projection correction specifically includes: performing... - Orthogonal eigenvalue decomposition yields an orthogonal eigenvector matrix and an eigenvalue diagonal matrix; all eigenvalues ​​in the eigenvalue diagonal matrix less than a preset positive definite constant are then decomposed. The eigenvalues ​​are uniformly truncated to the preset positive definite constant. The corrected eigenvalue diagonal matrix is ​​obtained; based on the orthogonal eigenvector matrix and the corrected eigenvalue diagonal matrix, the estimated value of the correction matrix is ​​reconstructed.

[0080] After obtaining the correction matrix estimate, the local gradient of the optimization cost function on the correction matrix estimate is recalculated based on the correction matrix estimate to obtain the corrected local gradient. The current gradient estimate of the current vehicle node and its corresponding target vehicle node is calculated by weighted average through a double randomness mixing matrix, and the gradient estimate of the current vehicle node is updated by combining the difference between the current local gradient and the corrected local gradient.

[0081] The updated gradient estimate is: ; In the formula, Indicates the first Vehicle node in the next iteration gradient estimator, This represents the mixing weights in the double randomness mixing matrix. Represents vehicle node The neighboring vehicle nodes (including itself). Represents vehicle node The extended set of neighbors (including itself). and They represent the first Second and third In the next iteration, the local gradient of the cost function with respect to the weight matrix is ​​optimized. Here, the first term is the weighted average of the gradient estimates of the neighboring vehicle nodes, and the last two terms are the gradients of the vehicle nodes. The change in the gradient estimator is such that the gradient estimator can track the global average gradient.

[0082] After obtaining the updated gradient estimate, the calibration matrix estimate and the updated gradient estimate are sent to the target vehicle node corresponding to the current vehicle node via vehicle-to-vehicle communication. At the same time, the corresponding updated data sent by the target vehicle node are received as input parameters for the next gradient estimate iteration update step.

[0083] The gradient estimate is updated in real time. When the gradient estimate is updated to the preset number of iterations, the current gradient estimate is obtained. The current weight matrix is ​​adaptively adjusted based on the current gradient estimate to obtain an adaptive weight matrix. Based on the adaptive weight matrix, and combined with the optimization cost function, safety constraint set and target vehicle dynamics model in the bidirectional lane-changing distributed model predictive control framework, the optimal control sequence of each current vehicle node in the prediction time domain is obtained.

[0084] Step S160: Control the corresponding vehicle to perform longitudinal and lateral coordinated movement according to the optimal control sequence of each vehicle node, and complete the bidirectional lane-changing operation of multiple queues of vehicles.

[0085] The above-described multi-queue vehicle bidirectional lane-changing cooperative control method will be described below with reference to a specific embodiment. However, it should be noted that this specific embodiment is only for better illustrating the present invention and does not constitute an improper limitation of the present invention.

[0086] like Figure 3 As shown, there are 3 queues with a total of 22 vehicles. Queue 1 includes vehicles 1-8, queue 2 includes vehicles 9-15, and queue 3 includes vehicles 16-22. Among them, 3 vehicles cut in (from right to left: vehicle 1, vehicle 2, and vehicle 3) and 3 vehicles cut out (from right to left: vehicle 6, vehicle 12, and vehicle 19) complete the cut-in / cut-out operation within a 25-second time window.

[0087] The initial cruising speed of all platoons is 20 m / s, the expected spacing between vehicles within a platoon is 10 m, the expected spacing between platoons is 30 m, and the sampling period is 0.1 s. Basic vehicle information: mass 1500 kg, front and rear wheelbases 1.2 m and 1.6 m respectively, moment of inertia... The lateral stiffness of the front and rear wheels is 50,000 N / rad and 60,000 N / rad, respectively.

[0088] The vehicles in the three queues are designated as Vehicle 1 to Vehicle 22 from right to left. The bidirectional cut-in / cut-out configuration reflects the complexity of spatiotemporal coupling. Specifically, Vehicle 1 cuts in at 5 seconds from the outer lane between the target gap vehicle 4 and the target gap vehicle 5 in queue 1; Vehicle 2 cuts in at 9 seconds between the target gap vehicle 10 and the target gap vehicle 11 in queue 2; and Vehicle 3 cuts in at 13 seconds between the target gap vehicle 17 and the target gap vehicle 18 in queue 3. Each cut-in operation lasts 4 seconds. Simultaneously, Vehicle 6 begins to cut out of queue 1 at 8 seconds, Vehicle 12 cuts out of queue 2 at 12 seconds, and Vehicle 19 cuts out of queue 3 at 16 seconds, with each cut-out lasting 4 seconds. This timing design results in spatiotemporal overlap between entry and exit points during the periods of 8-9 seconds, 12-13 seconds, and 16-17 seconds. The relative positions of entering vehicle 1 and entering vehicle 2 are closest around the 9-second mark, representing a key test point for verifying the effectiveness of the cross-interference constraint. The initial speeds of the entering vehicles are set to 21 m / s, 20 m / s, and 19 m / s, slightly higher than or close to the queue speed to facilitate speed matching. Exiting vehicles, on the other hand, need to adjust to a target speed of 22 m / s in the outer lane during the exit process.

[0089] The DMPC controller parameters were set considering tracking performance, computational efficiency, and actuator constraints. Longitudinal control employs a 20-step predictive time domain, with the initial weight matrix as follows: [lead vehicle node...] , , , Follow vehicle nodes , , , The driving torque is constrained to ±3000 N·m. Lateral control also employs a 20-step prediction time domain, with front wheel steering angle constrained to ±0.25 rad and steering angle change rate constrained to ±0.1 rad / s. The parameters of the online adaptive weighting algorithm are set as follows: preset positive definite constants. =0.01, gradient tracking step size 0.05, preset iteration count 15, and the hybrid matrix adopts Metropolis-Hastings design. This embodiment uses LT (Leader-Tail) topology simulation as an example for illustration.

[0090] like Figures 4-6 The figure shows the overall longitudinal dynamic response of three queues under the bidirectional lane-switching cooperative control method in the LT communication topology. From... Figure 4The speed curves clearly show that the three queues maintain a stable cruising speed of 20 m / s before the lane change, with the phase difference between each queue reflecting a spacing of approximately 30 m. When the merging vehicle node 1 begins to merge at 5 seconds, queue 1 experiences significant speed fluctuations, with the speed of the leading vehicle nodes (vehicle nodes 1-4) decreasing by approximately 4 m / s to around 16 m / s. This deceleration is necessary to make room for the merging vehicle node. Notably, this speed disturbance is largely confined within queue 1. Queues 2 and 3 remain relatively stable during this phase, exhibiting only minor adjustment fluctuations (approximately 0.5 m / s), validating the effectiveness of the multi-queue topology in suppressing disturbance propagation. The speed trajectories of the three merging vehicle nodes show that they approach the target position at a speed slightly higher than the queue's speed before merging, and quickly adjust to synchronize with the target queue at the start of the merge, with a smooth and abrupt speed adjustment process.

[0091] The complexity of bidirectional operations is particularly pronounced during the 8-9 second overlap phase. At this time, vehicle node 6 is cutting out while vehicle node 2 is about to cut in, and queues 1 and 2 are simultaneously undergoing topology reconstruction. Figure 5 As can be seen from the torque curve, the torque at node 7 of the original rear vehicle experiences a significant adjustment between 8 and 12 seconds, reaching a peak value of approximately This is because after vehicle node 6 is cut off, the original following vehicle node 7 needs to undergo topology reconstruction, switching from vehicle node 7 to vehicle node 6, during which the target vehicle is switched and the spacing is readjusted. Vehicles in queue 2 also experienced similar torque fluctuations after vehicle 2 was cut off. The torque changes of the three cut-off vehicles were relatively smooth, with a peak of approximately... The speed is significantly smaller than that of vehicles in the queue, ensuring that vehicles cutting out can gradually transition from queue speed to outer lane speed, avoiding drastic acceleration and deceleration.

[0092] Figure 6 The acceleration curves further validated the smoothness of the control strategy. The peak acceleration of vehicles in the platoon was generally controlled within ±5 m / s², meeting the comfort requirements of passenger cars. Between 5 and 17 seconds, the acceleration fluctuations in platoon 1 were the most dramatic, especially at the target gap front vehicle node 4 and target gap rear vehicle node 5, requiring frequent adjustments to maintain safe distances and coordinated movement. Entering vehicle nodes experienced a brief acceleration process at the initial stage of lane changes to match the target platoon speed; the peak acceleration of the three entering vehicle nodes was approximately 3-4 m / s², lasting about 1-2 seconds. The acceleration changes of exiting vehicle nodes were more gradual, with peak values ​​not exceeding 2 m / s², because the difference between the target speed (22 m / s) in the outer lane and the platoon speed (20 m / s) was small, requiring only a 2 m / s speed increment. After 20 seconds, the speeds of all vehicles gradually converged to steady-state values, the disturbances largely dissipated, and the system entered a new stable configuration.

[0093] like Figure 7 The image shows a panoramic view of the three-dimensional spatiotemporal trajectory visualization of the two-way lane-changing process. From the top-down projection, it is clear that the three merging vehicle nodes smoothly transition from the outer lane (Y=0) to the inner lane (Y=3.5m), their lateral trajectories exhibiting a graceful S-shaped curve. The tangents at the start and end points are nearly horizontal, with the greatest curvature in the middle section. Correspondingly, the three terminating vehicle nodes complete the lane change in the opposite direction, moving from the inner lane to the outer lane, their trajectory shapes mirroring those of the merging vehicle nodes. Of particular note is the spatiotemporal proximity of merging vehicle node 2 and terminating vehicle node 1. From the 3D perspective, the longitudinal position difference between the two vehicle nodes during the 9-10 second period is approximately 50m, and the lateral position difference is approximately half the lane width (1.75m). This is the moment with the highest risk of cross-interference during the entire simulation. Despite this, the cross-interference constraints successfully avoid the risk of collision, as can be seen from the image, the trajectories of the two vehicles maintain a clear spatial separation throughout.

[0094] like Figure 8 The diagram shows the adaptive evolution of the weights of the following vehicle nodes in the LT topology. (a) is the Q-weight graph of the following vehicle nodes, (b) is the R-weight graph of the following vehicle nodes, (c) is the F-weight graph of the following vehicle nodes, and (d) is the G-weight graph of the following vehicle nodes. Figure 8 It can be seen that the Q-weight of the following vehicle node is used to control the state error and remains close to zero throughout the simulation. This is because the following vehicle node in the algorithm pays little attention to the absolute state error. The control input weight R remains at 1 throughout the simulation, indicating that the basic requirements for the control input remain stable. The weight F is used to control the assumed output error. The F value is periodically adjusted between 9 and 10.5. The adjustment sequence of different queues corresponds to their respective lane-changing events: the following vehicle node in queue 1 adjusts during 5-9 seconds, the following vehicle node in queue 2 responds during 9-13 seconds, and the following vehicle node in queue 3 adjusts during 13-17 seconds. When a lane change is detected in the queue, the system increases the F weight to pay more attention to the deviation between the assumed output and the actual output, ensuring prediction accuracy and control smoothness. The weight G is used to control the state error of the previous vehicle node, and its change is the most significant, fluctuating between 4.5 and 5.5. When the cutting vehicle node starts to cut in, the G value of the following vehicle node in the corresponding queue rises rapidly to a peak of about 5.5, which strengthens the attention to the state error of the previous vehicle node to adapt to topology reconstruction. After the lane change is completed, the G value gradually falls back to a new steady-state level of about 4.5-5.0, which reflects the system's dynamic adjustment capability to the coordination needs between nodes.

[0095] Figure 9 This is a system block diagram of a multi-queue vehicle bidirectional lane-changing cooperative control system provided in an embodiment of the present invention. Figure 9As shown, the system includes an information acquisition module, a topology construction module, a model update module, a framework establishment module, an optimal control sequence acquisition module, and a lane-changing operation control module, wherein: The information acquisition module is used to acquire in real time the driving status information and lane change request information of the target vehicle in multiple queues of vehicles.

[0096] The topology construction module is used to determine the lane-changing mode based on the lane-changing request information of the target vehicle, and to construct a two-way time-varying communication topology based on the lane-changing mode.

[0097] The model update module is used to update the original vehicle dynamics model based on the target vehicle's current driving state information, thereby obtaining the target vehicle dynamics model.

[0098] The framework building module is used to construct a bidirectional lane-changing distributed model predictive control framework based on bidirectional time-varying communication topology and target vehicle dynamics model.

[0099] The optimal control sequence acquisition module is used to update the weight parameters in the bidirectional lane-changing distributed model predictive control framework using a gradient tracking-based online adaptive weight algorithm to obtain the optimal control sequence for each vehicle node.

[0100] The lane-changing operation control module is used to control the corresponding vehicles to perform longitudinal and lateral coordinated movements according to the optimal control sequence of each vehicle node, so as to complete the bidirectional lane-changing operation of multiple queues of vehicles.

[0101] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of the invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of the invention, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the invention. The sequence numbers of the above-described embodiments of the invention are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0102] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0103] In the several embodiments provided by this invention, it should be understood that the disclosed methods can be implemented in other ways. The methods disclosed in the several method embodiments provided by this invention can be arbitrarily combined without conflict to obtain new method embodiments. The features disclosed in the several method embodiments provided by this invention can be arbitrarily combined without conflict to obtain new method embodiments.

[0104] The above description is merely an embodiment 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 coordinated control of bidirectional lane changing for multi-queue vehicles, characterized in that, include: Real-time acquisition of driving status information and lane change request information of target vehicles in multiple queues; The lane-changing mode is determined based on the lane-changing request information of the target vehicle, and a two-way time-varying communication topology is constructed based on the lane-changing mode. The original vehicle dynamics model is updated based on the target vehicle's current driving status information to obtain the target vehicle dynamics model; Based on the bidirectional time-varying communication topology and the target vehicle dynamics model, a bidirectional lane-changing distributed model predictive control framework is constructed. The weight parameters in the bidirectional lane-changing distributed model predictive control framework are updated using a gradient tracking-based online adaptive weight algorithm to obtain the optimal control sequence for each vehicle node. Based on the optimal control sequence of each vehicle node, the corresponding vehicle is controlled to perform longitudinal and lateral coordinated movement to complete the bidirectional lane-changing operation of multiple queues of vehicles. The step of constructing a bidirectional lane-changing distributed model predictive control framework based on the bidirectional time-varying communication topology and the target vehicle dynamics model includes: Traverse each topology node in the bidirectional time-varying communication topology, determine each topology node as the current vehicle node in turn, and obtain the driving status information of the vehicle corresponding to the current vehicle node. Based on the bidirectional time-varying communication topology, all neighboring vehicle nodes of the current vehicle node are taken as the target vehicle node, and the driving status information of the vehicle corresponding to the target vehicle node is obtained. The driving status information of the vehicle corresponding to the current vehicle node and the driving status information of the vehicle corresponding to the target vehicle node are input into the target vehicle dynamics model to obtain the prediction model of the current vehicle node. Based on the prediction model, an optimized cost function for the current vehicle node is constructed, and a set of safety constraints for the current vehicle node is established. By combining the optimization cost function and safety constraint set of all current vehicle nodes, a bidirectional lane-changing distributed model predictive control framework is constructed. The set of security constraints includes an in-queue security constraint set, an entry security constraint set, an exit security constraint set, and a cross-interference security constraint set; wherein... The set of safety constraints within the queue is: （4.5） In the formula, Indicates the first The th queue A following vehicle The vertical position of time. Indicates the first The th queue A following vehicle The vertical position of time. Indicates the minimum safe distance between adjacent vehicles in the queue; Indicates the first The total number of following vehicles in each queue; The set of safety constraints includes: Cut into the gap between the vehicle and the target vehicle The distance constraint between them is: （4.6） In the formula, Indicates the distance between the vehicle cutting in and the target vehicle. exist The distance of time, Indicates the target clearance for the preceding vehicle. exist The vertical position of time. Indicates the vehicle cutting in The vertical position of time. This indicates the length of the vehicle ahead of the target clearance. This indicates the minimum safe distance between the vehicle cutting in and the vehicle in front of the target; After cutting into the gap between the vehicle and the target, the vehicle... The distance constraint between them is: （4.7） In the formula, Indicates the gap between the vehicle cutting in and the target vehicle. exist The distance of time, Indicates the vehicle cutting in The vertical position of time. Indicates the target clearance after the vehicle exist The vertical position of time. This indicates the length of the vehicle body that is being cut into. This indicates the minimum safe distance between the vehicle cutting in and the target vehicle; The cut-out safety constraint set includes: The distance constraint between the vehicle cutting out and the nearest preceding vehicle in the outer lane to the vehicle cutting out is: （4.8） In the formula, This indicates that the vehicle cutting out is closest to the vehicle in the outer lane that is in the lane where the vehicle cutting out is positioned. The distance of time, This indicates the vehicle in the outer lane that is closest to the position where the vehicle cutting out is inserted. The vertical position of time. Indicates that the vehicle is switched out. The vertical position of time. This indicates the length of the vehicle in the outer lane that is closest to the position where the cutting-out vehicle will enter. This indicates the minimum safe distance between the vehicle cutting out and the nearest vehicle in the outer lane to the vehicle that cut out from its insertion position; The distance constraint between the vehicle cutting out and the nearest following vehicle in the outer lane is: （4.9） In the formula, This indicates that the vehicle cutting out is closest to the vehicle behind it in the outer lane at the point where the vehicle cutting out entered. The distance of time, Indicates that the vehicle is switched out. The vertical position of time. This indicates the vehicle closest to the cutting-out vehicle's insertion position in the outer lane. The vertical position of time. This indicates the length of the vehicle body that has been cut out. This indicates the minimum safe distance between the vehicle cutting out and the nearest following vehicle in the outer lane; The set of cross-interference safety constraints, when both an inbound vehicle and an outbound vehicle are present simultaneously, has the following corresponding cross-interference safety constraints: （4.10） In the formula, Indicates the vehicle cutting in The vertical position of time. Indicates that the vehicle is switched out. The vertical position of time. Indicates the vehicle cutting in The horizontal position of the moment. Indicates that the vehicle is switched out. The horizontal position of the moment. This indicates the minimum safe distance for cross-interference. Represents the two-dimensional Euclidean norm; The set of safety constraints is as follows: （4.11） In the formula, express A unified set of security constraints at any given time. Represents the set of safety constraints within the queue. express The set of security constraints for each moment of entry. express The moment to cut out the safety constraint set, express Constantly interfering safety constraint set; Two-way lane changing DMPC is: （4.12） In the formula, express The total cost function of all vehicle nodes at time 1. Indicates the total number of queues. Indicates the first The number of vehicles following in each queue Indicates the first The optimized cost function for the navigator node of each queue. Indicates the first The first queue The optimized cost function for each vehicle following a node. This represents the optimization cost function for the vehicle node. This represents the optimized cost function for cutting out vehicle nodes. express The set of vehicle nodes that are accessed at any given moment. express The set of vehicle nodes that are cut off at any given moment; The step involves using a gradient-tracking-based online adaptive weight algorithm to update the weight parameters in the bidirectional lane-changing distributed model predictive control framework, thereby obtaining the optimal control sequence for each vehicle node. This includes: exist - Initialize the weight matrix of each current vehicle node in the bidirectional lane-changing distributed model predictive control framework within the positive definite cone to obtain the current weight matrix; By combining the current weight matrix with the optimization cost function, safety constraint set, and target vehicle dynamics model in the bidirectional lane-changing distributed model predictive control framework, a temporary optimal control sequence is generated. Based on the temporary optimal control sequence, calculate the local gradient of the optimization cost function with respect to the current weight matrix, and perform an iterative update step of the gradient estimate based on the local gradient; The gradient estimate is updated in real time. When the gradient estimate is updated in real time, the current gradient estimate is obtained. Based on the current gradient estimate, the current weight matrix is ​​adaptively adjusted to obtain an adaptive weight matrix; Based on the adaptive weight matrix, and combined with the optimization cost function, safety constraint set, and target vehicle dynamics model in the bidirectional lane-changing distributed model predictive control framework, the optimal control sequence of each current vehicle node in the prediction time domain is obtained. The step of performing gradient estimation iterative update based on the local gradient includes: Based on the bidirectional time-varying communication topology in the bidirectional lane-changing distributed model predictive control framework, a dual-randomness hybrid matrix is ​​generated. Based on the aforementioned double randomness mixing matrix, a weighted average is calculated on the weight matrix estimates of each current vehicle node and its corresponding target vehicle node to obtain the fusion matrix. The fusion matrix is ​​updated using gradient descent along the negative direction of the current local gradient, and the updated fusion matrix is ​​then... - Positive definite cone projection correction yields an estimated value of the correction matrix; Based on the estimated value of the correction matrix, the local gradient of the optimization cost function with respect to the estimated value of the correction matrix is ​​recalculated to obtain the corrected local gradient; The gradient estimates of the current vehicle node and its corresponding target vehicle node are calculated by weighted average using a double randomness mixing matrix, and the gradient estimate of the current vehicle node is updated by combining the difference between the current local gradient and the corrected local gradient. The updated fusion matrix is ​​then processed. - Positive definite cone projection correction yields an estimated correction matrix, including: Perform the updated fusion matrix - Orthogonal eigenvalue decomposition yields an orthogonal eigenvector matrix and an eigenvalue diagonal matrix; All values ​​in the eigenvalue diagonal matrix that are less than a preset positive definite constant are considered. The eigenvalues ​​are uniformly truncated to the preset positive definite constant. The corrected eigenvalue diagonal matrix is ​​obtained. Based on the orthogonal eigenvector matrix and the corrected eigenvalue diagonal matrix, the estimated value of the correction matrix is ​​reconstructed. -Corresponding to the positive definite cone - Positive definite constraint conditions are: ； In the formula, This indicates a pre-defined positive definite constant. , , , They represent the first The first queue The absolute convergence weight matrix, fuel economy weight matrix, driving comfort weight matrix, and relative convergence weight matrix of each following vehicle node.

2. The multi-queue vehicle bidirectional lane-changing cooperative control method according to claim 1, characterized in that, The target vehicle includes an incoming vehicle and an outgoing vehicle; the topology node corresponding to the incoming vehicle is designated as the incoming vehicle node, and the topology node corresponding to the outgoing vehicle is designated as the outgoing vehicle node. The construction of a bidirectional time-varying communication topology based on the lane-switching mode includes: When the lane-changing mode is the cut-in mode, a communication link addition operation is performed based on the cut-in vehicle node to update the communication edge set; When the lane-changing mode is the cut-out mode, a communication link deletion operation is performed based on the cut-out vehicle node to update the communication edge set; When the lane-changing mode is a two-way cut-in / cut-out mode, a communication link hybrid operation is performed based on the cutting-in vehicle node and the cutting-out vehicle node to update the communication edge set; the communication link hybrid operation includes a communication link addition operation and a communication link deletion operation. By combining the lane-changing modes, the vehicle nodes corresponding to different lane-changing modes, and the updated set of communication edges, a bidirectional time-varying communication topology is generated under different lane-changing modes.

3. The multi-queue vehicle bidirectional lane-changing cooperative control method according to claim 2, characterized in that, The communication link addition operation includes: adding a communication link between the cutting vehicle node and the target gap preceding vehicle node and a communication link between the cutting vehicle node and the target gap following vehicle node in the communication edge set; wherein, the target gap preceding vehicle node is the preceding neighbor vehicle node of the planned insertion position of the cutting vehicle, and the target gap following vehicle node is the following neighbor vehicle node of the planned insertion position of the cutting vehicle. The communication link deletion operation includes: deleting the communication link between the cutting-out vehicle node and the original preceding vehicle node and the communication link between the cutting-out vehicle node and the original following vehicle node in the communication edge set, and simultaneously establishing a bypass communication link between the original preceding vehicle node and the original following vehicle node; wherein, the original preceding vehicle node is the vehicle node that followed the cutting-out vehicle before the cutting-out operation was performed, and the original following vehicle node is the vehicle node that followed the cutting-out vehicle before the cutting-out operation was performed.

4. The multi-queue vehicle bidirectional lane-changing cooperative control method according to claim 1, characterized in that, The step of updating the original vehicle dynamics model based on the target vehicle's current driving state information to obtain the target vehicle dynamics model includes: Based on the target vehicle's current driving status information, determine the current longitudinal status information and the current lateral status information; The current longitudinal state information and the current lateral state information are used as initial state vectors and input into the original vehicle dynamics model. The Coriolis coupling term is calculated based on the current lateral velocity and the current yaw rate. The characteristic parameters of the target vehicle are obtained, and a state-space equation containing longitudinal and lateral coupling is constructed by combining the Coriolis coupling term to generate a dynamic model of the target vehicle.

5. A multi-queue vehicle bidirectional lane-changing cooperative control system, characterized in that, The system includes an information acquisition module, a topology construction module, a model update module, a framework establishment module, an optimal control sequence acquisition module, and a lane-changing operation control module, wherein: The information acquisition module is used to acquire in real time the driving status information and lane change request information of the target vehicle in multiple queues of vehicles; The topology construction module is used to determine the lane-changing mode based on the lane-changing request information of the target vehicle, and to construct a two-way time-varying communication topology based on the lane-changing mode. The model update module is used to update the original vehicle dynamics model based on the driving state information of the target vehicle at the current moment, and obtain the target vehicle dynamics model; The framework establishment module is used to construct a bidirectional lane-changing distributed model predictive control framework based on the bidirectional time-varying communication topology and the target vehicle dynamics model. The optimal control sequence acquisition module is used to update the weight parameters in the bidirectional lane-changing distributed model predictive control framework using a gradient tracking-based online adaptive weight algorithm to obtain the optimal control sequence for each vehicle node. The lane-changing operation control module is used to control the corresponding vehicles to perform longitudinal and lateral coordinated movements according to the optimal control sequence of each vehicle node, so as to complete the bidirectional lane-changing operation of multiple queues of vehicles. The step of constructing a bidirectional lane-changing distributed model predictive control framework based on the bidirectional time-varying communication topology and the target vehicle dynamics model includes: Traverse each topology node in the bidirectional time-varying communication topology, determine each topology node as the current vehicle node in turn, and obtain the driving status information of the vehicle corresponding to the current vehicle node. Based on the bidirectional time-varying communication topology, all neighboring vehicle nodes of the current vehicle node are taken as the target vehicle node, and the driving status information of the vehicle corresponding to the target vehicle node is obtained. The driving status information of the vehicle corresponding to the current vehicle node and the driving status information of the vehicle corresponding to the target vehicle node are input into the target vehicle dynamics model to obtain the prediction model of the current vehicle node. Based on the prediction model, an optimized cost function for the current vehicle node is constructed, and a set of safety constraints for the current vehicle node is established. By combining the optimization cost function and safety constraint set of all current vehicle nodes, a bidirectional lane-changing distributed model predictive control framework is constructed. The set of security constraints includes an in-queue security constraint set, an entry security constraint set, an exit security constraint set, and a cross-interference security constraint set; wherein... The set of safety constraints within the queue is: （4.5） In the formula, Indicates the first The th queue A following vehicle The vertical position of time. Indicates the first The th queue A following vehicle The vertical position of time. This indicates the minimum safe distance between adjacent vehicles within the queue; Indicates the first The total number of following vehicles in each queue; The set of safety constraints includes: Cut into the gap between the vehicle and the target vehicle The distance constraint between them is: （4.6） In the formula, Indicates the distance between the vehicle cutting in and the target vehicle. exist The distance of time, Indicates the target clearance for the vehicle in front. exist The vertical position of time. Indicates the vehicle cutting in The vertical position of time. This indicates the length of the vehicle ahead of the target clearance. This indicates the minimum safe distance between the vehicle cutting in and the vehicle ahead of the target; After cutting into the gap between the vehicle and the target, the vehicle... The distance constraint between them is: （4.7） In the formula, Indicates the gap between the vehicle cutting in and the target vehicle. exist The distance of time, Indicates the vehicle cutting in The vertical position of time. Indicates the target clearance after the vehicle exist The vertical position of time. This indicates the length of the vehicle body that is being cut into. This indicates the minimum safe distance between the vehicle cutting in and the target vehicle; The cut-out safety constraint set includes: The distance constraint between the vehicle cutting out and the nearest preceding vehicle in the outer lane to the vehicle cutting out is: （4.8） In the formula, This indicates that the vehicle cutting out is closest to the vehicle in the outer lane that is in the lane where the vehicle cutting out is positioned. The distance of time, This indicates the vehicle in the outer lane that is closest to the position where the vehicle cutting out is inserted. The vertical position of time. Indicates that the vehicle is switched out. The vertical position of time. This indicates the length of the vehicle in the outer lane that is closest to the position where the cutting-out vehicle will enter. This indicates the minimum safe distance between the vehicle cutting out and the nearest vehicle in the outer lane to the vehicle that cut out from its insertion position; The distance constraint between the vehicle cutting out and the nearest following vehicle in the outer lane is: （4.9） In the formula, This indicates that the vehicle cutting out is closest to the vehicle behind it in the outer lane at the point where the vehicle cutting out entered. The distance of time, Indicates that the vehicle is switched out. The vertical position of time. This indicates the vehicle closest to the cutting-out vehicle's insertion position in the outer lane. The vertical position of time. This indicates the length of the vehicle body that has been cut out. This indicates the minimum safe distance between the vehicle cutting out and the nearest following vehicle in the outer lane; The set of cross-interference safety constraints, when both an inbound vehicle and an outbound vehicle are present simultaneously, has the following corresponding cross-interference safety constraints: （4.10） In the formula, Indicates the vehicle cutting in The vertical position of time. Indicates that the vehicle is switched out. The vertical position of time. Indicates the vehicle cutting in The horizontal position at any moment Indicates that the vehicle is switched out. The horizontal position at any moment This indicates the minimum safe distance for cross-interference. Represents the two-dimensional Euclidean norm; The set of safety constraints is as follows: （4.11） In the formula, express A unified set of security constraints at any given time. Represents the set of safety constraints within the queue. express The set of security constraints for each moment of entry. express The moment to cut out the safety constraint set, express Constantly interfering safety constraint set; Two-way lane changing DMPC is: （4.12） In the formula, express The total cost function of all vehicle nodes at time 1. Indicates the total number of queues. Indicates the first The number of vehicles following in each queue Indicates the first The optimized cost function for the navigator node of each queue. Indicates the first The first queue The optimized cost function for each vehicle following a node. This represents the optimization cost function for the vehicle node. This represents the optimized cost function for cutting out vehicle nodes. express The set of vehicle nodes that are accessed at any given moment. express The set of vehicle nodes that are cut off at any given moment; The step involves using a gradient-tracking-based online adaptive weight algorithm to update the weight parameters in the bidirectional lane-changing distributed model predictive control framework, thereby obtaining the optimal control sequence for each vehicle node. This includes: exist - Initialize the weight matrix of each current vehicle node in the bidirectional lane-changing distributed model predictive control framework within the positive definite cone to obtain the current weight matrix; By combining the current weight matrix with the optimization cost function, safety constraint set, and target vehicle dynamics model in the bidirectional lane-changing distributed model predictive control framework, a temporary optimal control sequence is generated. Based on the temporary optimal control sequence, calculate the local gradient of the optimization cost function with respect to the current weight matrix, and perform an iterative update step of the gradient estimate based on the local gradient; The gradient estimate is updated in real time. When the gradient estimate is updated in real time, the current gradient estimate is obtained. Based on the current gradient estimate, the current weight matrix is ​​adaptively adjusted to obtain an adaptive weight matrix; Based on the adaptive weight matrix, and combined with the optimization cost function, safety constraint set, and target vehicle dynamics model in the bidirectional lane-changing distributed model predictive control framework, the optimal control sequence of each current vehicle node in the prediction time domain is obtained. The step of performing gradient estimation iterative update based on the local gradient includes: Based on the bidirectional time-varying communication topology in the bidirectional lane-changing distributed model predictive control framework, a dual-randomness hybrid matrix is ​​generated. Based on the aforementioned double randomness mixing matrix, a weighted average is calculated on the weight matrix estimates of each current vehicle node and its corresponding target vehicle node to obtain the fusion matrix. The fusion matrix is ​​updated using gradient descent along the negative direction of the current local gradient, and the updated fusion matrix is ​​then... - Positive definite cone projection correction yields an estimated value of the correction matrix; Based on the estimated value of the correction matrix, the local gradient of the optimization cost function with respect to the estimated value of the correction matrix is ​​recalculated to obtain the corrected local gradient; The gradient estimates of the current vehicle node and its corresponding target vehicle node are calculated by weighted average using a double randomness mixing matrix, and the gradient estimate of the current vehicle node is updated by combining the difference between the current local gradient and the corrected local gradient. The updated fusion matrix is ​​then processed. - Positive definite cone projection correction yields an estimated correction matrix, including: Perform the updated fusion matrix - Orthogonal eigenvalue decomposition yields an orthogonal eigenvector matrix and an eigenvalue diagonal matrix; All values ​​in the eigenvalue diagonal matrix that are less than a preset positive definite constant are considered. The eigenvalues ​​are uniformly truncated to the preset positive definite constant. The corrected eigenvalue diagonal matrix is ​​obtained; Based on the orthogonal eigenvector matrix and the corrected eigenvalue diagonal matrix, the estimated value of the correction matrix is ​​reconstructed. -Corresponding to the positive definite cone - Positive definite constraint conditions are: ； In the formula, This indicates a pre-defined positive definite constant. , , , They represent the first The first queue The absolute convergence weight matrix, fuel economy weight matrix, driving comfort weight matrix, and relative convergence weight matrix of each following vehicle node.

Images