A method for overtaking decision-making and control of intelligent connected vehicle groups

Abstract

This invention relates to a method for overtaking decision control of intelligent connected vehicle groups, comprising: S1, processing the road where the intelligent connected vehicle group is located into a grid and establishing a vehicle group decision control model based on the moving grid; S2, according to the vehicle group decision control model, using a rolling time-domain programming method, outputting the current vehicle behavior decision information corresponding to the vehicle group, for controlling the driving state of each vehicle in the group; S3, determining whether a new vehicle has appeared in the moving grid; if yes, returning to step S1 to start a new round of decision control; otherwise, proceeding to step S4; S4, determining whether the current vehicle behavior decision information has been executed by the vehicle group; if yes, returning to step S1 to start a new round of decision control; otherwise, returning to step S2. Compared with the prior art, this invention can quickly and accurately control intelligent connected vehicles to safely and efficiently overtake obstacle vehicles in overtaking scenarios, thereby improving the driving efficiency of intelligent connected vehicle groups.

Description

Technical Field

[0001] This invention relates to the field of intelligent connected vehicle control technology, and in particular to a method for decision-making and control of overtaking in a group of intelligent connected vehicles. Background Technology

[0002] With the rapid development of modern technology, intelligent connected vehicles are gradually coming into view. Because intelligent connected vehicles are equipped with advanced sensors, advanced controllers, and advanced actuators, and integrate modern network communication technology, they have a certain degree of autonomous perception and the ability to communicate with other networked devices. Therefore, based on the information obtained from network transmission, the controller makes decisions, forms control commands, and delivers them to the actuators for execution.

[0003] Intelligent connected vehicle convoy decision control is a novel decision control approach based on intelligent connected vehicles and a crucial component of current intelligent transportation technologies. Its purpose is to connect and coordinate intelligent connected vehicles through vehicle-to-everything (V2X) technology, transforming individual vehicle intelligence into collective intelligence. This further enhances the practicality of intelligent connected vehicles and ultimately achieves more advanced and intelligent urban traffic management and control. To achieve this goal, a central control center typically uses multi-vehicle communication (MVCC) technology to collect vehicle information within the convoy, process it, and send vehicle control commands to control the acceleration, braking, and steering of the intelligent connected vehicles within the convoy.

[0004] However, in practical applications, especially when facing overtaking scenarios, due to the complexity of traffic conditions and the presence of many uncertainties, existing intelligent connected vehicle group decision-making and control methods cannot effectively cope with all situations and suffer from poor computational timeliness. This greatly reduces the practicality of intelligent connected vehicle groups, affects their driving efficiency, and significantly hinders the future development of this technology. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology and provide an intelligent connected vehicle group overtaking decision control method, which can quickly and accurately control intelligent connected vehicles to safely and efficiently overtake obstacle vehicles in overtaking scenarios, thereby improving the driving efficiency of intelligent connected vehicle groups.

[0006] The objective of this invention can be achieved through the following technical solution: a method for overtaking decision-making and control of intelligent connected vehicle groups, comprising the following steps:

[0007] S1. The roads where the intelligent connected vehicle group is located are processed into a grid, and a vehicle group decision control model based on the mobile grid is established.

[0008] S2. Based on the vehicle group decision control model, a rolling time-domain programming method is adopted to output the current vehicle behavior decision information corresponding to the vehicle group, which is used to control the driving status of each vehicle in the vehicle group.

[0009] S3. Determine whether a new vehicle has appeared in the moving grid. If yes, return to step S1 to start a new round of decision control; otherwise, execute step S4.

[0010] S4. Determine whether the current vehicle behavior decision information has been executed by the vehicle group. If the determination is yes, return to step S1 to start a new round of decision control; otherwise, return to step S2.

[0011] Furthermore, the rasterization process in step S1 includes the following steps: determining the rasterization starting point, determining the raster size, determining the rasterization direction, and determining the raster movement speed.

[0012] Furthermore, the gridding starting point is specifically the last row of vehicles within the vehicle group;

[0013] The grid size includes grid width and grid length, wherein the grid width is the same as the lane width, and the grid length is the same as the average length of each row of vehicles in the vehicle group;

[0014] The specific rasterization direction is as follows: the driving direction of the vehicle group is defined as the forward direction, and the rasterization is performed forward from the starting grid.

[0015] The grid movement speed is specifically defined as follows: if the current vehicle group decision command is an overtaking command, the grid movement speed is the speed of the overtaken target vehicle; if the current vehicle group decision command is another command, the grid movement speed is the vehicle group movement speed.

[0016] Furthermore, the vehicle decision control model based on mobile gridding in step S1 includes constraints and an objective function. The constraints include no clone constraint, movement constraint, collision avoidance constraint within the vehicle group, collision avoidance constraint outside the vehicle group, and initial state constraint.

[0017] The objective function includes the objective cost, the vertical movement cost, and the horizontal movement cost.

[0018] Furthermore, the "no clone constraint" means that for any given time step, any vehicle in the vehicle group can only occupy one grid cell.

[0019]

[0020]

[0021]

[0022]

[0023] Where, n GIt refers to the number of connected-automated vehicles (CAVs) within the intelligent connected vehicle fleet, including... OK, Column, i.e. For the number of CAV rows within a group of intelligent connected vehicles, Let i be the CAV column number within the intelligent connected vehicle group, and i be the CAV number, i∈[1,n]. G ], k is the time step number after discretization of the continuous time domain, k∈[1,N], and N is the planning step size. Let be the total number of rows in the raster. For step k, the occupancy state of the i-th CAV in row p. This represents the total number of columns in the grid. This represents the occupancy status of the i-th CAV in row p and column q at step k, where p is the raster row number. q grid column numbering,

[0024]

[0025] Furthermore, the movement constraint means that for any two adjacent time steps, the same vehicle can only move one grid cell laterally or one grid cell vertically, and cannot move diagonally or more than one grid cell laterally / vertically. Therefore, for The longitudinal movement constraint is:

[0026]

[0027]

[0028] The lateral movement constraint is:

[0029]

[0030]

[0031] The diagonal movement constraint is:

[0032]

[0033]

[0034]

[0035]

[0036] in, p1 and p2 are any two ordered, non-adjacent row numbers. q1 and q2 are any two non-adjacent sequential numbers, n HV Let j be the number of target vehicles being overtaken within the grid network. All target vehicles being overtaken are manually driven vehicles (HVs), and j is the HV number, j∈[1,n]. HV ], This represents the possession state of the k-th step and the j-th HV in the p-th row. This represents the occupancy status of the k-th step and the j-th HV in the q-th column.

[0037] Furthermore, the aforementioned collision avoidance constraint within the vehicle group means that, for any given time step, no two CAVs within the intelligent connected vehicle group can occupy the same grid cell. Therefore, for

[0038]

[0039] Where i1∈[1,n G ],i2∈[1,n G ], i1 and i2 are CAV numbers, and i1 ≠ i2;

[0040] The aforementioned external collision avoidance constraint for the vehicle group means that, at any given time step, a connected vehicle and the vehicle being overtaken cannot simultaneously occupy the same grid cell. Therefore, for

[0041]

[0042] Where, p j,k Let q be the row number of the target vehicle j at step k. j,k Let be the column number of the target vehicle j at step k.

[0043] Furthermore, the initial state constraint means that in the initial state, i.e., k=1, the initial positions of all vehicles in the intelligent connected vehicle group are known, therefore for

[0044]

[0045]

[0046] Furthermore, the objective function is specifically:

[0047]

[0048] Among them, W tar The weighting factor of the target cost, W lon Weighting factor for vertical movement costs, W lat The weighting factor for lateral movement costs. To achieve the target cost, only when the vehicle group reaches the front of the grid network... Only after it has been driven forward can the cost be minimized; therefore, the target cost is used to drive the vehicle forward. For vertical movement costs, This refers to the cost of lateral movement.

[0049] Furthermore, the current vehicle behavior decision information in step S2 specifically refers to the grid positions occupied by different vehicles in the vehicle group at the current moment.

[0050] Compared with the prior art, the present invention has the following advantages:

[0051] I. This invention performs grid-based processing on the roads where intelligent connected vehicle groups are located and establishes a vehicle group decision-making and control model based on the moving grid. Based on this model, a rolling time-domain programming approach is used to output the current vehicle behavior decision information corresponding to the vehicle group, thereby controlling the driving state of each vehicle within the group. This achieves motion planning for vehicles within the group based on the moving grid. When facing obstructions from vehicles ahead, it can quickly and accurately control intelligent connected vehicles to safely and efficiently overtake obstacles, effectively improving the driving efficiency of the intelligent connected vehicle group.

[0052] Second, this invention uses a moving grid for vehicle motion planning within a vehicle group, transforming a time-domain problem into a spatial-domain problem. This makes the established vehicle group decision-making and control model more intuitive and faster to solve. Furthermore, using a moving grid as the background significantly improves the maneuverability and adaptability of intelligent connected vehicle groups in traffic flow.

[0053] Third, when constructing the vehicle group decision control model, this invention fully considers many constraints, including the absence of a clone constraint, movement constraint, collision avoidance constraint within the vehicle group, collision avoidance constraint outside the vehicle group, and initial state constraint. This makes the optimal solution of the model obtained in the subsequent solution approach the real situation, improves the practicality of this invention, and makes it well applicable to general urban expressways, highways and other scenarios.

[0054] Fourth, this invention employs a rolling time-domain control update method. After one planning cycle, the vehicle group begins executing control commands. The rolling grid network is updated, the current planning cycle ends, and the next planning cycle begins only when a new vehicle appears in the moving grid, or when the vehicle group has completed the commands for the current planning cycle. Simultaneously, during a planning cycle, all planning instructions are executed, meaning the prediction domain and control domain have the same length. Rolling time-domain control ensures that the solution at each time step is the optimal solution for that time step, thereby guaranteeing the optimality of the final planning and control result. Attached Figure Description

[0055] Figure 1This is a schematic diagram of the method flow of the present invention;

[0056] Figure 2 This is a schematic diagram illustrating the application process of an example.

[0057] Figure 3 This is a schematic diagram of a moving rasterization.

[0058] Figure 4 A schematic diagram of time-domain planning for vehicle group rolling;

[0059] Figure 5 This is a schematic diagram showing the correspondence between the decision control output of the vehicle group and the vehicle behavior. Detailed Implementation

[0060] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0061] Example

[0062] like Figure 1 As shown, an intelligent connected vehicle group overtaking decision-making and control method includes the following steps:

[0063] S1. The roads where the intelligent connected vehicle group is located are processed into a grid, and a vehicle group decision control model based on the mobile grid is established.

[0064] S2. Based on the vehicle group decision control model, a rolling time-domain programming method is adopted to output the current vehicle behavior decision information corresponding to the vehicle group, which is used to control the driving status of each vehicle in the vehicle group.

[0065] S3. Determine whether a new vehicle has appeared in the moving grid. If yes, return to step S1 to start a new round of decision control; otherwise, execute step S4.

[0066] S4. Determine whether the current vehicle behavior decision information has been executed by the vehicle group. If the determination is yes, return to step S1 to start a new round of decision control; otherwise, return to step S2.

[0067] This embodiment applies the above-described technical solution, such as Figure 2 As shown, the specific content includes:

[0068] The roads where the intelligent connected vehicle fleet is located are processed into a grid, and on this basis, a vehicle fleet decision and control model based on mobile grid is established.

[0069] The process of gridding the roads where intelligent connected vehicles are located is also known as mobile gridding, such as... Figure 3 As shown, the main steps include selecting the starting point for rasterization, determining the size of the raster, moving the raster from the origin, and determining the moving speed of the raster.

[0070] The starting point for gridding is the last row of vehicles in the selected vehicle group.

[0071] The width of the grid is denoted as w. cell The width of the grid is the same as the lane width. The length of the grid is denoted as l. cell It is the same as the average length of each row of vehicles in the controlled vehicle group.

[0072] The direction of travel for the vehicle group is defined as the forward direction, and the vehicle moves forward from the starting grid cell. The number of columns in the grid is denoted as... The number of lanes is denoted as n lane ,definition The number of rows of intelligent connected vehicles is recorded as ( Figure 3 middle To ensure the vehicle group can complete overtaking maneuvers, the number of rows in the grid network should be [number missing].

[0073] The grid's moving speed is related to the output of the intelligent connected vehicle group's state decision layer. If the output decision command is for the vehicle group to overtake, the grid's moving speed is the speed of the overtaken target vehicle. If the output decision command is another command, the grid's moving speed is the vehicle group's moving speed.

[0074] This transforms the decision-making and control problem of intelligent connected vehicle groups during overtaking into controlling the movement of the group within a grid, ultimately ensuring a safe and smooth arrival at the leading grid.

[0075] The process of establishing a mobility-based vehicle group decision-making and control model and performing rolling time-domain planning includes the following steps:

[0076] 1) Constructing constraints

[0077] There is no clone constraint. That is, at any time step, any vehicle in the swarm can only occupy one grid cell. Therefore, for

[0078]

[0079]

[0080] Where, n G It refers to the number of connected-automated vehicles (CAVs) within the intelligent connected vehicle fleet, including... OK, Column, i.e. For the number of CAV rows within a group of intelligent connected vehicles, Let be the column number of CAVs within the intelligent connected vehicle group. Let i be the CAV number, i∈[1,n]. G]. k is the time step number after discretization of the continuous time domain, k∈[1,N], and N is the planning step size. Let be the total number of rows in the raster. For step k, the occupancy state of the i-th CAV in row p. This represents the total number of columns in the grid. To determine the occupancy state of the i-th CAV in row p and column q at step k, p is the raster row number. q grid column numbering,

[0081] Movement constraints. For any two adjacent time steps, the same vehicle can only move one grid cell laterally or one grid cell vertically. Diagonal movement is not allowed, and movement cannot be more than one grid cell laterally or vertically.

[0082] Therefore, for There is a longitudinal movement constraint:

[0083]

[0084] There is a lateral movement constraint:

[0085]

[0086] There is a diagonal movement constraint:

[0087]

[0088]

[0089]

[0090]

[0091] in, p1 and p2 are any two ordered, non-adjacent row numbers. q1 and q2 are any two non-adjacent sequential numbers. n HV This represents the number of target vehicles (also known as target vehicles) within the grid network, all of which are manually driven vehicles (i.e., HVs).

[0092] j is the HV number, j∈[1,n] HV ]. This represents the occupying state of the j-th HV at step k in row p. This represents the occupation state of the j-th HV in the k-th step in the q-th column.

[0093] Collision avoidance constraints within the vehicle group. This means that at any given time step, no two CAVs within the intelligent connected vehicle group can occupy the same grid cell. Therefore, for

[0094]

[0095] Where i1∈[1,n G ],i2∈[1,n G ], i1 and i2 are CAV numbers, and i1≠i2.

[0096] External collision avoidance constraints for the vehicle group. This means that at any given time step, a connected vehicle and the vehicle being overtaken cannot simultaneously occupy the same grid cell. Therefore, for

[0097]

[0098] Where, p j,k Let q be the row number of the target vehicle j at step k. j,k Let be the column number of the target vehicle j at step k.

[0099] Initial state constraints. In the initial state, i.e., k=1, the initial positions of all vehicles in the intelligent connected vehicle group are known. Therefore, for

[0100]

[0101]

[0102] 2) Construct the objective function

[0103] The objective function is of the following form:

[0104]

[0105] Among them, W tar Weighting coefficients for target cost. W lon Weighting factor for vertical movement costs. W lat Weighting coefficients for lateral movement costs.

[0106] The objective function is mainly composed of It consists of three parts, each representing the target cost (this cost only applies when the vehicle group reaches the front of the grid network). Only after the vehicle moves can the cost be minimized. Therefore, the target cost can be considered as the cost of moving the vehicle forward, longitudinal movement cost, and lateral movement cost. To prevent unnecessary lateral swaying of the vehicle, or even invalid results such as vehicle position swapping, longitudinal / lateral movement costs are set. Vehicle movement will cause a change in its grid position, thus generating movement costs.

[0107] 3) Rolling time-domain planning

[0108] Employing a control update method based on the rolling time domain, such as Figure 4 As shown, after a planning cycle ends, the vehicle group begins executing control commands. The mobile grid is updated and the planning cycle ends only when a new vehicle appears within the mobile grid, or when the vehicle group has completed executing the commands for the current planning cycle. During a planning cycle, all planned commands are executed, meaning the prediction domain and control domain have the same length.

[0109] 4) Output vehicle behavior

[0110] Steps 1) to 3) establish a decision control model for an intelligent connected vehicle convoy overtaking decision control method. The output result is the position of different vehicles in the convoy on the grid at a certain moment. This result is not directly related to the behavior of individual vehicles. In order for the output result to be usable by the individual vehicle decision control module, it is necessary to establish a correspondence between the behavior of individual vehicles and the output result of the model, as follows: Figure 5 As shown.

[0111] In summary, this technical solution first performs gridded processing on the road where the intelligent connected vehicle group is located, i.e., moving gridded processing. Then, based on this, a rolling time-domain programming approach is used to solve the vehicle group decision-making and control model based on the moving gridded processing, obtaining vehicle behavior decision information to control the driving state of vehicles within the group. This technical solution has the advantages of high computational efficiency and strong adaptability, and can be applied to complex overtaking scenarios, effectively controlling the intelligent connected vehicle group to safely and efficiently overtake obstacle vehicles, greatly improving the driving efficiency of the intelligent connected vehicle group.

Claims

1. A method for decision-making and control of overtaking in an intelligent connected vehicle group, characterized in that, Includes the following steps: S1. The roads where the intelligent connected vehicle group is located are processed into a grid, and a vehicle group decision control model based on the moving grid is established. The grid processing includes the following steps: determining the grid starting point, determining the grid size, determining the grid direction, and determining the grid moving speed. The vehicle decision control model based on mobile gridding includes constraints and an objective function. The constraints include no clone constraint, movement constraint, collision avoidance constraint within the vehicle group, collision avoidance constraint outside the vehicle group, and initial state constraint. The objective function includes the objective cost, the vertical movement cost, and the lateral movement cost; The objective function is specifically: in, The weighting coefficients for the target cost. The weighting factor for vertical movement costs. The weighting factor for lateral movement costs. To achieve the target cost, only when the vehicle group reaches the front of the grid network... Only after it has been driven forward can the cost be minimized; therefore, the target cost is used to drive the vehicle forward. For vertical movement costs, For lateral movement costs; It refers to the number of connected autonomous vehicles (CAVs) within the intelligent connected vehicle fleet, including... OK, Column, i.e. For the number of CAV rows within a group of intelligent connected vehicles, This refers to the number of CAV columns within a group of intelligent connected vehicles. for serial number, , It is the time step numbering after discretization of the continuous time domain. , To plan the step length, Let be the total number of rows in the raster. For the first Step, the first The CAV in the first The occupancy status of a row. This represents the total number of columns in the grid. In the first Step, the first The CAV in the first line, number The occupancy status of the column. Number the grid rows. , Grid column numbering, ; S2. Based on the vehicle group decision control model, a rolling time-domain programming method is adopted to output the current vehicle behavior decision information corresponding to the vehicle group, which is used to control the driving status of each vehicle in the vehicle group. S3. Determine whether a new vehicle has appeared in the moving grid. If yes, return to step S1 to start a new round of decision control; otherwise, execute step S4. S4. Determine whether the current vehicle behavior decision information has been executed by the vehicle group. If the determination is yes, return to step S1 to start a new round of decision control; otherwise, return to step S2.

2. The intelligent connected vehicle group overtaking decision-making and control method according to claim 1, characterized in that, The gridding starting point is specifically the last row of vehicles in the vehicle group; The grid size includes grid width and grid length, wherein the grid width is the same as the lane width, and the grid length is the same as the average length of each row of vehicles in the vehicle group; The specific rasterization direction is as follows: the driving direction of the vehicle group is defined as the forward direction, and the rasterization is performed forward from the starting grid. The grid movement speed is specifically defined as follows: if the current vehicle group decision command is an overtaking command, the grid movement speed is the speed of the overtaken target vehicle; if the current vehicle group decision command is another command, the grid movement speed is the vehicle group movement speed.

3. The intelligent connected vehicle group overtaking decision-making and control method according to claim 1, characterized in that, The "no clone constraint" means that at any time step, any vehicle in the vehicle group can only occupy one grid cell. , : in, It refers to the number of connected autonomous vehicles (CAVs) within the intelligent connected vehicle fleet, including... OK, Column, i.e. For the number of CAV rows within a group of intelligent connected vehicles, This refers to the number of CAV columns within a group of intelligent connected vehicles. for serial number, , It is the time step numbering after discretization of the continuous time domain. , To plan the step length, Let be the total number of rows in the raster. For the first Step, the first The CAV in the first The occupancy status of a row. This represents the total number of columns in the grid. In the first Step, the first The CAV in the first line, number The occupancy status of the column. Number the grid rows. , Grid column numbering, .

4. The intelligent connected vehicle group overtaking decision-making and control method according to claim 3, characterized in that, The movement constraint means that for any two adjacent time steps, the same vehicle can only move one grid cell laterally or one grid cell vertically; it cannot move diagonally or more than one grid cell laterally or vertically. Therefore, for , The longitudinal movement constraint is: The lateral movement constraint is: The diagonal movement constraint is: in, , , , Number any two ordered, non-adjacent rows. , , , For any two non-adjacent numbers, there are sequential numbers. The number of target vehicles being overtaken within the grid network, where all target vehicles are manually driven vehicles (HV). for serial number, , For the first Step, First HV in the first The occupancy status of a row. For the first Step, First HV in the first The occupancy status of the column.

5. The intelligent connected vehicle group overtaking decision-making and control method according to claim 4, characterized in that, The aforementioned collision avoidance constraint within the vehicle group refers to the rule that, at any given time step, no two CAVs within the intelligent connected vehicle group can occupy the same grid cell. Therefore, for , , , , : in, , For CAV number, and ; The aforementioned external collision avoidance constraint for the vehicle group means that, at any given time step, a connected vehicle and the vehicle being overtaken cannot simultaneously occupy the same grid cell. Therefore, for , , : in, For the target vehicle In the The row number where the step is located. For the target vehicle In the The column number where the step is located.

6. The intelligent connected vehicle group overtaking decision-making and control method according to claim 5, characterized in that, The initial state constraint refers to the constraint in the initial state, i.e. At that time, the initial positions of all vehicles in the intelligent connected vehicle group are known, therefore for , : 。 7. A method for intelligent connected vehicle group overtaking decision control according to any one of claims 1 to 6, characterized in that, The current vehicle behavior decision information in step S2 specifically refers to the grid positions occupied by different vehicles in the vehicle group at the current moment.

Images