A vehicle self-organization and coordination control method based on finite state machine

Abstract

The application provides a vehicle self-organization and cooperative control method based on a finite state machine, relates to construction of a single-vehicle state model M1 and an organization state model M2, and state switching and evolution of the two models; the single-vehicle state of the single-vehicle state model M1 refers to that a CAV does not need to communicate with other vehicles, does not need to cooperate, and the CAV adopts a free driving mode or an obstacle avoidance driving mode; the organization state of the organization state model M2 refers to that the CAV communicates with other CAVs, multiple CAVs form an organization, and the vehicles freely drive, form a platoon or cooperatively avoid collision. The application effectively processes cooperative decision and control problems among networked automatic driving vehicles under a high-density traffic flow complex interaction scene, greatly expands a use scene of a cooperative automatic driving algorithm, is more robust, simultaneously disassembles an optimization problem, improves solving efficiency, and has superiority in solving compared with distributed control.

Description

Technical Field

[0001] This invention relates to the field of cooperative autonomous driving technology, and in particular to a self-organizing and cooperative control method for connected autonomous vehicles based on finite state machines. Background Technology

[0002] In recent years, Cooperative-Automated Driving Systems (C-ADS) and related algorithm technologies have been extensively studied. Compared with single-vehicle autonomous driving, cooperative autonomous driving can leverage communication technologies to achieve comprehensive cooperation such as V2V and V2I, thereby effectively solving problems such as conflict deadlock and conservative interaction faced by single-vehicle autonomous driving.

[0003] Current approaches to cooperative autonomous driving mainly include two methods: solving for motion control variables and determining the passage order. The former typically employs centralized optimization methods to control vehicle movement, improving efficiency, safety, and comfort, while the latter calculates the order in which multiple vehicles pass through conflict points to ensure conflict-free driving. However, in real-world applications, these methods often suffer from high computational overhead, lack of real-time performance, poor control precision, and the risk of collisions.

[0004] Furthermore, in the real world, traffic scenarios are complex and ever-changing, traffic flow density varies, and the high degree of uncertainty caused by human-machine mixed driving environments means that connected autonomous vehicles need to take different measures according to different environments to deal with potential risks. Summary of the Invention

[0005] The purpose of this invention is to propose a self-organizing and cooperative control method for connected autonomous vehicles based on finite state machines. The vehicle collects state information of surrounding vehicles, finds potential cooperative partners, and forms an organization through information exchange. Vehicles within the organization determine the control quantities that each vehicle should take through centralized cooperative optimization, thereby realizing the self-organization and cooperation of multiple CAVs in complex road environments.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A vehicle self-organization and cooperative control method based on finite state machines involves constructing two vehicle state models: a single-vehicle state model M1 and an organized state model M2, as well as the state switching and evolution of these two models.

[0008] In the single-vehicle state model M1, the single-vehicle state refers to a situation where the CAV does not need to communicate with or coordinate with other vehicles. The vehicle can adopt either free-roaming or obstacle-avoidance driving modes. In the single-vehicle state, the CAV performs the following steps:

[0009] The S11 CAV collects information about surrounding vehicles.

[0010] The S12 CAV vehicle is performing a search operation.

[0011] The S13 CAV vehicle is performing an information check.

[0012] S14 CAV vehicle is undergoing a risk inspection.

[0013] The S15 CAV is performing driving maneuvers.

[0014] S16 CAV vehicle is undergoing a status transition.

[0015] Organizational state model M2 refers to the CAV communicating with other CAVs, forming an organization. Vehicles can drive freely, travel in platoons, or cooperate to avoid collisions. In this organized state, CAVs perform the following steps:

[0016] S21 CAVs within the organization perform information sharing actions.

[0017] S22. CAV performs information check actions within the organization.

[0018] S23 The organization's CAV performs risk inspection actions.

[0019] S24 The CAV within the organization performs driving actions.

[0020] S25 The CAV within the organization performs the action of leaving the organization.

[0021] The S26 cloud controller executes organizational splitting actions.

[0022] S27 CAV execution state transition within the organization.

[0023] Compared with the prior art, the present invention has the following advantages

[0024] (1) The self-organization and cooperative control method for connected autonomous vehicles based on finite state machines proposed in this invention achieves the selection of cooperative objects through vehicle self-organization at the upper layer and the cooperative control through optimization at the lower layer. The two-layer model realizes the decomposition and solution of the cooperative problem, effectively handling the cooperative decision-making and control problem between connected autonomous vehicles in complex interaction scenarios of high-density traffic flow.

[0025] (2) The vehicle self-organizing finite state machine proposed in this invention allows each traffic participant to autonomously discover conflict problems and establish connections with the cooperating object without specifying the cooperating object, which greatly expands the application scenarios of the cooperative autonomous driving algorithm.

[0026] (3) This invention realizes hierarchical collaborative control of connected autonomous vehicles. Compared with centralized control, this method does not have a central node and is more robust. At the same time, it decomposes the optimization problem and improves the solution efficiency. Compared with distributed control, this method only performs centralized control on vehicles with serious conflicts, and the solution is more superior. Attached Figure Description

[0027] Figure 1 This is an overall framework diagram of the present invention;

[0028] Figure 2 This is a flowchart illustrating the execution of the single-vehicle state and action set described in this invention.

[0029] Figure 3 This is a flowchart illustrating the execution of the organizational state and action set described in this invention.

[0030] Figure 4 These are scene illustrations and demonstration illustrations for embodiments of the present invention. Detailed Implementation

[0031] A vehicle self-organization and cooperative control method based on finite state machines involves constructing two vehicle state models: a single vehicle state model M1 and an organized state model M2, as well as the state switching and evolution of the two models.

[0032] In the single-vehicle state model M1, the single-vehicle state refers to a situation where the CAV does not need to communicate with or coordinate with other vehicles. The vehicle can adopt either free-roaming or obstacle-avoidance driving modes. In the single-vehicle state, the CAV performs the following steps:

[0033] The S11 CAV collects information about surrounding vehicles.

[0034] The S12 CAV vehicle is performing a search operation.

[0035] The S13 CAV vehicle is performing an information check.

[0036] S14 CAV vehicle is undergoing a risk inspection.

[0037] The S15 CAV is performing driving maneuvers.

[0038] S16 CAV vehicle is undergoing a status transition.

[0039] Organizational state model M2 refers to the CAV communicating with other CAVs, forming an organization. Vehicles can drive freely, travel in platoons, or cooperate to avoid collisions. In this organized state, CAVs perform the following steps:

[0040] S21 CAVs within the organization perform information sharing actions.

[0041] S22. CAV performs information check actions within the organization.

[0042] S23 The organization's CAV performs risk inspection actions.

[0043] S24 The CAV within the organization performs driving actions.

[0044] S25 The CAV within the organization performs the action of leaving the organization.

[0045] The S26 cloud controller executes organizational splitting actions.

[0046] S27 CAV execution state transition within the organization.

[0047] State definition: According to finite state machine theory, CAV vehicles are divided into "individual vehicles" and "organizations," with each state performing different actions. The two states can transition between each other based on the result of the action.

[0048] Definition of action: Based on the finite state machine theory, actions such as "information check", "risk check" and "driving" are set for the "vehicle" and "organization" states respectively. The vehicle will execute the relevant action program in sequence during the movement.

[0049] Perception range and neighbor range: Perception range refers to the area that the sensors on the CAV can detect, and the vehicle can obtain obstacle information within this range. Neighbor range is a user-defined distance threshold used for CAV grouping; the neighbor range is smaller than the perception range.

[0050] Driving Modes: During the "driving" action, the acceleration control amount is obtained according to different formulas for different states of the CAV, which correspond to different driving modes. The driving modes include "single vehicle - free driving", "single vehicle - obstacle avoidance driving", "organized - free driving", "organized - cooperative driving", and "organized - platoon driving".

[0051] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments. This embodiment's technical solution is applicable to scenarios such as unsignalized roundabouts and ramp merging. The method of the present invention collects information about surrounding vehicles through CAV (Carrier Availability Vehicle) onboard sensing devices, processes the information with a cloud controller, and uniformly controls the vehicles when coordination is required. It is technically feasible.

[0052] Example

[0053] A vehicle self-organization and cooperative control method based on finite state machines involves constructing two vehicle state models: a single-vehicle state model M1 and an organized state model M2, as well as the state switching and evolution of the two models.

[0054] The process is as follows:

[0055] Initial State: The initial state of vehicle movement should be a single vehicle state. At the start of each round, the vehicle determines whether to execute step S1 or step S2 based on its own state. State transitions are as follows: Figure 1 As shown.

[0056] S1: As Figure 2 As shown, when the vehicle is in single-vehicle mode, motion control is achieved through the following steps.

[0057] S11: CAV (Carrier Access Vehicle) Information Acquisition. Let this vehicle be Veh0. During its operation, this vehicle acquires its own information, including position (x0, y0), speed v0, and heading angle. Information such as: Acquire information on all vehicles {Veh1,Veh2,…,Vehn} within the perception range, including their positions {(x1,y1),(x2,y2),…,(xn,yn)}, speeds {v1,v2,…,vn}, and navigation angles {…,y1,(x2,y2),…,(xn,yn)}. }

[0058] S12: The CAV performs an organization search action. It searches for vehicles within its neighboring range. If other CAVs exist, it sends an organization request to those vehicles.

[0059] S13: This vehicle performs an information check. It checks whether it has received and responded to any organization requests from other vehicles; it also checks whether its own request has been approved. If an organization is formed, this vehicle's status changes to "organized" in the next round.

[0060] S14: This vehicle performs a risk check. Calculates risk indicators for other vehicles within the vehicle's perception range, including traffic safety evaluation indicators: TTC (Time to Collision) and PET (Time to Peripheral).

[0061]

[0062]

[0063] In the TTC calculation formula, This represents the distance between the current vehicle and the target vehicle. The speed of this vehicle and the following vehicle in the target vehicle. The speed of the vehicle in front is the speed of the vehicle in front of the target vehicle. The speed of the vehicle behind should be greater than the speed of the vehicle in front.

[0064] In the formula for calculating PET, This represents the distance from the vehicle to the point of conflict. This represents the distance from the target vehicle to the point of conflict. and These represent the speed of this vehicle and the speed of the target vehicle, respectively.

[0065] like or Therefore, it is considered that there is a risk, among which, and The risk assessment thresholds are the collision time threshold and the post-intrusion time threshold. In this implementation example, both are set to 3 seconds. If the risk index is less than the threshold, it is considered that there is a potential conflict between the vehicle and the target vehicle, the vehicle is recorded as the conflicting vehicle, and an organization request is sent to the target vehicle.

[0066] S15: This vehicle initiates a driving action. In S14, if there are no conflicting vehicles, this vehicle adopts the "single vehicle - free driving" mode. Otherwise, this vehicle adopts the "single vehicle - obstacle avoidance driving" mode.

[0067] Furthermore, in S15, the calculation formula for "single vehicle - free movement" is as follows:

[0068] The vehicle's acceleration is determined using an IDM model. By inputting vehicle parameters and model parameters, the acceleration that the vehicle should take at the current moment is obtained. The vehicle parameters include the vehicle's desired speed v0 and the distance h between the vehicle and the vehicle in front. The model parameters include the acceleration exponent δ.

[0069]

[0070] The formula for calculating the minimum clearance s*(v) in formula (3) is as follows:

[0071]

[0072] In formulas (3) and (4), s0 is the minimum stopping distance of the vehicle, T is the expected time interval, v is the speed of the vehicle, Δv is the speed difference between the vehicle in front and the vehicle, a is the maximum acceleration of the vehicle, and b is the expected deceleration of the vehicle.

[0073] Furthermore, in S15, the calculation formula for "single vehicle obstacle avoidance driving" is as follows:

[0074] This example uses the rolling optimization method to optimize and solve for acceleration. The vehicle dynamics model uses a point mass model.

[0075]

[0076]

[0077]

[0078]

[0079]

[0080] in, To accelerate the vehicle, Let be the speed of the vehicle at time t. This represents the distance traveled by the vehicle within a given time step. and Let x and y be the positions of the vehicle at time t. The heading angle of the vehicle remains unchanged in the prediction time domain, and Δt is the time step. Formulas (5), (6), (7), (8), and (9) describe the changes in the future position coordinates of the vehicle as the vehicle's future acceleration changes.

[0081] In this example, obstacle trajectory prediction is required. The current kinematic state information of the obstacle is input, and the vehicle's position coordinates over a future period are output, ensuring that the vehicle maintains a safe distance from the target vehicle at every time step during its future journey. This example uses a constant acceleration method.

[0082]

[0083]

[0084]

[0085]

[0086]

[0087] in, and Obstacles The horizontal and vertical coordinates of the position at time t Obstacles Heading angle, Δt is the time step, Obstacles acceleration at time t Obstacles Initial acceleration, Obstacles The velocity at time t. Formulas (10), (11), (12), (13), and (14) predict the obstacle. Location coordinates for a period of time in the future.

[0088] In this example, the optimization solution formula uses the rolling optimization method as an example:

[0089]

[0090] in , , These are the optimization objectives for efficiency, safety, and comfort, respectively, all of which are functions of acceleration. The weights corresponding to the three optimization objectives are used to optimize the acceleration that the vehicle should adopt in the future period of time through equation (15). , , The expression is as follows:

[0091]

[0092]

[0093]

[0094] The following constraints must be met: acceleration range constraint, speed range constraint, and vehicle spacing safety constraint, which are as follows:

[0095]

[0096]

[0097] The dist() function calculates the distance between two workshops, using the following formula:

[0098]

[0099] in, These represent the coordinates, acceleration, and velocity of the vehicle (i=0) and the conflicting vehicle i at time t, respectively. These are, respectively, the safe distance, the lower limit of the vehicle's acceleration, the upper limit of the vehicle's acceleration, the lower limit of speed, and the upper limit of speed. and These represent the desired vehicle speed and the desired distance between the two vehicles, respectively.

[0100] S16: This vehicle performs a state transition. Based on the results of the above actions, this vehicle will adopt either a standalone state or an organized state in the next round, and will decide whether to execute step S1 or step S2.

[0101] S2: As Figure 3 As shown, when the vehicle is in the organized state, motion control is completed through the following steps.

[0102] S21: Information shared by all vehicles within the organization. This includes position {(x0,y0), (x1,y1), (x2,y2),…,(xn,yn)}, speed {v0,v1,v2,…,vn}, and heading angle { , Information such as...

[0103] S22: All vehicles within the organization perform information check actions. Check if the vehicle has received an organization request from an external vehicle and respond accordingly; check if its own request has been approved. If an organization is formed, external vehicles join the organization or the organization merges.

[0104] S23: All vehicles within the organization undergo risk checks. Each vehicle calculates its risk indicators (TTC and PET) relative to surrounding vehicles, using the same formula as in S14. If an indicator exceeds a threshold, a potential conflict with that vehicle is considered to exist. Specifically, potential conflicts for Veh0 can be categorized into three types, each employing a different strategy.

[0105] (1) There is a conflict with the CAV within the organization. At this time, the cloud controller records the conflict (Veh0, Vehi), where Vehi is the conflicting vehicle, and the conflict will be centrally coordinated by the cloud controller.

[0106] (2) There is a conflict with a CAV outside the organization. At this time, the CAV sends an organization request to the vehicle, and the cloud controller records the conflict (Veh0, Vehi). In the S25 action, the vehicle is regarded as a dynamic obstacle.

[0107] (3) Conflict with other vehicles outside the organization. The cloud controller records the conflict (Veh0, Vehi), and the vehicle cannot cooperate and will be regarded as a dynamic obstacle.

[0108] Based on the obtained conflicts (Veh i, Veh j), the cloud controller treats it as an undirected graph, extracts the disconnected subgraphs based on vertex link relationships, and thus obtains the group of cooperative objects. For example... Figure 4 As shown, in the example, {car ego, car 2, car 3} can be considered as a group of collaborative objects. For each group of collaborative objects, the cloud controller will perform parallelized centralized collaborative optimization control.

[0109] S24: All vehicles within the organization execute driving actions. There are three driving modes in the organization state: the "organization-cooperative driving" mode uses centralized optimization to determine the motion control quantities of each vehicle; the "organization-platoon driving" mode uses feedback control to complete vehicle aggregation; and the "organization-free driving" mode uses the IDM model to calculate the control quantities.

[0110] Furthermore, in S24, if a vehicle within the organization does not belong to any collaborative object group and there is no vehicle ahead, the vehicle can adopt the "organization-free driving" mode, as shown in formulas (3) and (4).

[0111] Furthermore, in S24, if a vehicle within the organization does not belong to any collaborative group but has a vehicle ahead, this vehicle adopts an "organization-platooning" mode, and the control quantity is calculated from the vehicle platooning model. This example uses a linear feedback law, and the calculation formula is as follows:

[0112]

[0113]

[0114] Where: i represents the current vehicle, i-1 represents the vehicle in front, e represents the difference between the distance between the vehicles and the desired distance, and vkprev represents the speed of the current vehicle at the previous moment.

[0115] Furthermore, in S24, an "organization-cooperative driving" mode is adopted for vehicles within the collaborative object group. The cloud controller centrally optimizes and then sends control commands to each vehicle. In the cooperative driving mode, the acceleration of vehicle i within the organization at time t is obtained by the following formula. The optimization objective is shown in formula (25).

[0116]

[0117] Where VEH represents all vehicles cooperating within the organization, and h represents the optimization time domain. These are indicators for safety, efficiency, and comfort, respectively. The weights of the three are respectively.

[0118] The following constraints must be met:

[0119]

[0120]

[0121]

[0122] Formula (26) represents the vehicle dynamics formula, which describes the dynamic state of the vehicle at time t. (including location) speed ) and input (acceleration) The relationship between ) is shown in formulas (5) to (9).

[0123] Formulas (27) to (29) represent the velocity range, acceleration range and safety distance constraint, respectively, as in formulas (19) to (21).

[0124] The specific formula for optimizing the objective is as follows:

[0125]

[0126]

[0127]

[0128]

[0129] Unlike "single-vehicle obstacle avoidance driving," in "organization-cooperative driving," the controlled object changes from one vehicle to multiple vehicles, and the optimization objective is the sum of the objectives of multiple vehicles. The execution effect is as follows: Figure 4 As shown in segment 3, ego, car 2, and car 3 are vehicles in the same collaborative group within the organization. The movement of the three vehicles is calculated by the cloud controller, and the result is that ego accelerates first and car 3 decelerates to give way.

[0130] S25: All vehicles within the organization shall be subject to the decision to leave the organization. When there is no conflict between the vehicle and surrounding vehicles, the vehicle shall make a decision based on the location information of the vehicle and surrounding vehicles and the destination information. If the distance between vehicles is greater than the threshold, the vehicle shall leave the current organization and the vehicle status shall change to single vehicle status.

[0131] Furthermore, in S25, a rule-based approach is used as an example to determine whether a vehicle has left the organization. The specific formula is as follows:

[0132]

[0133] in, This is the duration of the vehicle's presence within the organization, ensuring that the organization's membership does not change rapidly. For example, this vehicle is located near the nearest vehicle within the organization, ensuring that members of the organization are spatially close. and All values ​​are thresholds. Vehicles that leave the organization will revert to a single-vehicle state in the next round. The execution effect is as follows: Figure 4 In segments 3 and 4, the ego car detaches from the organization and transforms into a single vehicle.

[0134] S26: The cloud controller performs an organization split determination. The cloud controller centrally acquires the location and destination information of all vehicles and uses clustering methods to divide the vehicle organization into smaller organizations. If a connected autonomous vehicle is in conflict with other vehicles, that vehicle is not allowed to leave the organization until it is free of conflict with other vehicles.

[0135] Furthermore, in S26, a combination of rule-based and clustering methods is used as the basis for organizational division. The clustering method is the DBSCAN method. The specific formula is as follows:

[0136]

[0137] in The scanning radius is d1, which is the vehicle spatial distance in this example; This is the minimum number of points to contain, which is set to 1 in this example.

[0138] If only one vehicle leaves the organization during a split, that vehicle will become a single vehicle in the next round. If multiple adjacent vehicles simultaneously separate from the organization during a split, these vehicles will leave the original organization to form a new organization, which will become an organization in the next round. However, the new organization will need to reallocate cloud controllers.

[0139] S27: All vehicles within the organization undergo a state transition. Based on the results of the above actions, the vehicles will adopt either a single-vehicle state or an organizational state in the next round, and it will be determined whether to execute process S1 or S2.

[0140] The above description is merely a description of preferred embodiments of this application and is not intended to limit the scope of this application in any way. Any changes or modifications made by those skilled in the art based on the above-disclosed technical content should be considered as equivalent and valid embodiments and fall within the scope of protection of the technical solution of this application.

Claims

1. A vehicle self-organizing and cooperative control method based on finite state machines, characterized in that, This involves constructing two vehicle state models: a single-vehicle state model M1 and an organizational state model M2, as well as the state switching and evolution of these two models; In the single-vehicle state model M1, the single-vehicle state refers to the CAV not needing to communicate with or coordinate with other vehicles; the vehicle adopts a free-roaming or obstacle-avoiding driving mode; in the single-vehicle state, the CAV vehicle executes step S1 as follows: The S11 CAV collects information about surrounding vehicles. The S12 CAV vehicle is performing a search operation. S13 CAV vehicle performs information check procedures; S14 CAV vehicle undergoes risk inspection procedures; The S15 CAV is performing driving maneuvers. S16 CAV vehicle is undergoing a status transition; In the organizational state model M2, the organizational state refers to the communication between the CAV itself and other CAVs, with multiple CAVs forming an organization; the vehicles can drive freely, drive in formation, or cooperate to avoid collisions; in the organizational state, the CAV vehicles execute step S2 as follows: S21. Within the organization, CAVs perform actions to share information. S22 CAV performs information check actions within the organization; S23. The organization's CAV performs risk inspection actions; S24: CAVs within the organization perform driving maneuvers. S25 The CAV within the organization performs a departure maneuver from the organization; The S26 cloud controller executes organizational splitting actions; S27 CAV execution state transition within the organization; In S11, this vehicle is denoted as Veh0. During its operation, this vehicle acquires its own vehicle information, including its position (x0, y0), speed v0, and heading angle. Acquire information on all vehicles {Veh1,Veh2,…,Vehn} within the perception range, including their positions {(x1,y1),(x2,y2),…,(xn,yn)}, speeds {v1,v2,…,vn}, and heading angles { }; In S12, search for vehicles within the neighboring range of this vehicle. If other CAVs exist, send an organization request to this vehicle. In S13, check if the vehicle has received an organization request from other vehicles and respond to it; check if its own request has been approved; if an organization is to be formed, the vehicle's status will change to "organization" in the next round. In S14, risk indicators for other vehicles within the perception range of this vehicle are calculated, including traffic safety evaluation indicators: TTC collision time and PET post-intrusion time; In the TTC calculation formula, This represents the distance between the current vehicle and the target vehicle. The speed of this vehicle and the following vehicle in the target vehicle. The speed of this vehicle is the speed of the vehicle in front of the target vehicle, and the speed of the vehicle behind should be greater than the speed of the vehicle in front. In the formula for calculating PET, This represents the distance from the vehicle to the point of conflict. This represents the distance from the target vehicle to the point of conflict. and These represent the speed of this vehicle and the speed of the target vehicle, respectively. like or Therefore, it is considered that there is a risk, among which, and The risk assessment thresholds are the collision time threshold and the post-intrusion time threshold. If the risk index is less than the threshold, it is considered that there is a potential conflict between the vehicle and the target vehicle. The vehicle is recorded as the conflicting vehicle and an organization request is sent to the target vehicle. In S15, based on the indicators obtained in S14, if there are no conflicting vehicles, this vehicle adopts the "single vehicle - free driving" mode; otherwise, this vehicle adopts the "single vehicle - obstacle avoidance driving" mode. S16: Based on the results of the above actions, this vehicle will adopt either a single vehicle state or an organized state in the next round, and decide whether to execute step S1 or step S2. In S21, all vehicles within the organization share information, including position {(x0,y0), (x1,y1), (x2,y2),…,(xn,yn)}, speed {v0,v1,v2,…,vn}, and heading angle { , ; In S22, check whether the vehicle has received an organization request from an external vehicle and respond to it; check whether its own request has been approved; if an organization is formed, external vehicles join the organization or the organization is merged. In S23, the organization calculates the risk indicators TTC and PET between its internal vehicles and surrounding vehicles; if the indicators exceed the threshold, it is considered that there is a potential conflict between the internal vehicles and the surrounding vehicles. In S24, there are three driving modes in the organization state: the "organization-cooperative driving" mode uses centralized optimization to determine the motion control quantities of each vehicle; the "organization-platooning driving" mode uses feedback control to complete vehicle aggregation; and the "organization-free driving" mode uses the IDM model to calculate the control quantities. In S25, when there is no conflict between the vehicle and surrounding vehicles, the vehicle makes a judgment based on the position information of the vehicle and surrounding vehicles and the destination information. If the distance between vehicles is greater than the threshold, the vehicle leaves the current organization and the vehicle status changes to single vehicle status. In S26, the cloud controller centrally acquires the location information and destination information of all vehicles and uses a clustering method to divide the vehicle organization into smaller organizations. If a connected autonomous vehicle has a conflict with other vehicles, the connected autonomous vehicle is not allowed to leave the organization until it has no conflict with other vehicles. In S27, based on the results of the above actions, the vehicle will adopt a single-vehicle state or an organized state in the next round, and decide whether to execute the S1 process or the S2 process. In S23, potential conflicts for Veh0 are categorized into three types, and different strategies are employed for each: (1) There is a conflict with the CAV vehicle within the organization; at this time, the cloud controller records the conflict (Veh0, Vehi), where Vehi is the conflicting vehicle, and the conflict will be centrally coordinated by the cloud controller. (2) There is a conflict with a CAV vehicle outside the organization; at this time, the CAV vehicle sends an organization request to the CAV vehicle outside the organization, the cloud controller records the conflict (Veh0, Vehi), and in the S25 action, the CAV vehicle outside the organization is regarded as a dynamic obstacle; (3) There is a conflict with other vehicles outside the organization; the cloud controller records the conflict (Veh0, Vehi), and other vehicles outside the organization cannot cooperate and will be regarded as dynamic obstacles; Based on obtaining all conflicts (Vehi, Vehj), the cloud controller treats it as an undirected graph, extracts the disconnected subgraphs based on vertex link relationships, and then obtains the group of cooperative objects. In S24, if a vehicle within the organization does not belong to any collaborative object group and there is no vehicle ahead, this vehicle adopts the "organization-free driving" mode. If a vehicle within the organization does not belong to any collaborative object group, but there is a vehicle ahead, this vehicle adopts the "organization-platooning driving" mode, and the control quantity is calculated by the vehicle platooning model; For vehicles within the collaborative group, an "organization-collaborative driving" mode is adopted; the cloud controller centrally optimizes and then sends control commands to each vehicle. In S25, a rule-based approach is used as the basis for determining departure from the organization, and the specific formula is as follows: in, This is the duration of the vehicle's presence within the organization, ensuring that the organization's membership does not change rapidly. This is the distance between this vehicle and the nearest vehicle within the organization, ensuring that organization members are spatially close. and All are threshold values; vehicles that leave the organization will be converted to single-vehicle status in the next round. In S26, a combination of rule-based and clustering methods is used as the basis for organizational division; the clustering method is the DBSCAN method, and the specific formula is as follows: in The scan radius; The minimum number of points contained; If only one vehicle leaves the organization during a split, that vehicle will become a single vehicle in the next round. If multiple adjacent vehicles leave the organization simultaneously during a split, these vehicles will leave the original organization to form a new organization, which will become an organization in the next round. However, the new organization will need to redistribute cloud controllers.

2. The vehicle self-organizing and cooperative control method based on finite state machines as described in claim 1, characterized in that, In S15, the calculation formula for "single vehicle - free movement" is as follows: The vehicle's acceleration is determined using an IDM model: by inputting vehicle parameters and model parameters, the acceleration that the vehicle should take at the current moment is obtained; the vehicle parameters include the vehicle's desired speed v0 and the distance h between the vehicle and the vehicle in front; the model parameters include the acceleration exponent δ. Minimum clearance in the workshop in formula (3) The calculation formula is as follows: In formulas (3) and (4), s0 is the minimum stopping distance of the vehicle, T is the expected time interval, v is the speed of the vehicle, Δv is the speed difference between the vehicle in front and the vehicle, a is the maximum acceleration of the vehicle, and b is the expected deceleration of the vehicle.

3. The vehicle self-organizing and cooperative control method based on finite state machines as described in claim 1, characterized in that, In S15, the calculation formula for "single vehicle obstacle avoidance driving" is as follows: The acceleration is optimized using a rolling optimization method, and the vehicle dynamics model adopts a point mass model. in, To accelerate the vehicle, Let be the speed of the vehicle at time t. This represents the distance traveled by the vehicle within a given time step. and Let x and y be the positions of the vehicle at time t. The heading angle of the vehicle remains unchanged in the prediction time domain, and Δt is the time step. Formulas (5), (6), (7), (8), and (9) describe the changes in the future position coordinates of the vehicle with the future acceleration of the vehicle. Obstacle trajectory prediction: Input the current kinematic state information of the obstacle, output the vehicle's position coordinates within a future period, ensuring that the vehicle maintains a safe distance from the target vehicle at every time step during its future journey; a constant acceleration method is used. in, and Obstacles The horizontal and vertical coordinates of the position at time t Obstacles Heading angle, Δt is the time step, Obstacles acceleration at time t Obstacles Initial acceleration, Obstacles At time t, the velocity; formulas (10), (11), (12), (13), and (14) predict the obstacle. Location coordinates for a future period of time; The optimization solution formula adopts a rolling optimization method: in , , These are the optimization objectives for efficiency, safety, and comfort, respectively, all of which are functions of acceleration. Assuming the weights corresponding to the three optimization objectives, the acceleration that the vehicle should adopt in the future is determined by Equation (15). , , The expression is as follows: The following constraints must be met: acceleration range constraint, speed range constraint, and vehicle spacing safety constraint, which are as follows: The dist() function calculates the distance between two workshops, using the following formula: in, These represent the coordinates, acceleration, and velocity of the vehicle (i=0) and the conflicting vehicle i at time t, respectively. These are, respectively, the safety distance, the lower limit of the vehicle's acceleration, the upper limit of the vehicle's acceleration, the lower limit of speed, and the upper limit of speed; and This represents the desired vehicle speed and the desired distance between the two vehicles.

Images