Automatic driving vehicle dispatching system for unsignalized intersection based on game optimization

Abstract

The application discloses an automatic driving vehicle intelligent scheduling method and system for a signal-free intersection, and belongs to the technical field of automatic driving and intelligent traffic control. In view of the problems of low traffic efficiency and unfair traffic at the existing signal-free intersection, the application proposes a multi-vehicle cooperative scheduling method based on game theory optimization. The method obtains the state information of each vehicle through vehicle networking communication, constructs an optimization model with the minimum weighted maximum waiting time as the target, and comprehensively considers the scheduling fairness and efficiency. Explicit safety constraints, vehicle fleet consistency constraints and global waiting time constraints are set to ensure that the scheduling result meets the safety and systematic requirements. At the same time, adjustable priority weights are given to different types of vehicles to realize differentiated scheduling control of emergency vehicles, official vehicles and the like. The method has the advantages of high interpretability, clear structure, strong adaptability and the like, and has good engineering realizability and popularization value.

Description

Technical Field

[0001] This invention relates to the fields of intelligent transportation and autonomous driving technology, and in particular to an autonomous vehicle dispatching system for unsignalized intersections based on game theory optimization. This system is primarily applied to multi-vehicle traffic management scenarios in a vehicle-road cooperative environment, falling under the technical scope of intelligent connected vehicle (ICV) dispatching and control systems and the intelligent transformation of urban traffic infrastructure. Specifically, it is applicable to automated dispatching and control systems that achieve safe, efficient, and fair vehicle passage at intersections lacking traffic light control by optimizing dispatching rules. Background Technology

[0002] Intersections, as key nodes in urban road traffic systems, are the main areas where vehicles converge, cross, and pass. Traditional intersection traffic management typically relies on traffic lights, traffic signs, or manual direction. However, in some areas, especially where traffic infrastructure is inadequate or signal equipment malfunctions, a large number of intersections remain without traffic signals. Due to the lack of a unified scheduling mechanism, these intersections are highly prone to traffic conflicts, resulting in low traffic efficiency and even traffic accidents, becoming a significant constraint on urban traffic safety and operational efficiency.

[0003] With the development of autonomous driving and intelligent connected vehicle technologies, vehicles can achieve real-time information sharing and collaborative decision-making through vehicle-to-vehicle (V2V / V2I) technology, thus providing a technological foundation for intelligent scheduling at unsignalized intersections. Existing research attempts to improve scheduling efficiency and safety through trajectory prediction, collision warning, or priority passing strategies based on simple rules, but these methods still have many limitations. On the one hand, these methods often lack a unified optimization objective, making it difficult to coordinate the passage order of multiple vehicles from a system perspective; on the other hand, scheduling strategies generally struggle to explicitly model key constraints such as safety intervals and vehicle consistency. Furthermore, existing methods generally neglect passage fairness, potentially leading to some vehicles waiting for extended periods, and lack the ability to differentiate control over vehicles with different priorities. While some reinforcement learning-based scheduling methods possess adaptability, they suffer from poor interpretability and uncontrollable models, hindering practical deployment and engineering implementation.

[0004] Therefore, there is an urgent need for an intelligent scheduling system for signalless intersections that can achieve multi-vehicle collaboration, fair passage, and efficiency balance while ensuring safety constraints. This system should also have clear modeling logic, optimization objectives, and highly interpretable scheduling strategies to meet the practical needs of future autonomous driving and vehicle-road cooperative scenarios. Summary of the Invention

[0005] 1. Purpose of the invention:

[0006] To address the problems mentioned in the background, the present invention aims to provide a game-theoretic optimization-based autonomous vehicle scheduling system for unsignalized intersections. This system addresses issues such as low traffic efficiency, high collision risk, and unfair traffic conditions at existing unsignalized intersections. It acquires vehicle status information through vehicle-to-everything (V2X) networks, constructs a scheduling model with the goal of minimizing the weighted maximum waiting time, and combines safety constraints and priority settings to achieve multi-vehicle collaborative, safe, efficient, and fair traffic scheduling. This system is suitable for intelligent transportation scenarios in autonomous driving and vehicle-road cooperative environments.

[0007] 2. Technical Solution:

[0008] This invention proposes a game-theoretic optimization-based autonomous vehicle dispatching system for unsignalized intersections. This system enables multi-vehicle cooperative traffic control in a vehicle-to-everything (V2X) environment. Its core components include vehicle state perception, dispatching model construction, constraint setting, and dispatching instruction generation modules, specifically comprising the following steps:

[0009] Step 1: Obtain the dynamic status information of all autonomous vehicles approaching the intersection through the vehicle-to-everything (V2X) communication system, including key data such as the current position, speed, and predicted arrival time at the intersection for each vehicle. Let x be the longitudinal and lateral positions of vehicle i. i With y i Its speed is v i The estimated time to reach the intersection is t. ai ;

[0010] Step 2: Based on the above status information, define the departure time of each vehicle from the intersection as t. di Then the waiting time for vehicle i is expressed as:

[0011] t wi =t di -t ai (1)

[0012] If the width of the intersection is w d If a vehicle passes through an intersection at a constant speed, the waiting time can be expressed as:

[0013]

[0014] Among them, t si This represents the waiting time of vehicle i; if it does not stop, then t represents the waiting time. si =0;

[0015] Step 3: Construct a scheduling objective function to achieve a balance between efficiency and fairness. Introduce vehicle priority weights w. i Then the scheduling cost of vehicle i is:

[0016] J i =wi ·t wi (3)

[0017] This invention aims to minimize the maximum weighted waiting time of all vehicles and constructs the following scheduling model:

[0018]

[0019] Among them, t w This represents the waiting time vector for all vehicles. Let F and M be the set of feasible solutions, representing the set of vehicles in the convoy and the set of individual vehicles, respectively.

[0020] Step 4: To ensure no collisions occur while vehicles are entering the intersection, this invention incorporates a time interval safety constraint. The time difference between any two vehicles entering the intersection area should satisfy:

[0021]

[0022] This constraint prevents different vehicles from entering the conflict zone simultaneously.

[0023] Step 5: Further introduce spatial safety distance constraints to ensure that the minimum distance between any two vehicles in the intersection area is not less than a preset threshold d. min The constraint is expressed as:

[0024]

[0025] Where, d min The following geometric requirements should be met:

[0026]

[0027] Here, L and W represent the length and width of the vehicle, respectively, to ensure that physical contact can be avoided even in the worst-case scenario;

[0028] Step 6: For vehicles traveling in a convoy, the system requires them to maintain a consistent speed to avoid interference within the convoy. Let v Fi and v Fi-1 Let i and y represent the speeds of the i-th car in the convoy and the car in front of it, respectively, which must satisfy:

[0029] (v Fi -v Fi-1 ) 2 ≤ε v (8)

[0030] Where, ε v To maximize the permissible speed difference and ensure stable operation of the convoy;

[0031] Step 7: The spatial distance between vehicles in the convoy must also remain consistent. Let d be the desired distance between the i-th vehicle and the vehicle in front of it. r The actual position difference is x Fi -x Fi-1 It must meet the following requirements:

[0032] [(x Fi -x Fi-1 )-d r ] 2 ≤ε d (9)

[0033] Where, ε d This indicates the permissible spacing deviation to ensure the integrity of the formation;

[0034] Step 8: To limit overall system delays, a global waiting time constraint is introduced. The total waiting time for all vehicles must not exceed the maximum threshold T. max ,Right now:

[0035]

[0036] Where 1 is a unit vector, t w This is the waiting time vector;

[0037] Step 9: Input the above objective function and all constraints into the scheduling solution module. This module uses a solver based on optimization algorithms (such as nonlinear programming or mixed integer programming) to obtain the optimal entry time window and control strategy for each vehicle;

[0038] Step 10: The system converts the optimization results into scheduling instructions, including parameters such as passage time and speed suggestions, and sends them to each autonomous vehicle through V2V or V2I interfaces, thereby achieving safe, coordinated and efficient traffic scheduling within the intersection.

[0039] Furthermore, in step 1, the vehicle status information includes not only the vehicle's two-dimensional spatial position p i =(x i ,y i ) and velocity v i It also includes the fleet number c to which the vehicle belongs. i Status acquisition timestamp τ i and task level code κ i , where κ i This information is used to differentiate the urgency of vehicle tasks (e.g., regular, public transport, emergency vehicles). This information can be uniformly constructed into a state vector:

[0040] s i =[x i ,y i ,v i ,c i,τ i ,κ i (11)

[0041] Furthermore, in step 2, the travel time T for vehicles to cross the intersection... i Defined as the integral of the path time required for a vehicle to travel from the intersection entrance to the exit:

[0042]

[0043] Where, x entry and x exit These represent the start and end lateral positions of the intersection conflict zone, v i (t) is the velocity function of vehicle i within this interval. If we assume the vehicle travels at a constant speed v... i If passed, then there is Among them l c This represents the length of the conflict zone.

[0044] Furthermore, in step 3, the priority weight w i Besides being used for weighting the objective function, it can also be used to define the scheduling penalty function P. i As shown below:

[0045]

[0046] Among them, t wi Let γ be the waiting time for vehicle i, and let γ be the penalty coefficient set by the scheduling system, reflecting the greater scheduling cost caused by the delay for high-priority vehicles.

[0047] Furthermore, in step 4, the time conflict is represented as a safe time window matrix T. safe Elements:

[0048] T safe (i,j)=max(0,Δt min -|t ai -t aj |) (14)

[0049] Among them, t ai and t aj Let Δt be the estimated arrival time of vehicles i and j. min This is the minimum allowed time interval. If T safe (i,j)>0 indicates that there is a potential time conflict between the two vehicles.

[0050] Furthermore, in step 5, the vehicle spatial proximity matrix D(i,j) is defined as:

[0051]

[0052] Where, p i=(x i ,y i ) and p j =(x j ,y j Let be the spatial coordinates of vehicles i and j, respectively, and ε be a small positive constant to prevent division by zero. The larger this index is, the closer the two vehicles are, and it is used to assess the risk of spatial conflict.

[0053] Furthermore, in step 6, the fleet speed consistency index σ v Defined as:

[0054]

[0055] Where, N f v represents the number of vehicles in the fleet. Fi Let v be the speed of the i-th car in the convoy. F1 σ represents the speed of the lead car in the convoy. v The smaller the value, the more consistent the speed of the convoy.

[0056] Furthermore, in step 7, the vehicle spacing consistency index δ d Defined as:

[0057]

[0058] Where, x Fi Let d be the lateral position of the i-th car in the convoy. r δ represents the target spacing. d It reflects the normalized error between the actual spacing and the ideal spacing, and is used to assess the compactness of the fleet.

[0059] Furthermore, in step 8, the total waiting time upper limit constraint is expressed as a penalty term in the form of a soft constraint:

[0060]

[0061] Where n is the total number of vehicles participating in the scheduling, and T max This is the preset maximum total waiting time. This item can be incorporated into the scheduling optimization objective to avoid solution failure due to hard constraints.

[0062] Furthermore, in step 9, the min-max problem is reconstructed into the following linearly equivalent form using the auxiliary variable z:

[0063]

[0064] Here, z represents the upper bound of the maximum weighted waiting time, and t is the set of scheduling times for all vehicles. The original "minmax" problem is transformed into a linear form, which facilitates optimization. This optimization problem can be computed online on edge nodes or a centralized server using solvers such as IPOPT and Gurobi.

[0065] Furthermore, in step 10, the scheduling output instruction set CMD is executed. i Represented as:

[0066]

[0067] in, Recommend the time for vehicles to enter the intersection. To suggest a passage speed, μ i This is the control mode, with values ​​including coast (constant speed), hold (wait), or accelerate (accelerate).

[0068] 3. The game-theoretic optimization-based autonomous vehicle dispatching system for unsignalized intersections of this invention has the following advancements and advantages compared to existing technologies:

[0069] (1) By constructing an optimization model with the goal of minimizing the weighted maximum waiting time, this invention effectively solves the problems of low traffic efficiency and excessive waiting time for some vehicles at unsignalized intersections, and achieves a dynamic balance between fairness and overall efficiency in the multi-vehicle scheduling process.

[0070] (2) The present invention introduces explicit time interval constraints and spatial distance constraints in the scheduling process, which successfully solves the path conflict and collision risk problem when multiple vehicles pass through the intersection at the same time, and improves the operational safety of the entire system in complex traffic environments.

[0071] (3) By introducing a traffic priority weight mechanism, this invention effectively solves the problem that different types of vehicles (such as ambulances and buses) cannot reflect traffic priority, enabling the scheduling strategy to flexibly support differentiated services and task priority responses, and adapt to the needs of real traffic management. Attached Figure Description

[0072] Figure 1 This is a schematic diagram of a typical traffic scenario at an unsignalized intersection as described in this invention, illustrating the convergence of multiple autonomous vehicles (including convoys and individual vehicles) from different directions within a four-way intersection area, with conflict areas marked.

[0073] Figure 2 This is a system structure block diagram of the present invention, which shows the composition of the entire scheduling system, including the relationship and information flow direction of the status information acquisition module, the optimized scheduling module, the control command generation module, and the vehicle-mounted execution terminal;

[0074] Figure 3This is a flowchart of the scheduling result generation and instruction issuance of the present invention, which describes the process from scheduling solution to generating vehicle control instructions, including the processing flow of information such as vehicle recommended entry time, recommended passage speed and control mode identifier; Detailed Implementation

[0075] The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and specific examples:

[0076] Figure 2 This is a system architecture diagram of the game-theoretic optimization-based autonomous vehicle scheduling method for unsignalized intersections described in this invention. The specific implementation steps of this method include:

[0077] S1. The dispatch system obtains vehicle status information from the on-board terminal, including the vehicle's current location coordinates p. i =(x i ,y i ), velocity v i Estimated arrival time at the intersection t ai Vehicle category (e.g., regular car, bus, ambulance, etc.) and its fleet number (c) i All information is collected in real time via V2V or V2I communication modules, and a unified state vector s is constructed. i ;

[0078] S2. Based on the vehicle's current position and planned travel direction, the system establishes a scheduling path and a geometric model of the conflict zone, and delineates the spatial extent of the conflict zone. The vehicle's travel path within the intersection is assumed to be a linear trajectory, and the travel time is determined by both the length of the passage area and the vehicle speed. A preliminary calculation of the travel time T is performed. i ;

[0079] S3. Construct a scheduling optimization model with the objective function of minimizing the weighted maximum waiting time of all vehicles, and establish the following min-max optimization problem:

[0080]

[0081] Where w i Let t be the priority weight of vehicle i. di For the time it takes to leave the intersection, t ai Give its estimated arrival time.

[0082] S4. To ensure the safety and rationality of the scheduling results, the system introduces several constraints during the modeling process. First, in the time dimension, the time difference between any two vehicles entering the intersection must not be less than the minimum safe time interval, i.e.

[0083] |t ai -t aj |≥Δt min(twenty two)

[0084] This is used to avoid collisions caused by vehicles entering the conflict zone simultaneously. Spatially, it requires that the Euclidean distance between any two vehicles satisfies...

[0085] |p i -p j |2≥d min (twenty three)

[0086] Where d min The threshold is calculated based on vehicle size and safe passage distance. For vehicles forming a platoon, the system further constrains the stability of their speed and spacing, requiring the speed difference between adjacent platoon members to meet certain conditions.

[0087]

[0088] At the same time, the positional spacing should be close to the target value d. r ,Right now

[0089]

[0090] Finally, to prevent overall system congestion, the total waiting time for all vehicles is limited to a maximum of T. max ,Right now

[0091]

[0092] The above constraints work together in the scheduling solution process, ensuring the overall performance of the scheduling results in terms of security, stability and traffic efficiency.

[0093] S5. Input the above optimization objective and constraints into the scheduling solution module, and solve using a nonlinear optimization algorithm or mixed-integer programming to obtain the optimal entry time for each vehicle. Compared with the recommended travel speed

[0094] S6. Generate vehicle control commands based on the scheduling results. The format of the control commands is as follows:

[0095]

[0096] Where, μ i This indicates the control mode, and the possible values ​​are "coast" (pass at a constant speed), "hold" (wait), and "accelerate" (pass at an accelerated speed).

[0097] S7. The dispatching command is sent to the vehicle execution terminal through the V2X interface. The vehicle enters the intersection passage area according to the command, and at the same time, the execution status is uploaded in real time for use in the next cycle of dispatching, forming a dispatching closed loop.

[0098] The present invention has been described according to specific embodiments, but is not limited to the above-described embodiments. Equivalent modifications or substitutions made by those skilled in the art without departing from the scope of the claims are all within the scope of protection of the present invention, and the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A game-theoretic optimization-based automated vehicle dispatching system for unsignalized intersections, characterized in that, The system includes: a status information acquisition module, an optimized scheduling module, a control command generation module, and an on-board execution terminal; The state information acquisition module is used to acquire the dynamic state data of each autonomous vehicle in real time through communication between the on-board equipment and the roadside equipment. This data includes the current location coordinates, driving speed, estimated arrival time at the intersection, vehicle type, and priority level. This information is then organized into a unified state vector for scheduling modeling. The optimization scheduling module is used to construct a multi-vehicle conflict coordination scheduling problem based on game theory. A min-max optimization model is defined with the objective of minimizing the weighted maximum waiting time of all vehicles. The objective function is: in, Indicates vehicle The passage priority weight, Its estimated time to reach the intersection, The actual departure time is used for scheduling. This optimized scheduling module introduces multiple traffic scheduling constraints during the solution process, including ensuring that the time difference between the arrival of any two vehicles is not less than a set threshold. Time interval constraints, minimum lateral or longitudinal distance between vehicles not less than the safe distance The constraints include spatial safety distance constraints, fleet consistency constraints (ensuring the speed and spacing differences between adjacent vehicles in the same fleet do not exceed the allowable error range), and the total waiting time of all vehicles not exceeding the system's allowable upper limit. Global wait time constraints; Used to prevent any two vehicles from entering the intersection conflict zone at the same time; The control command generation module is used to convert the scheduling optimization results into a set of structured control parameters, including the recommended entry time into the intersection, the recommended speed, and the identifier of the control strategy to be adopted for each vehicle. The on-board execution terminal is used to receive and parse the control commands, adjust the vehicle speed and position in real time to ensure that it passes through the intersection smoothly according to the predetermined time sequence and path, and at the same time feeds back the execution status to the system through V2X communication to support dynamic rescheduling in the next cycle, thereby realizing system-level closed-loop scheduling control. The control parameter set generated by the control command generation module includes the following three items: in, For vehicles Recommended entry time, To recommend a passage speed, These are control mode parameters, with values ​​including "pass at a constant speed", "wait", or "accelerate through".

2. The system according to claim 1, characterized in that, The time interval constraint set by the optimized scheduling module is specifically: for any two autonomous vehicles with a risk of conflict... and The scheduling entry time satisfies: in, This is a preset minimum safe time interval used to avoid simultaneous entry into the intersection conflict zone.

3. The system according to claim 1, characterized in that, The spatial safety constraint set by the optimized scheduling module is that all vehicles must maintain a distance of no less than the minimum distance at any given time. That is, any vehicle and satisfy: in, and vehicles and vehicles Spatial coordinates, These are safety thresholds set based on vehicle size and dynamic safety requirements.

4. The system according to claim 1, characterized in that, When modeling vehicles in a convoy, the optimized scheduling module further sets speed consistency and spacing consistency constraints, requiring adjacent vehicles within the convoy to... and The following relationship must be satisfied: in, For the first The speed of the car Its position coordinates, For the desired vehicle distance, and These are the error tolerances for speed and spacing, respectively.

5. The system according to claim 1, characterized in that, The optimized scheduling module is equipped with a global waiting time constraint to limit the total waiting time of all vehicles from exceeding a system-defined threshold. in, For vehicles The estimated arrival time, Schedule its transit time. This represents the maximum allowable waiting time.

Images