Vehicle decision method for intersection with mixed connected and non-connected vehicles based on game theory

Abstract

This invention presents a vehicle decision-making method for connected mixed-traffic intersections based on game theory, belonging to the field of vehicle intelligent interaction and traffic safety technology. It utilizes vehicle-to-vehicle (V2V) communication and roadside units (RSUs) to acquire real-time information on the status of surrounding vehicles and driver characteristics, identifying potential conflict points. Based on a first-come, first-served principle, a global priority identification algorithm is executed to construct an acyclic priority sequence for all vehicles. Differentiated game models are adopted according to the interaction type: when the two players are CAV (Carrier Ability) and CMV (Carrier Traffic Vehicle), a non-cooperative game strategy maximizing the minimum value is used; when both players are CAV, a cooperative game strategy is used to maximize their total benefit. Finally, the optimal action is used to control the vehicles. This invention significantly improves intersection traffic efficiency while ensuring safety.

Description

Technical Field

[0001] This invention belongs to the field of automotive intelligent interaction and traffic safety technology, specifically relating to a vehicle decision-making method for connected mixed traffic intersections based on game theory. Background Technology

[0002] With the rapid development of autonomous driving and vehicle-to-everything (V2X) technologies, the mixed traffic environment of connected autonomous vehicles (CAVs) and connected human-driven vehicles (CMVs) has become an inevitable stage in urban traffic development. Intersections are key nodes in urban traffic networks and are also one of the main bottlenecks leading to traffic congestion and accidents. T-junctions, as typical traffic bottlenecks, are characterized by uneven directional flow and concentrated turning conflicts, resulting in more frequent spatiotemporal conflicts between vehicles. Traditional signal control or static priority allocation methods are difficult to adapt to dynamic mixed traffic environments.

[0003] Existing vehicle decision-making methods in mixed-traffic driving environments largely rely on AI-based approaches such as reinforcement learning. While these methods possess some environmental adaptability, they generally suffer from strong "black box" characteristics and poor interpretability, making it difficult to meet the transparency and safety requirements of practical deployments. Furthermore, their focus on single-vehicle optimization lacks effective global priority allocation logic, easily leading to traffic deadlocks or inefficiency. Additionally, driving style is an uncertain factor influencing interaction decisions in CMVs. However, current research has failed to fully perceive and integrate this factor, failing to achieve an effective correlation between driving style and interaction decisions.

[0004] Therefore, there is an urgent need in the existing technology for a vehicle decision-making method for unsignalized T-junctions that can effectively address the uncertainty of CMV behavior, combine decision transparency and environmental adaptability, and achieve a balance between safety and efficiency. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and to provide a vehicle decision-making method for connected mixed traffic intersections based on game theory, which aims to solve the problems mentioned in the background art.

[0006] This invention provides a vehicle decision-making method for connected mixed-traffic intersections based on game theory, comprising the following steps, which are performed sequentially: Step 1: Use vehicle-to-everything (V2X) technology and roadside units to collect real-time status information of all vehicles within the T-junction area as input information for vehicle decision-making; Step 2: The roadside unit identifies potential conflict points between vehicles in real time based on real-time vehicle status information and the geometry of the T-junction. Step 3: Based on preset principles, construct an acyclic global priority sequence for vehicles within the conflict area of ​​the T-junction; Step 4: Construct the profit function and calibrate the parameters in the profit function using an open-source dataset; Step 5: Based on the type of interaction, use different game theory models to make decisions and obtain the optimal acceleration command; Step Six: Perform the two-vehicle game sequentially according to the global priority sequence, and convert the optimal acceleration command obtained in Step Five into the actual control quantity, which is then executed through the vehicle drive-by-wire system.

[0007] The vehicle-to-everything (V2X) technology described in step one enables vehicles to exchange status information via V2V or V2I communication. The roadside unit is located at the T-junction and is used to control the acceleration of the connected autonomous vehicle for the next step. The control area is a circle centered on the center of the T-junction. The real-time vehicle status information includes dynamic input information and static input information. The dynamic input information includes lateral and longitudinal position, lateral and longitudinal velocity, and lateral and longitudinal acceleration. The static input information includes vehicle type, lane of the vehicle, driving style, intersection length, road width, and road boundary information.

[0008] The method for identifying potential conflict points between vehicles in step two is as follows: Step 2-1: If the driving paths of the two vehicles overlap, then there is a potential conflict point between the two vehicles at the current moment. Calculate the distance between the two vehicles and the potential conflict point. Step 2-2: Calculate the time T it takes for both vehicles to reach the potential conflict point at their current speed and acceleration. , In the formula, These are the vehicle's position coordinates. The coordinates of the potential conflict point. For the length of the vehicle, The speed of the vehicle; Steps 2-3: Calculate the time difference between the two vehicles and the potential conflict point. If the time difference is greater than the threshold... If both vehicles are in a safe state, they will continue driving in their current state and the process will end; if the threshold is less than 1, the process will end. If it is in a dangerous state, it will be prioritized.

[0009] The method for constructing the global priority sequence in step three is as follows: Step 3-1: Record the two vehicles in a dangerous situation as ( i , j ) Conflict pair, when vehicles i Compare j When closer to the point of conflict, the leader is identified. i and followers j， Record i > j ; Step 3-2: Define the global priority set R、Includes the set of all two-vehicle leader-follower pairs. Q、 Global Leader Set L、 Global follower set F and the temporary leadership gathering L_r ; Step 3-3: Traverse the collection Q Add the leader of each conflict pair to the global leader set. L Add the followers of each conflict pair to the global follower set. F ; Steps 3-4: From the set L Extract all vehicles not included in set F in the current iteration and add them to the temporary leader set. L_r In the middle; if set L_r If empty, it indicates a circular dependency, meaning each vehicle has the same priority to pass through. Q A leader is randomly selected from a pair of relationships to join. L_r To break the deadlock; Steps 3-5: Assemble the temporary leaders L_r Vehicles in the list are added to the global priority set in sequence. R In, and will all with L_r The relationship between the vehicle and the leader is crucial for the development of the vehicle. Q Remove from; Steps 3-3 to 3-5: Repeat steps 3-3 to 3-5 until all vehicles have been processed, and finally output the global priority set. R。

[0010] The profit function described in step four is: , In the formula: For safety and profit, For efficiency and profit, For comfort benefits, , , They are vehicles i Safety benefit weighting factor, efficiency benefit weighting factor, and comfort benefit weighting factor; The security benefits are as follows: , In the formula, It was during the period of subsequent occupation. It is the post-occupation time threshold; The efficiency gain is: , In the formula, It is the current speed of vehicle i. It is a road speed limit; The aforementioned comfort benefits are: , In the formula, It is the acceleration of vehicle i.

[0011] The method for calibrating the parameters in the profit function using an open-source dataset in step four is as follows: Step 4-1: Using the real dataset of the inD intersection drone dataset, three human driving styles are obtained through the k-Shape temporal clustering method: conservative, general, and aggressive. Step 4-2: Set the weight parameters to be calibrated Chromosomal encoding is performed to form a genetic individual; Step 4-3: Randomly generate 60 sets of parameters as the initial population, serving as the starting point for iterative optimization; Step 4-4: Based on the real dataset of the inD intersection, run the revenue function to perform dynamic simulation; Steps 4-5: Calculate the root mean square error (RMSE) between the simulated vehicle speed and the actual trajectory vehicle speed. , in, To simulate vehicle speed, This refers to the actual vehicle speed along the trajectory. Step 4-6: Stop iterating if any of the following conditions are met, and proceed to step 4-7: ①The fitness value reaches the threshold: ≤ 1( m / s ), ② Reaching the maximum number of iterations: 300 generations; Step 4-7: If the current step is not skipped, perform selection, crossover, and mutation operations on the current population to generate a new generation of population, and repeat step 4-3. Steps 4-8: Stop the iteration, output the optimal parameter combination, and complete the parameter calibration of the function.

[0012] The decision variable in the game model described in step five is the discrete change of acceleration from its minimum to its maximum value.

[0013] The interaction types described in step five include interactions between connected autonomous vehicles (CAVs) and connected human-driven vehicles (CMVs), and interactions between connected autonomous vehicles (CAVs) themselves.

[0014] The interaction between the connected autonomous vehicle (CAV) and the connected human-driven vehicle (CMV) adopts a non-cooperative game theory approach. While considering the CMV's reaction strategy, the lower bound of the CAV's payoff is maximized, calculated by the following formula: , , The constraints are as follows: , In the formula, vehicle i The revenue is expressed as Its acceleration is denoted as Its current speed is ;vehicle i The optimal strategy is to use This indicates that other vehicles j In known vehicles i The optimal set of policies under policy conditions is denoted as ,vehicle j The range of selectable strategies is as follows The minimum and maximum permissible accelerations of the vehicle are respectively and ,vehicle i The desired headway and desired maximum speed are expressed as follows: and ,vehicle i Acquired vehicles j The expected headway and maximum speed are respectively and ; The interaction between the connected autonomous vehicles (CAVs) adopts a cooperative game theory approach. The core constraint is that the combined benefit achieved by any two vehicles through cooperation must not be less than the minimum guaranteed benefit each vehicle can obtain if negotiations break down and cooperation fails. This is calculated using the following formula: , The constraints are as follows: , , In the formula, and Let i and j be the respective payoffs of each car in a non-cooperative game. Let i be the optimal acceleration of each vehicle in the cooperative game between vehicles i and j.

[0015] Step six describes the vehicle drive-by-wire system controlling the acceleration of the connected autonomous vehicle (CAV). Based on the vehicle's kinematic model, it updates the vehicle's speed and position in real time to ensure smooth decision execution.

[0016] The beneficial effects of this invention are as follows: This invention adopts game theory as the core decision-making framework, which has clear mathematical logic and transparent decision-making rules, meeting the requirements for security and verifiability in actual deployment. In addition, it combines the global priority identification algorithm with decision-making, fundamentally avoiding deadlock and collision in multi-vehicle interaction. By utilizing differentiated game theory and a payoff function that reflects driving style, it realizes the dynamic interaction of connected autonomous vehicles (CAVs) in real mixed traffic environments, ultimately significantly improving the traffic efficiency of intersections while ensuring safety. Attached Figure Description

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0018] Figure 1 This is a schematic diagram of the vehicle decision-making method for connected mixed traffic intersections based on game theory according to the present invention.

[0019] Figure 2 This is a schematic diagram of a signalless T-junction environment under a networked mixed traffic environment, as described in a specific embodiment of the present invention.

[0020] Figure 3 This is a schematic diagram of the parameter calibration process of the present invention.

[0021] Figure 4 This is a schematic diagram of the parameter calibration results of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] This invention presents a vehicle decision-making method for connected mixed-traffic intersections based on game theory. It utilizes vehicle-to-vehicle (V2V) communication and roadside units (RSUs) to acquire real-time information on the status of surrounding vehicles and driver characteristics, identifying potential conflict points. Based on a first-come, first-served principle, a global priority identification algorithm is executed to construct an acyclic priority sequence for all vehicles. Differentiated game models are employed according to the interaction type: when the two players are CAV and CMV, a non-cooperative game strategy maximizing the minimum value is used; when both players are CAV, a cooperative game strategy is used to maximize their total benefit. Finally, the vehicle is controlled based on the optimal action obtained from the solution.

[0024] like Figure 1 As shown, the vehicle decision-making method for connected mixed-traffic intersections based on game theory includes the following steps: S1: Connected information collection, using vehicle-to-everything (V2X) technology and roadside units (RSUs) to collect real-time status information of all vehicles within the T-junction area, as input information for vehicle decision-making.

[0025] Specifically, vehicle-to-everything (V2X) technology refers to the exchange of status information between vehicles via V2V or V2I communication; the roadside unit (RSU) is located near the T-junction, and the RSU controls the acceleration of the CAV (Car Access Vehicle). The CAV control area is a circle centered at the center of the intersection, with a radius of... r =30m.

[0026] Specifically, the real-time status information of all vehicles includes: collecting the vehicle's current lateral and longitudinal position, lateral and longitudinal speed, and lateral and longitudinal acceleration as dynamic input information; and collecting vehicle type, lane to which the vehicle belongs, driving style, intersection length, road width, and road boundary information as static input information.

[0027] Step S2: Conflict detection. Based on the state information and intersection geometry, the RSU identifies potential conflict points between vehicles in real time.

[0028] Specifically, the preset paths (such as straight ahead and left turn) of all vehicles and the intersection layout are acquired in real time in a connected environment. The RSU identifies the paths and potential conflict points of all CAVs and CMVs within the area, such as intersection scenarios. Figure 2 As shown.

[0029] Specifically, the steps for identifying potential conflict points between vehicles are as follows: Step S2.1: If the driving paths of the two vehicles overlap, there is a potential conflict point between the two vehicles at the current moment. Calculate the distance between the two vehicles and the potential conflict point. Step S2.2: Calculate the time it takes for both vehicles to reach the potential conflict point at their current speed and acceleration; The arrival time of the aforementioned vehicles at the conflict point is calculated using the following formula:

[0030] In the formula, These are the vehicle's position coordinates. The coordinates of the potential conflict point. For the length of the vehicle, The speed of the vehicle.

[0031] Step S2.3: Calculate the time difference between the two vehicles and the potential conflict point. If the time difference is greater than the threshold... If both vehicles are in a safe state, they will continue driving in their current state and the process will end; if the threshold is less than 1, the process will end. If so, the situation is dangerous and step S3 continues.

[0032] Step S3: Global priority identification. Based on the "first-come, first-served" principle, construct an acyclic global priority sequence for all vehicles in the intersection conflict area.

[0033] Specifically, global priority identification determines a global passage sequence for all vehicles identified as being in a dangerous state in step S2. The specific determination steps are as follows: Step S3.1: Define the two vehicles in a dangerous state in step S2 above as a (i, j) conflict pair. When vehicle i is closer to the conflict point than vehicle j, it is determined to be the leader i and the follower j (denoted as i>j). Step S3.2: Initialize the global priority empty set R and the set Q containing all two-vehicle leader-follower relationship pairs; Step S3.3: Traverse set Q, add the leader of each conflict pair to the global leader set L, and add the follower of each conflict pair to the global follower set F; Step S3.4: Extract all vehicles from set L that are not included in set F in the current iteration, and use them as a temporary leader set r; if set r is empty, it means that there is a circular dependency, that is, each vehicle has the same passing priority, then randomly select a relation pair from Q and add it to r to break the deadlock. Step S3.5: Add the vehicles in the temporary leader set r to the global priority set R in sequence, and remove all relationship pairs with vehicles in r as leaders from Q; Step S3.6: Repeat steps S3.3 to 3.5 until all vehicles have been processed, and finally output the global priority set R.

[0034] Step S4: Establish a benefit function, design a benefit function that includes safety, efficiency and comfort, and use an open source dataset to calibrate the parameters in the function.

[0035] Specifically, the benefit function consists of safety, efficiency, and comfort, and is calculated using the following formula:

[0036] In the formula, For safety and profit, For efficiency and profit, For comfort benefits, , , They are vehicles i Safety benefit weighting factor, efficiency benefit weighting factor, and comfort benefit weighting factor; The safety benefit is calculated using the following formula:

[0037] In the formula, It was during the period of subsequent occupation. It is the post-occupation time threshold; The efficiency gains are calculated using the following formula:

[0038] In the formula, It is the current speed of vehicle i. It is a road speed limit; The comfort gains are calculated using the following formula:

[0039] In the formula, It is the acceleration of vehicle i.

[0040] Specifically, the calibration flowchart for using open-source datasets to calibrate the parameters in a function is as follows: Figure 3 As shown: Step S4.1: Using the inD intersection real dataset, three CMV driving styles are obtained through the k-Shape temporal clustering method: conservative, average, and aggressive. Step S4.2: Set the weight parameters to be calibrated Chromosome encoding is performed to create gene individuals that can be processed by the algorithm; Step S4.3: Randomly generate 60 sets of parameters as the initial population, serving as the starting point for iterative optimization; Step S4.4: Based on the real dataset of the inD intersection, run the above game function to perform dynamic simulation; Step S4.5: Calculate the root mean square error (RMSE) between the simulated vehicle speed and the actual trajectory vehicle speed. The specific calculation formula is as follows:

[0041] in, To simulate vehicle speed, This refers to the actual vehicle speed along the trajectory. Step S4.6: If any of the following conditions are met, stop the iteration and skip step S4.6 to proceed to step S4.7: ①The fitness value reaches the threshold: ≤ 1( m / s ), ② Reaching the maximum number of iterations: 300 generations; Step S4.7: If the current step is not skipped, perform selection, crossover, and mutation operations on the current population to generate a new generation of population, and then repeat step S4.3. Step S4.8: Stop the iteration and output the optimal combination of parameters, which completes the parameter calibration of the function.

[0042] Specifically, the parameter calibration results are as follows: Figure 4 As shown, the safety benefit weight factor, efficiency benefit weight factor, and comfort benefit weight factor corresponding to the three driving styles were obtained. Among them, the aggressive driver values ​​efficiency more, with an efficiency weight as high as 0.5, which is the highest among the three. The conservative driver values ​​safety more, with a safety weight of 0.62, which is much higher than the other styles. The average driver seeks a balance between safety and efficiency, with weights of 0.45 and 0.40, respectively.

[0043] Step S5: Game-theoretic interaction decision-making. Different game models are used to make decisions based on the type of interaction.

[0044] Specifically, the decision variable in the game is acceleration, which is the discrete change of acceleration from its minimum to its maximum value.

[0045] Game theory decision-making includes the interaction between CAV and CMV, and the interaction between CAVs, as follows: The interaction between CAV and CMV employs a non-cooperative game theory approach. While considering the CMV's response strategy, the lower bound of CAV's payoff is maximized, calculated using the following formula:

[0046]

[0047] The constraints are as follows:

[0048] In the formula, vehicle i The revenue is expressed as Its acceleration is denoted as Its current speed is .vehicle i The optimal strategy is to use This indicates that other vehicles j In known vehicles i The optimal set of policies under policy conditions is denoted as ,vehicle j The range of selectable strategies is as follows The minimum and maximum permissible accelerations of the vehicle are respectively... and ,vehicle i The desired headway and desired maximum speed are expressed as follows: and ,vehicle i Acquired vehicles j The expected headway and maximum speed are respectively and ; The interaction between CAVs employs cooperative game theory to maximize the interests of both parties. The core constraint is that the combined benefit achieved by any two vehicles through cooperation must not be less than the minimum guaranteed benefit each party can obtain if negotiations break down and cooperation fails (i.e., the benefit at the point of cooperation breakdown). This is calculated using the following formula:

[0049] The constraints are as follows:

[0050]

[0051] In the formula, and Let i and j be the respective payoffs of the two cars in a non-cooperative game.

[0052] Step S6: Vehicle control execution. The two vehicles engage in a game of strategy according to the global priority sequence. The optimal acceleration command obtained in step S5 is converted into an actual control quantity and executed through the vehicle drive-by-wire system to achieve safe and efficient passage.

[0053] Specifically, for any vehicle A, it becomes either a leader or a follower in different conflicts in the global priority sequence, and the control decision that prioritizes becoming a follower ensures safe and orderly passage. Specifically, control execution involves controlling the acceleration of the CAV. This control process is based on the vehicle's kinematic model and updates the vehicle's speed and position in real time to ensure smooth execution of decisions.

[0054] In summary, the game theory-based vehicle decision-making method for connected mixed-traffic intersections proposed in this invention uses vehicle-to-everything (V2X) technology and roadside units (RSUs) to perceive the environment in real time, establish a payoff function that balances safety and efficiency, construct a differentiated game model, and achieve safe and efficient operation of unsignalized T-junctions.

[0055] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0056] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A vehicle decision-making method for connected mixed-traffic intersections based on game theory, characterized by: Includes the following steps, And the following steps are performed in sequence: Step 1: Use vehicle-to-everything (V2X) technology and roadside units to collect real-time status information of all vehicles within the T-junction area as input information for vehicle decision-making; Step 2: The roadside unit identifies potential conflict points between vehicles in real time based on real-time vehicle status information and the geometry of the T-junction. Step 3: Based on preset principles, construct an acyclic global priority sequence for vehicles within the conflict area of ​​the T-junction; Step 4: Construct the profit function and calibrate the parameters in the profit function using an open-source dataset; Step 5: Based on the type of interaction, use different game theory models to make decisions and obtain the optimal acceleration command; Step Six: Perform the two-vehicle game sequentially according to the global priority sequence, and convert the optimal acceleration command obtained in Step Five into the actual control quantity, which is then executed through the vehicle drive-by-wire system.

2. The vehicle decision-making method for connected mixed-traffic intersections based on game theory according to claim 1, characterized in that: The vehicle-to-everything (V2X) technology described in step one enables vehicles to exchange status information via V2V or V2I communication. The roadside unit is located at the T-junction and is used to control the acceleration of the connected autonomous vehicle for the next step. The control area is a circle centered on the center of the T-junction. The real-time vehicle status information includes dynamic input information and static input information. The dynamic input information includes lateral and longitudinal position, lateral and longitudinal velocity, and lateral and longitudinal acceleration. The static input information includes vehicle type, lane of the vehicle, driving style, intersection length, road width, and road boundary information.

3. The vehicle decision-making method for connected mixed-traffic intersections based on game theory according to claim 1, characterized in that: The method for identifying potential conflict points between vehicles in step two is as follows: Step 2-1: If the driving paths of the two vehicles overlap, then there is a potential conflict point between the two vehicles at the current moment. Calculate the distance between the two vehicles and the potential conflict point. Step 2-2: Calculate the time T it takes for both vehicles to reach the potential conflict point at their current speed and acceleration. ， In the formula, These are the vehicle's position coordinates. The coordinates of the potential conflict point. For the length of the vehicle, The speed of the vehicle; Steps 2-3: Calculate the time difference between the two vehicles and the potential conflict point. If the time difference is greater than the threshold... If both vehicles are in a safe state, they will continue driving in their current state until the end. If less than the threshold If it is in a dangerous state, it will be prioritized.

4. The vehicle decision-making method for connected mixed-traffic intersections based on game theory according to claim 1, characterized in that: The method for constructing the global priority sequence in step three is as follows: Step 3-1: Record the two vehicles in a dangerous situation as ( i , j ) Conflict pair, when vehicles i Compare j When closer to the point of conflict, the leader is identified. i and followers j， Record i > j ; Step 3-2: Define the global priority set R、 Includes the set of all two-vehicle leader-follower pairs. Q、 Global Leader Set L、 Global follower set F and the temporary leadership gathering L_r ; Step 3-3: Traverse the collection Q Add the leader of each conflict pair to the global leader set. L Add the followers of each conflict pair to the global follower set. F ; Steps 3-4: From the set L Extract all vehicles not included in set F in the current iteration and add them to the temporary leader set. L_r In the middle; if set L_r If empty, it indicates a circular dependency, meaning each vehicle has the same priority to pass through. Q A leader is randomly selected from a pair of relationships to join. L_r To break the deadlock; Steps 3-5: Assemble the temporary leaders L_r Vehicles in the list are added to the global priority set in sequence. R In, and will all with L_r The relationship between the vehicle and the leader is crucial for the development of the vehicle. Q Remove from; Steps 3-3 to 3-5: Repeat steps 3-3 to 3-5 until all vehicles have been processed, and finally output the global priority set. R。 5. The vehicle decision-making method for connected mixed-traffic intersections based on game theory according to claim 1, characterized in that: The profit function described in step four is: ， In the formula: For safety and profit, For efficiency and profit, For comfort benefits, , , They are vehicles i Safety benefit weighting factor, efficiency benefit weighting factor, and comfort benefit weighting factor; The security benefits are as follows: ， In the formula, It was during the period of subsequent occupation. It is the post-occupation time threshold; The efficiency gain is: ， In the formula, It is the current speed of vehicle i. It is a road speed limit; The aforementioned comfort benefits are: ， In the formula, It is the acceleration of vehicle i.

6. The vehicle decision-making method for connected mixed-traffic intersections based on game theory according to claim 1, characterized in that: The method for calibrating the parameters in the profit function using an open-source dataset in step four is as follows: Step 4-1: Using the real dataset of the inD intersection drone dataset, three human driving styles are obtained through the k-Shape temporal clustering method: conservative, general, and aggressive. Step 4-2: Set the weight parameters to be calibrated Chromosomal encoding is performed to form a genetic individual; Step 4-3: Randomly generate 60 sets of parameters as the initial population, serving as the starting point for iterative optimization; Step 4-4: Based on the real dataset of the inD intersection, run the revenue function to perform dynamic simulation; Steps 4-5: Calculate the root mean square error (RMSE) between the simulated vehicle speed and the actual trajectory vehicle speed. ， in, To simulate vehicle speed, This refers to the actual vehicle speed along the trajectory. Step 4-6: Stop iterating if any of the following conditions are met, and proceed to step 4-7: ①The fitness value reaches the threshold: ≤ 1( m / s ), ② Reaching the maximum number of iterations: 300 generations; Step 4-7: If the current step is not skipped, perform selection, crossover, and mutation operations on the current population to generate a new generation of population, and repeat step 4-3. Steps 4-8: Stop the iteration, output the optimal parameter combination, and complete the parameter calibration of the function.

7. The vehicle decision-making method for connected mixed-traffic intersections based on game theory according to claim 1, characterized in that: The decision variable in the game model described in step five is the discrete change of acceleration from its minimum to its maximum value.

8. The vehicle decision-making method for connected mixed-traffic intersections based on game theory according to claim 1, characterized in that: The interaction types described in step five include interactions between connected autonomous vehicles (CAVs) and connected human-driven vehicles (CMVs), and interactions between connected autonomous vehicles (CAVs) themselves.

9. The vehicle decision-making method for connected mixed-traffic intersections based on game theory according to claim 8, characterized in that: The interaction between the connected autonomous vehicle (CAV) and the connected human-driven vehicle (CMV) adopts a non-cooperative game theory approach. While considering the CMV's reaction strategy, the lower bound of the CAV's payoff is maximized, calculated by the following formula: ， ， The constraints are as follows: ， In the formula, vehicle i The revenue is expressed as Its acceleration is denoted as Its current speed is ;vehicle i The optimal strategy is to use This indicates that other vehicles j In known vehicles i The optimal set of policies under policy conditions is denoted as ,vehicle j The range of selectable strategies is as follows The minimum and maximum permissible accelerations of the vehicle are respectively and ,vehicle i The desired headway and desired maximum speed are expressed as follows: and ,vehicle i Acquired vehicles j The expected headway and maximum speed are respectively and ; The interaction between the connected autonomous vehicles (CAVs) adopts a cooperative game theory approach. The core constraint is that the combined benefit achieved by any two vehicles through cooperation must not be less than the minimum guaranteed benefit each vehicle can obtain if negotiations break down and cooperation fails. This is calculated using the following formula: ， The constraints are as follows: ， ， In the formula, and Let i and j be the respective payoffs of each car in a non-cooperative game. Let i be the optimal acceleration of each vehicle in the cooperative game between vehicles i and j.

10. The vehicle decision-making method for connected mixed-traffic intersections based on game theory according to claim 1, characterized in that: Step six describes the vehicle drive-by-wire system controlling the acceleration of the connected autonomous vehicle (CAV). Based on the vehicle's kinematic model, it updates the vehicle's speed and position in real time to ensure smooth decision execution.

Images