A cooperative traffic method for unsignalized intersections based on IVCPS under mixed passenger and freight traffic

Abstract

The present invention relates to a method for cooperative passage at an unsignalized intersection based on IVCPS under mixed passenger and freight traffic, and belongs to the field of intelligent connected vehicles. The method comprises: judging the vehicle density of the unsignalized intersection in the information space. When the vehicle density is low, a real vehicle model is constructed based on a rectangular model, the unsignaled intersection is modeled by a three-vehicle model, and the vehicle trajectory is modeled at the same time, and the trajectory model is solved to obtain the vehicle passage order and speed planning; when the vehicle density is high, the vehicles in the unsignaled intersection area are platooned and cooperatively controlled by a queue-based two-layer control planning method. Among them, in the two-layer control planning method, the upper layer plans the passage order of the fleet by improving the MCTS algorithm, and the lower layer plans the vehicle trajectory by the MPC algorithm. The present invention can adaptively adjust the control scheme according to the vehicle density, optimize the passage efficiency of the unsignaled intersection, and reduce fuel consumption.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent connected vehicles and relates to an IVCPS-based collaborative traffic method for an unsignaled intersection under mixed passenger and freight traffic. Background Art

[0002] Unsignalized intersections are a crucial component of the road traffic system. Vehicles from different directions converge at these intersections, creating significant conflicts and interference. Vehicles must make complex decisions when navigating unsignalized intersections, leading to frequent accidents and reduced efficiency. Frequent vehicle starts and stops increase energy consumption. Urban unsignalized intersections near ports, industrial parks, and construction sites often contain both buses and trucks. Because buses and trucks differ significantly from cars in size, acceleration and deceleration performance, speed, and energy consumption, significant interference between these vehicles can lead to mobility bottlenecks. This can further increase accidents, reduce efficiency, and increase energy consumption. Therefore, coordinated traffic control schemes should be developed for mixed passenger and freight traffic at unsignalized intersections to improve safety, efficiency, and reduce fuel consumption.

[0003] In recent years, with the rapid development of computer, communication, and automation technologies, CPS and vehicle-road-cloud integration have been applied to the intelligent vehicle sector, resulting in the unique IVCPS intelligent vehicle cyber-physical system (ICPS) technology. By utilizing IVCPS technology, information acquisition, processing, and transmission between the vehicle, road, and cloud can be mapped one-to-one between the physical and cyber spaces of unsignalized intersections. Vehicle information such as physical dimensions, speed, and acceleration, traffic status information such as volume and congestion, and spatiotemporal information such as the timestamp of a vehicle's arrival at a specific moment and its location can be obtained. Within the framework of the IVCPS intelligent vehicle cyber-physical system, a collaborative control method is designed for mixed passenger and freight unsignalized intersections, which is expected to improve safety, address low traffic efficiency, and increase vehicle fuel consumption at unsignalized intersections.

[0004] The existing cooperative traffic method for unsignalized intersections has the following shortcomings:

[0005] 1. Existing research on coordinated traffic flow at unsignalized intersections focuses on a single spatiotemporal scale, a single vehicle type, and a single traffic density. It fails to fully leverage IVCPS technology to establish a framework that can span multiple scales, accommodate different vehicle types, and adaptively adjust control methods based on vehicle density. Consequently, it fails to fully optimize unsignalized intersection performance.

[0006] 2. Existing research on cooperative traffic flow at unsignalized intersections focuses on designing a method that is compatible with different traffic flows. However, when vehicle density is low, these methods may waste a large amount of time and space resources and fail to consider the control decision-making issues brought about by the differences between different vehicle types.

[0007] 3. When the vehicle density at unsignalized intersections is high, existing research institutes have proposed traditional collaborative traffic methods based on, for example, centralized control. Traditional methods struggle to combine information space and physical space at a single scale to form a hierarchical, well-structured control approach. Furthermore, these methods suffer from high computational complexity, difficulty solving problems, a failure to balance traffic efficiency and fuel consumption, and difficulty adapting to mixed passenger and freight traffic. Summary of the Invention

[0008] In view of this, the purpose of the present invention is to provide a cooperative traffic method for unsignalized intersections based on IVCPS under mixed passenger and freight traffic, so that the control system can adaptively adjust the control scheme according to vehicle density under dynamically changing traffic conditions and vehicle types, thereby optimizing the traffic efficiency of unsignalized intersections and reducing fuel consumption.

[0009] In order to achieve the above object, the present invention provides the following technical solutions:

[0010] A collaborative traffic method for unsignalized intersections based on IVCPS under mixed passenger and freight traffic conditions is proposed. This method transmits vehicle-side and road section information to an information space through on-board sensors and roadside units. The vehicle density at the unsignalized intersection is judged in the information space, and different control strategies are adopted according to different vehicle densities to achieve collaborative traffic at the unsignalized intersection.

[0011] When vehicle density is low, the unsignalized intersection is segmented, and a real vehicle model is constructed based on a rectangular model. The unsignalized intersection is modeled using a three-vehicle model. Simultaneously, vehicle trajectories are modeled, and the trajectory model is solved to obtain the vehicle passage order and speed planning, enabling vehicles to coordinate through the unsignalized intersection.

[0012] When the vehicle density is high, the unsignalized intersection is divided into sections, and the vehicles within the unsignalized intersection area are formed into platoons. A queue-based two-layer control planning method is used for collaborative control, allowing the vehicles to pass through the unsignalized intersection in a coordinated manner. In this two-layer control planning method, the upper layer plans the platoon's passage sequence using an improved MCTS algorithm, and the lower layer plans the vehicle trajectories using an MPC algorithm.

[0013] Furthermore, at low vehicle density, the unsignalized intersection is divided into an observation section, an uncontrolled section, a target optimization section, and a conflict zone. In the uncontrolled section, vehicles are not controlled and travel at a constant speed until they reach the target optimization section. In the target optimization section and the conflict zone, vehicles are controlled until they leave the conflict zone.

[0014] Based on the segment division, the unsignalized intersection is modeled using a three-car model, including:

[0015] With the center of the unsignalized intersection as the origin O, an XOY coordinate system is established, and the direction of two conflicting areas in the direction of movement of a single vehicle is defined as the Y axis; vehicles from three directions enter the target optimization section at the same time, and the lane width of the unsignalized intersection is defined as W, and the vehicle width is w i,j , the vehicle length is l i,j , the vehicle speed and acceleration are v i,j (t) and a i,j (t), where i represents the lane, j represents the vehicle in the lane, and t represents time.

[0016] The conflict area between the first vehicle and the second vehicle is defined as the first conflict area, and the conflict area between the second vehicle and the third vehicle is defined as the second conflict area; i1, i2, and i3 represent the lanes where the first vehicle, the second vehicle, and the third vehicle are located, respectively; j1, j2, and j3 represent the vehicle numbers of the first vehicle, the second vehicle, and the third vehicle in the lanes, respectively.

[0017] Define the initial time and initial position of the trajectory planned by the cloud for the first vehicle as t0 and Define the end time and end position of the first vehicle trajectory planning as and represents the time when the front of the first vehicle arrives at the first conflict area; the initial time and initial position of the trajectory planning of the second vehicle in the cloud are defined as t0 and Define the end time and end position of the second vehicle trajectory planning as follows: and represents the time when the front of the second vehicle arrives at the first conflict area; the initial time and initial position of the third vehicle trajectory planning in the cloud are defined as t0 and The end time and end position of the third vehicle trajectory planning are defined as follows: and Indicates the time when the front end of the third vehicle arrives at the second conflict area.

[0018] Furthermore, a fifth-order polynomial is used as the vehicle trajectory model, which is expressed as:

[0019] s i,j (t) = ri,j +r i,j,1 t i,j +r i,j,2 t i,j 2 +r i,j,3 t i,j 3 +r i,j,4 t i,j 4 +r i,j,5 t i,j 5

[0020] Where r i,j 、r i,j,1 、r i,j,2 、r i,j,3 、r i,j,4 、r i,j,5 All represent coefficients, t i,j represents the travel time of vehicle j in lane i; the first-order and second-order differentials of the above equation are obtained to obtain the functions of velocity and acceleration with respect to time:

[0021] v i,j (t) = r i,j,1 +2r i,j,2 t i,j +3r i,j,3 t i,j 2 +4r i,j,4 t i,j 3 +5r i,j,5 t i,j 4

[0022] a i,j (t) = 2r i,j,2 +6r i,j,3 t i,j +12r i,j,4 t i,j 2 +20r i,j,5 t i,j 3

[0023] The position, velocity, and acceleration can be converted into time-dependent equations using a quintic polynomial. The trajectory of each vehicle is obtained by solving the optimal end time for each vehicle, where the end time is the moment when each vehicle arrives at the conflict area.

[0024] The trajectory planning problem is transformed into a multi-objective optimization problem. The designed objective functions include the shortest travel time objective function, the fuel consumption objective function, and the comfort objective function. At the same time, constraints are constructed. Among them, the travel time objective function is optimized to minimize the end time. Multiple objective functions are converted into a single objective function through weighted summation, and then the simulated annealing algorithm is used to solve the single objective function to obtain the optimal end time.

[0025] Furthermore, when the vehicle density is high, the unsignalized intersection is divided into an observation section, a formation section, a decision section and a conflict area; vehicles are arranged into queues in the formation section; and collaborative control is performed in the decision section using the two-layer control planning method.

[0026] In the platoon section, arranging vehicles into a platoon includes: setting the distance between vehicle j and vehicle j+1 to be D S , the maximum vehicle distance between vehicle j and vehicle j+1 is D H , when the following relationship is satisfied between vehicle j and vehicle j+1, vehicles j and vehicle j+1 are grouped into the same fleet:

[0027]

[0028] Where, v i,j (t0) represents the speed of the vehicle at time t0, v i,j,min Indicates the minimum speed of the vehicle, L B Indicates the length of the observation segment.

[0029] Furthermore, in the two-layer control planning method, the upper layer plans the passage order of the fleet through an improved MCTS algorithm, including: in the MCTS algorithm, each node of the search tree represents a state, and each state contains the position and speed information of all vehicles at the current moment; the MCTS algorithm repeatedly executes the four stages of selection, expansion, simulation and backtracking, and stops after reaching the predetermined calculation time or number of iterations.

[0030] In the selection phase, starting from the root node, an optimal child node is selected for expansion through the UCT algorithm;

[0031] In the expansion phase, starting from the selected optimal child node, a new vehicle passage sequence is simulated, a new child node is generated and added to the search tree;

[0032] In the simulation phase, for the expanded child nodes, starting from the root node, the process of vehicles passing through the unsignalized intersection is simulated by randomly selecting the passage order of vehicles, and the simulation process is continued downward along a path of the search tree passing through the expanded child nodes until all vehicles pass through the unsignalized intersection;

[0033] In the backtracking phase, the simulation result is traced back to the root node through the path selected in the simulation phase, and the values ​​and access times of the passed nodes are updated at the same time.

[0034] The UCT algorithm is used to evaluate the value of each child node in the selection phase, which is expressed as:

[0035]

[0036] Where, β u represents the total reward value of node u, p u represents the number of times node u is visited, P represents the total number of times the parent node is visited, and C represents the trade-off coefficient used to adjust the exploration part and the utilization part, where the utilization part is The exploration section is The average delay of the vehicle is selected as β u Calculation of value:

[0037]

[0038] Where, t i,j,ac represents the actual travel time of a single vehicle through an unsignalized intersection, t i,j,min Indicates the minimum travel time of a single vehicle.

[0039] Furthermore, the UCT algorithm is improved by using a dynamic window method to adjust the C value and defining the sliding window size W m , adjustment factor k m , number of iterations n, initial C value C init and a positive number η;

[0040] Calculate the average delay time AD within the sliding window W :

[0041]

[0042] Calculate the overall average delay time AD T :

[0043]

[0044] Then the C value is adjusted by the following formula:

[0045]

[0046] C=max(C,η)

[0047] Furthermore, the two-layer control planning method, in which the lower layer plans the vehicle trajectory through the MPC algorithm, includes:

[0048] 1) Define the vehicle's state variables, including position, velocity, and acceleration;

[0049] 2) Define the vehicle's dynamic model in discrete form:

[0050]

[0051] v i,j (t+1)=v i,j (t)+a i,j (t)αT

[0052] Where s i,j (t) The position of vehicle j in lane i at time t, v i,j (t) represents the speed of vehicle j in lane i at time t, αT represents the simulation step size, and a i,j (t) represents the acceleration of vehicle j in lane i at time t;

[0053] 3) Define optimization objectives and constraints; the optimization objectives include vehicle fuel consumption and comfort, and the constraints include vehicle speed, acceleration constraints, and vehicle spacing constraints;

[0054] 4) Based on the current state and the predicted control input sequence, the future state of the vehicle within multiple time steps is calculated to generate a series of control input sequences, namely candidate trajectories, and the state sequence under the generated control input sequence is predicted in the vehicle dynamics model; each candidate trajectory is evaluated, the corresponding objective function value is calculated, and all constraints are checked for satisfaction; and the trajectory with the minimum objective function value and satisfying all constraints is selected from the candidate trajectories as the optimal trajectory at the current moment; wherein the control input sequence includes the state sequence and the predicted sequence;

[0055] 5) Apply the state sequence of the optimal trajectory to the vehicle;

[0056] 6) Repeat the following steps in each control cycle:

[0057] Update Status: Get the latest status of the vehicle.

[0058] Reforecast: Based on the updated state, re-forecast the future state.

[0059] Re-optimization: Resolve the optimization problem to obtain a new optimal control input sequence.

[0060] The beneficial effects of the present invention are:

[0061] 1) The present invention leverages IVCPS technology to calculate the current intersection vehicle density in real time, while simultaneously identifying different vehicle types (including those with varying sizes and weights). It can span multiple scales, accommodate different vehicle types, and adaptively adjust control methods and sensing control ranges based on vehicle density.

[0062] 2) Considering that the actual collision zone of different vehicle types varies with their size when passing through an intersection, the present invention constructs a realistic vehicle model based on a rectangular model when vehicle density is low, reducing the control system's perception and control range and improving control accuracy. This allows the vehicle's controllable area to be expanded from the stop line of the current lane to the conflict zone. Furthermore, the fuel emission model is improved based on the differences between different vehicle types, reducing fuel consumption while ensuring high traffic efficiency.

[0063] 3) When vehicle density is high, this paper proposes a platoon-based two-level planning control method for unsignalized intersections. In the upper-level planning, an improved MCTS algorithm is used to improve both algorithm solution efficiency and vehicle flow efficiency. In the lower-level planning, a distributed control method is employed to optimize the fleet's speed and acceleration using fuel consumption as a metric. This method achieves a balance between high flow efficiency and low fuel consumption while maintaining low computational complexity.

[0064] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below with reference to the accompanying drawings, in which:

[0066] Figure 1 Schematic diagram of the division of different time and space scales under IVCPS;

[0067] Figure 2 Schematic diagram of the cooperative control framework for unsignalized intersections under IVCPS;

[0068] Figure 3 Schematic diagram of the cooperative traffic framework under low vehicle density;

[0069] Figure 4 It is a schematic diagram of the three-car model and the vehicle model;

[0070] Figure 5 Schematic diagram of the cooperative traffic framework under high vehicle density;

[0071] Figure 6 It is a structural diagram of the upper and lower level control planning method;

[0072] Figure 7 Schematic diagram of the MCTS algorithm search steps. DETAILED DESCRIPTION

[0073] The following describes the embodiments of the present invention by means of specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0074] Based on the multi-spatiotemporal scales of IVCPS technology, this paper proposes a multi-spatiotemporal scale collaborative traffic control system for unsignaled intersections. This system divides unsignaled intersections into multiple zones and transforms the system-level unsignaled vehicle collaborative task into multiple subsystem-level control tasks with a hierarchical progression. Relying on IVCPS technology, it is possible to calculate the current intersection vehicle density in real time and simultaneously identify different vehicle types (with inconsistent vehicle sizes and weights). Finally, a framework system is established that can span multiple scales, be compatible with different vehicle types, adaptively adjust control methods based on vehicle density, and perceive the control range.

[0075] Furthermore, considering that the actual collision zone of different vehicle types passing through an intersection varies with their size, a low-vehicle-density cooperative traffic framework is designed when the vehicle density is low. A real vehicle model is constructed based on a rectangular model. To reduce the control system's perception and control range and improve control accuracy, a cooperative traffic method based on a three-vehicle model is proposed. Taking into account the actual sizes of different vehicles, the vehicle's passage sequence and speed planning are converted into vehicle trajectories described by a fifth-order polynomial. This allows the vehicle's controllable area to be extended from the stop line of the current lane (previously studied) to the conflict zone. The fuel emission model is also improved based on the differences between different vehicle types to ensure optimal traffic efficiency while reducing fuel consumption. When the vehicle density is high, a high-vehicle-density cooperative communication framework is designed, along with vehicle platooning rules. The platooned vehicle queue is treated as a single, overlong vehicle. A queue-based two-level planning and control method for unsignalized intersections is proposed. In the upper-level planning, an improved MCTS algorithm is used to improve solution efficiency and determine the order of the fleet based on traffic efficiency. In the lower-level planning, a distributed control method is used to optimize the fleet's speed and acceleration based on fuel consumption. This method achieves a balance between traffic efficiency and fuel consumption while reducing computational complexity.

[0076] Based on the above, an embodiment of the present invention provides a method for cooperative passage at an unsignalized intersection under mixed passenger and freight traffic based on IVCPS. The method is specifically described as follows:

[0077] 1. Division of different spatiotemporal scales under IVCPS

[0078] IVCPS cyber-physical systems have six major characteristics, including complexity, heterogeneity, openness, closed-loop, emergence, and evolution. It is difficult to analyze and design a large complex system at the same time and space scale. For IVCPS, the scales can be divided into node level, unit level, subsystem level, system level, and regional level from small to large in terms of time and space. Figure 1 At different time and space scales, the information space and physical space include the following:

[0079] Table 1 Multi-scale division of intelligent vehicle cyber-physical system

[0080]

[0081]

[0082] The traffic control architecture for signal-free intersections based on IVCPS at multiple scales can well combine information space with physical space, and use vehicle-road-cloud integration technology to achieve precise control. The signal-free intersection targeted by this embodiment is a "422" scenario, i.e., 4 directions, 2 intersecting roads, and 2 lanes. Here, only one lane of a road in one direction is taken as an example, and the remaining lanes are the same. When the vehicle leaves the conflict area, it will no longer be controlled. This embodiment will be based on the traffic control framework for signal-free intersections at multiple scales, and the signal-free intersection area will be further divided into observation sections, control sections, and collision areas to form a corresponding collaborative control framework, such as Figure 2 shown.

[0083] Under this framework, the information space can receive information from the vehicle side and the road section, and formulate corresponding control strategies based on the information provided. In the physical space, smart cars can be interconnected with other vehicles and road equipment in real time. After receiving the information transmitted by the information space, the vehicles in the physical space will continuously adjust their driving status to ensure precise control of unsignaled intersections.

[0084] Within this framework, the complex task of coordinated traffic flow at unsignalized intersections is broken down and refined into observation sections, control sections, and collision zones, achieving both local and global optimization. The main functions of each section are as follows:

[0085] Observation section: After the vehicle arrives at the observation section, the system will use vehicle-road-cloud integration technology to judge the vehicle density of the unsignalized intersection at the current system level, and at the same time accelerate the vehicle to the maximum speed to pass through the observation section.

[0086] Control section: When a vehicle arrives at the control section, the system will execute the corresponding control strategy based on the vehicle's judgment of vehicle density in the observation section. Under high vehicle density, the control area is further subdivided into a formation area and a decision area. After forming a team in the formation area, the vehicles can pass through the unsignaled intersection in a queue, reducing the control calculation complexity of the system and reducing the fuel consumption of the vehicle. After the vehicle arrives at the decision area, an upper and lower control model is constructed. The upper layer plans the vehicle's passage sequence, and the lower layer optimizes the vehicle's speed to achieve reduced fuel consumption while ensuring passage efficiency. Under low vehicle density, the intersection is modeled based on the three-vehicle model, and the vehicle trajectory is modeled. Finally, the passage sequence and speed planning are solved by the solution algorithm.

[0087] Conflict area: Under high vehicle density, after the control section adjusts the vehicle status, vehicles in all directions will pass through the conflict area without conflict; under low vehicle density, vehicles will continue to be controlled in this area until they leave the conflict area without conflict.

[0088] To focus on the coordinated control of unsignalized intersections with mixed passenger and freight traffic, this embodiment makes the following assumptions for the scenario:

[0089] (1) All controlled vehicles are intelligent connected vehicles, which are divided into intelligent connected manually driven vehicles and intelligent connected autonomous driving vehicles. Non-motor vehicles and pedestrians are not considered for the time being;

[0090] (2) Connected vehicles in different scenarios strictly follow the corresponding control strategies, and there is no subjective interference of drivers in vehicle operations in intelligent connected vehicles;

[0091] (3) Both the intelligent vehicle terminal and the intelligent roadside terminal operate in an ideal state, without packet loss or communication delay.

[0092] (4) The selected environments are all within the coverage of the Internet, and there is no overtaking or lane changing behavior after the vehicle enters the scene.

[0093] 2. Cooperative Traffic Methods under Different Vehicle Densities

[0094] 1. Cooperative traffic method under low vehicle density

[0095] When the vehicle density at an unsignalized intersection is low, in order to reduce the system's perception control range, Figure 2 The collaborative control framework shown in Figure 1 divides the unsignalized intersection into observation section, uncontrolled section, target optimization section, and conflict area. Figure 3As shown in the figure, the uncontrolled section indicates that the vehicle will not be controlled externally in this section and will maintain its speed until it reaches the target optimized section. When the vehicle is in the target optimized section or the conflict zone, it will be controlled by the cloud until it leaves the conflict zone.

[0096] A cooperative traffic method based on the three-vehicle model is proposed, and the target optimization section and conflict area are modeled. Figure 4 shown. Figure 4 In the example, vehicles from three directions enter the target optimization section at the same time. The lane width is W and the width of the vehicle is w. i,j , length l i,j , the acceleration is a i,j (t), with a velocity of v i,j (t), where i represents the lane number, j represents the jth vehicle in the current lane, and t represents time. An XOY coordinate system is established at the center of the unsignalized intersection, with the direction of the two conflicting regions in the direction of motion of a single vehicle defined as the Y-axis. The lane numbers of vehicles 1, 2, and 3 are represented by i1, i2, and i3, respectively, and the vehicle numbers of vehicles 1, 2, and 3 in their current lanes are represented by j1, j2, and j3, respectively.

[0097] Assume that the initial time is t0, the conflict area between vehicle 1 and vehicle 2 is conflict area 1, and its size is The length and time of vehicle 1’s front end arriving at conflict area 1 are The length and time of the rear end of the vehicle leaving conflict area 1 are The position of the center of mass of vehicle 1 in the direction of motion at time t0 is The initial coordinates in the XOY coordinate system are The length and time of vehicle 2’s front end arriving at conflict area 1 are

[0098] The length and time of the rear end of the vehicle leaving conflict area 1 are The position of the center of mass of vehicle 2 in the direction of motion at time t0 is

[0099] The collision area between vehicle 2 and vehicle 3 is collision area 2, and its size is The length and time of vehicle 2's front end arriving at conflict area 2 are The length and time of the rear end of the vehicle leaving conflict area 2 are The length and time of vehicle 3’s front end arriving at conflict area 1 are The length and time of the rear end of the vehicle leaving conflict area 1 are The position of the center of mass of vehicle 3 in the direction of motion at time t0 is The initial coordinates in the XOY coordinate system are

[0100] For the three-vehicle model, define the initial time and initial position of the cloud-based trajectory planning for vehicle 1 As well as the vehicle's end time and end location Initial time and initial position of vehicle 2 trajectory planning As well as the vehicle's end time and end location Initial time and initial position of vehicle 3 trajectory planning As well as the vehicle's end time and end location After completing the trajectory planned by the cloud, the vehicle will maintain the speed at the end of the trajectory planning until all vehicles have left the conflict area. The solution to some of the parameters defined above is as follows:

[0101]

[0102] A fifth-order polynomial is used as the vehicle trajectory model:

[0103] s i,j (t) = r i,j +r i,j,1 t i,j +r i,j,2 t i,j 2 +r i,j,3 t i,j 3 +r i,j,4 t i,j 4 +r i,j,5 t i,j 5

[0104] By taking the first and second order differentials of the above equation, we can get the functions of velocity and acceleration with respect to time:

[0105] v i,j (t) = r i,j,1 +2r i,j,2 t i,j +3r i,j,3 t i,j 2 +4r i,j,4 t i,j 3 +5r i,j,5 t i,j 4

[0106] a i,j (t) = 2r i,j,2 +6r i,j,3 ti,j +12r i,j,4 t i,j 2 +20r i,j,5 t i,j 3

[0107] In the trajectory planning of the vehicle, if the initial time and initial position of the vehicle (t0, s i,j (t0)) and the end time and end position of the vehicle (t i,j,f , s i,j (t f )), then all unknown quantities can be solved. With the support of IVCPS technology, the system can obtain s i,j (t0). Therefore, the position, velocity, and acceleration equations can all be converted into equations related to time, and then the optimal end time of each vehicle can be solved. Thus, the trajectory of each vehicle is obtained, and the trajectory planning problem is transformed into a multi-objective problem, and the three-vehicle coordinated traffic problem is transformed into a multi-constraint problem. The process is as follows:

[0108] 1) Construction of multi-objective optimization function

[0109] Design of the shortest travel time objective function: In the cooperative traffic method designed based on the three-vehicle model, the shortest travel time of the cooperative traffic method should be the shortest time for the rear ends of the three vehicles to pass through the conflict area. Its objective function is expressed as:

[0110]

[0111] Fuel consumption objective function design: VT-Mirco is selected as the fuel consumption model. The concept of unit efficiency (MOE) is introduced into the model. The calculation formula is as follows:

[0112]

[0113] Among them, MOE e is the unit change in vehicle fuel consumption and emissions, MOE is the regression model e The regression coefficient for velocity in g power and acceleration in h power, v i,j is the instantaneous speed of the vehicle, a i,j is the instantaneous acceleration of the vehicle. Taking into account the differences in mass between different types of vehicles, this embodiment adjusts the VT-Mirco model to obtain the constructed energy consumption formula:

[0114]

[0115] Among them, M Xis the weight of different vehicle types, specifically, M A For passenger cars, M B Truck, M C For a car. Therefore, the fuel consumption objective function F2 is:

[0116]

[0117] Comfort objective function design: In order to ensure the stability and comfort of the vehicle during driving and reduce energy consumption, it is necessary to reduce the frequent acceleration and deceleration of the vehicle, that is, to reduce the changes in vehicle speed and acceleration.

[0118]

[0119] Where ΔT is the simulation step size. Based on the above objective function, the multi-objective function can be transformed into a single objective function optimization:

[0120]

[0121] in, is the weight coefficient, which can be adjusted according to actual needs.

[0122] 2) Constructing constraints

[0123] Collision constraints:

[0124]

[0125] Velocity and acceleration constraints:

[0126] v n,min ≤v i,j (t)≤v n,max

[0127] a n,min ≤a i,j (t)≤a n,max

[0128] Among them, n=1, 2, and 3 represent three different types of vehicles: sedan, truck, and bus, respectively. n,max and v n,min are the maximum speed and minimum speed of the road respectively, a n,max and a n,min They are the maximum acceleration and minimum acceleration restricted by the road.

[0129] 3) Model solution

[0130] The simulated annealing algorithm is used to solve the single-objective optimization function proposed above. Since the above-mentioned objective optimization function contains constraints, the constraint function is converted into a penalty factor, and the objective function is converted into:

[0131]

[0132] Among them, γ is the penalty factor, which takes 0 and infinite integers, G Z (x) represents the constraints mentioned above.

[0133] 2. Cooperative traffic method under high vehicle density

[0134] When the vehicle density at an unsignalized intersection is high, in order to improve the traffic efficiency of the unsignalized intersection and reduce the fuel consumption of vehicles, the unsignalized intersection is divided into observation section, formation section, decision section, and conflict area, such as Figure 5 shown.

[0135] Under this framework, vehicles form a platoon in the platoon section according to the platoon rules. Figure 5 The red vehicle in the middle represents the leading vehicle in the convoy, and the black vehicle represents the following vehicle. Figure 6 The upper and lower layer controllers shown in the figure use the improved MCTS algorithm to plan the passage sequence of the fleet at the upper layer, and the MPC algorithm to plan the vehicle trajectory at the lower layer.

[0136] In the platoon section, when the distance D between vehicle j and vehicle j+1 is S Less than the maximum vehicle distance D H And when the following equation is satisfied, the two vehicles are transformed into the same team:

[0137]

[0138] Among them, L B is the length of the observation segment. s 、D H The calculation formulas are expressed as follows:

[0139]

[0140] In the platooning section, two vehicles will not participate in other platoon formations until they form a stable platoon. A stable platoon means that the speed and acceleration of the vehicles in the platoon are consistent, and the spacing between vehicles is the desired spacing. Once a platoon enters the decision section, it is considered a single entity. The platoon's speed is the minimum speed of all vehicles in the platoon, the maximum acceleration and deceleration is the minimum acceleration and deceleration of all vehicles in the platoon, and the maximum and minimum speeds are the maximum and minimum speeds of all vehicles in the platoon.

[0141] 1) MCTS algorithm plans the passage sequence

[0142] The upper-level controller uses a modified MCTS algorithm to sort the vehicles' travel order. The basic idea is to simulate a large number of possible decision paths to evaluate the potential value of each decision and use the simulation results to guide the next search. The algorithm consists of four main phases: selection, expansion, simulation, and backpropagation.

[0143] The selection phase mainly performs the following steps: starting from the root node, an unexplored child node is selected according to a certain strategy (such as the UCB algorithm) until a leaf node is reached or a limiting condition is met.

[0144] The expansion phase mainly performs the following: if the currently selected node has unexplored child nodes, one or more child nodes are expanded according to a certain strategy (such as exploring unknown areas or prioritizing nodes with high potential value).

[0145] The simulation phase mainly performs the following steps: for the expanded child nodes, a certain number of simulations are performed through random simulation or heuristic simulation to obtain a result.

[0146] The backtracking phase mainly performs the following tasks: tracing the simulation results back to the root node through the node path, and updating the statistical information of each node for use in the next round of selection.

[0147] like Figure 7 Figure 2 shows an iteration of the MCTS-based strategy, which includes the above four steps: selection, expansion, simulation, and backtracking.

[0148] In this embodiment, the UCT algorithm is used to evaluate the value of each child node in the selection step:

[0149]

[0150] Among them, β u is the total reward value of node u, p u is the number of times node u is visited, P is the total number of times the parent node is visited, and C is the trade-off coefficient used to adjust the exploration part and the utilization part, where the utilization part is The exploration section is The average delay of the vehicle is selected as β u Calculation of value:

[0151]

[0152] Among them, t i,j,ac represents the actual travel time of a single vehicle through an unsignalized intersection, t i,j,minIndicates the minimum travel time of a single vehicle. In previous studies, the C value is often set to a constant. In this embodiment, the dynamic window method is used to adjust the C value, and the size of the sliding window is set to W. M =30, the adjustment factor is k m =0.05, the minimum positive number is η=0.5, and the initial value of C is C init =1.4. The average delay time within the sliding window is:

[0153]

[0154] Calculate the overall average delay:

[0155]

[0156] The calculation formula of C value is as follows:

[0157]

[0158] C=max(C,η)

[0159] 2) MPC controller plans vehicle trajectory

[0160] The vehicle trajectory planning process based on MPC is as follows:

[0161] 1. Define the vehicle's state variables: position, velocity, and acceleration;

[0162] 2. Define the vehicle's dynamic model in discrete form:

[0163]

[0164] v i,j (t+1)=v i,j (t)+a i,j (t)ΔT

[0165] 3. Define optimization objectives and constraints;

[0166] 4. Based on the current state and the predicted control input sequence, calculate the future state of the vehicle over multiple time steps, generate a series of possible control input sequences (candidate trajectories), and predict the state sequence under these control inputs based on the vehicle dynamics model. Evaluate each candidate trajectory, calculate its corresponding objective function value, and check whether it meets all constraints. From the candidate trajectories, select the trajectory with the minimum objective function value and that meets all constraints as the optimal trajectory at the current moment;

[0167] 5. Apply the first control input of the optimal trajectory to the vehicle;

[0168] 6. Repeat the steps in each control cycle:

[0169] Update Status: Get the latest status of the vehicle.

[0170] Reforecast: Based on the updated state, re-forecast the future state.

[0171] Re-optimization: Resolve the optimization problem to obtain a new optimal control input sequence.

[0172] The optimization objectives and constraints of the MPC controller are as follows:

[0173] Set the prediction horizon length of the MPC controller to N P , control the time domain length to be N C , then the objective function of the controller is designed as follows:

[0174]

[0175] The objective function is divided into two parts:

[0176] Vehicle fuel consumption objective function:

[0177] Comfort objective function design:

[0178] The constraints are: vehicle speed and acceleration constraints:

[0179] v n,min ≤v i,j (t)≤v n,max

[0180] a n,min ≤a i,j (t)≤a n,max

[0181] Vehicle spacing distance constraint:

[0182]

[0183] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for cooperative passage at an unsignalized intersection with mixed passenger and freight traffic based on IVCPS, characterized by: Vehicle-side and road section information is transmitted to the information space through on-board sensors and roadside units. The vehicle density at unsignalized intersections is determined in the information space, and different control strategies are adopted according to different vehicle densities to achieve coordinated passage at unsignalized intersections. When vehicle density is low, the unsignalized intersection is segmented, and a real vehicle model is constructed based on a rectangular model. The unsignalized intersection is modeled using a three-vehicle model. Simultaneously, vehicle trajectories are modeled, and the trajectory model is solved to obtain the vehicle passage order and speed planning, enabling vehicles to coordinate through the unsignalized intersection. At high vehicle density, unsignalized intersections are segmented and vehicles within the unsignalized intersection area are platooned. A platoon-based two-layer control planning method is used for coordinated control, enabling vehicles to coordinate through the unsignalized intersection. In this two-layer control planning method, the upper layer uses an improved MCTS algorithm to plan the platoon's passage sequence, while the lower layer uses an MPC algorithm to plan vehicle trajectories. The upper layer plans the passage order of the fleet through an improved MCTS algorithm, including: in the MCTS algorithm, each node of the search tree represents a state, and each state contains the position and speed information of all vehicles at the current moment; the MCTS algorithm repeatedly executes the four stages of selection, expansion, simulation, and backtracking, and stops after reaching a predetermined calculation time or number of iterations; In the selection phase, starting from the root node, an optimal child node is selected for expansion through the UCT algorithm; In the expansion phase, starting from the selected optimal child node, a new vehicle passage sequence is simulated, a new child node is generated and added to the search tree; In the simulation phase, for the expanded child nodes, starting from the root node, the process of vehicles passing through the unsignalized intersection is simulated by randomly selecting the passage order of vehicles, and the simulation process is continued downward along a path of the search tree passing through the expanded child nodes until all vehicles pass through the unsignalized intersection; In the backtracking phase, the simulation result is traced back to the root node through the path selected in the simulation phase, and the values ​​and access times of the passed nodes are updated at the same time.

2. The method for cooperative passage at an unsignalized intersection according to claim 1, characterized in that: When the vehicle density is low, the unsignalized intersection is divided into an observation section, an uncontrolled section, a target optimization section, and a conflict area; in the uncontrolled section, the vehicle is not controlled and the vehicle maintains its speed to the target optimization section; in the target optimization section and the conflict area, the vehicle is controlled until the vehicle leaves the conflict area.

3. The method for cooperative passage at an unsignalized intersection according to claim 2, characterized in that: Based on the segment division, the unsignalized intersection is modeled using the three-vehicle model. The model includes: taking the center of the unsignalized intersection as the origin O, establishing an XOY coordinate system, and defining the direction of two conflict areas in the direction of movement of a single vehicle as the Y axis; vehicles from three directions enter the target optimization segment at the same time, defining the lane width of the unsignalized intersection as W and the vehicle width as w i,j , the vehicle length is l i,j , the vehicle speed and acceleration are v i,j (t) and a i,j (t), where i represents the lane, j represents the vehicle in the lane, and t represents the time; The conflict area between the first and second vehicles is defined as the first conflict area, and the conflict area between the second and third vehicles is defined as the second conflict area; i1, i2, and i3 represent the lanes where the first, second, and third vehicles are located, respectively; j1, j2, and j3 represent the vehicle numbers of the first, second, and third vehicles in the lanes, respectively; Define the initial time and initial position of the trajectory planned by the cloud for the first vehicle as t0 and Define the end time and end position of the first vehicle trajectory planning as and represents the time when the front of the first vehicle arrives at the first conflict area; the initial time and initial position of the trajectory planning of the second vehicle in the cloud are defined as t0 and Define the end time and end position of the second vehicle trajectory planning as follows: and represents the time when the front of the second vehicle arrives at the first conflict area; the initial time and initial position of the third vehicle trajectory planning in the cloud are defined as t0 and The end time and end position of the third vehicle trajectory planning are defined as follows: and Indicates the time when the front end of the third vehicle arrives at the second conflict area.

4. The method for cooperative passage at an unsignalized intersection according to claim 1, characterized in that: A fifth-order polynomial is used as the vehicle trajectory model, which is expressed as: s i,j (t)=r i,j +r i,j,1 t i,j +r i,j,2 t i,j 2 +r i,j,3 t i,j 3 +r i,j,4 t i,j 4 +r i,j,5 t i,j 5 Where r i,j 、r i,j,1 、r i,j,2 、r i,j,3 、r i,j,4 、r i,j,5 All represent coefficients, t i,j represents the travel time of vehicle j in lane i; the first-order and second-order differentials of the above equation are obtained to obtain the functions of velocity and acceleration with respect to time: v i,j (t)=r i,j,1 +2r i,j,2 t i,j +3r i,j,3 t i,j 2 +4r i,j,4 t i,j 3 +5r i,j,5 t i,j 4 a i,j (t)=2r i,j,2 +6r i,j,3 t i,j +12r i,j,4 t i,j 2 +20r i,j,5 t i,j 3 The trajectory of each vehicle is obtained by solving the optimal end time of each vehicle, where the end time is the time when each vehicle arrives at the conflict area; The trajectory planning problem is transformed into a multi-objective optimization problem. The designed objective functions include the shortest travel time objective function, the fuel consumption objective function, and the comfort objective function. At the same time, constraints are constructed. Among them, the travel time objective function is optimized to minimize the end time. Multiple objective functions are converted into a single objective function through weighted summation, and then the simulated annealing algorithm is used to solve the single objective function to obtain the optimal end time.

5. The method for cooperative passage at an unsignalized intersection according to claim 1, characterized in that: When the vehicle density is high, the unsignalized intersection is divided into an observation section, a formation section, a decision section and a conflict area; vehicles are arranged into queues in the formation section; and collaborative control is performed in the decision section using the two-layer control planning method.

6. The method for cooperative passage at an unsignalized intersection according to claim 1, characterized in that: The UCT algorithm is used to evaluate the value of each child node in the selection phase, which is expressed as: Where, β u represents the total reward value of node u, p u represents the number of times node u is visited, P represents the total number of times the parent node is visited, and C represents the trade-off coefficient used to adjust the exploration part and the utilization part, where the utilization part is The exploration section is The average delay of the vehicle is selected as β u Calculation of value: Where, t i,j,ac represents the actual travel time of a single vehicle through an unsignalized intersection, t i,j,min represents the minimum travel time of a single vehicle, i represents lane i, and j represents the jth vehicle in the current lane.

7. The method for cooperative passage at an unsignalized intersection according to claim 6, characterized in that: The UCT algorithm is improved by using a dynamic window method to adjust the C value and defining the sliding window size W. m , adjustment factor k m , number of iterations n, initial C value C init and a positive number η; Calculate the average delay time AD within the sliding window W : Calculate the overall average delay time AD T : Then the C value is adjusted by the following formula: C=max(C,η).

8. The method for cooperative passage at an unsignalized intersection according to claim 1, characterized in that: The lower layer plans the vehicle trajectory through the MPC algorithm, including: 1) Define the vehicle's state variables, including position, velocity, and acceleration; 2) Define the vehicle's dynamic model in discrete form: v i,j (t+1)=v i,j (t)+a i,j (t)ΔT Where s i,j (t) The position of vehicle j in lane i at time t, v i,j (t) represents the speed of vehicle j in lane i at time t, ΔT represents the simulation step size, and a i,j (t) represents the acceleration of vehicle j in lane i at time t; 3) Define optimization objectives and constraints; 4) Based on the current state and the predicted control input sequence, the future state of the vehicle within multiple time steps is calculated to generate a series of control input sequences, namely candidate trajectories, and the state sequence under the generated control input sequence is predicted in the vehicle dynamics model; each candidate trajectory is evaluated, the corresponding objective function value is calculated, and all constraints are checked for satisfaction; and the trajectory with the minimum objective function value and satisfying all constraints is selected from the candidate trajectories as the optimal trajectory at the current moment; wherein the control input sequence includes the state sequence and the predicted sequence; 5) Apply the state sequence of the optimal trajectory to the vehicle; 6) Repeat the following steps in each control cycle: Update status: Get the latest status of the vehicle; Reforecast: Based on the updated state, re-forecast the future state; Re-optimization: Resolve the optimization problem to obtain a new optimal control input sequence.

Images