Lane-constraint-free intersection cooperative traffic control method based on Euler helix

Abstract

The invention relates to the field of automatic driving, in particular to a lane-constraint-free intersection cooperative traffic control method based on an Euler helix, and the method comprises the steps: adding an intelligent network connection vehicle into an ordered registration set; extracting the intelligent network connection vehicle from the ordered registration set, and establishing an optimizable lane-constraint-free track from an intersection entrance stop line to a target exit lane; the method comprises the following steps: acquiring intelligent network connected vehicle sensing data, and generating a state-action-reward tuple; inputting the state-action-reward tuple into a decision neural network and an evaluation neural network, and updating network parameters through dual-network cooperative training; and iteratively generating optimization control actions of each sub-track section, and combining to form a complete passing track of the intelligent network connection vehicle. According to the technical scheme, the intelligent network connection vehicles can interact and share the motion state and cooperatively generate the smooth collision-free passing track, the driving flexibility of the intelligent network connection vehicles in the intersection is improved, and the limitation of a traditional passing mode based on fixed lane line constraint is broken through.

Description

Technical Field

[0001] The present invention relates to the technical fields of computers, traffic big data and autonomous driving trajectory planning, and in particular to a coordinated traffic control method for an intersection without lane constraints based on an Euler spiral. Background Art

[0002] Intersections, as nodes where multidirectional traffic flows converge and diverge, are core hubs in urban road networks. Traditional intersection management faces dual challenges: on the one hand, the phase cycle duration and switching sequence of traditional signal lights cannot adapt to traffic fluctuations in real time; on the other hand, the reaction delay of human drivers exacerbates the contradiction between traffic efficiency and safety. With the development of intelligent connected vehicle technology, V2X communication can achieve 10-millisecond information exchange between vehicles, providing a technical foundation for breaking through traditional traffic models.

[0003] Currently, many researchers have proposed collision-avoiding methods for intelligent connected vehicle cooperative traffic at intersections based on fixed lane constraints, including adaptive traffic lights and right-of-way allocation strategies based on spatiotemporal conflict points. However, these methods are unable to overcome fixed lane constraints and can only resolve collision avoidance issues by slowing down or even stopping the vehicle, resulting in increased energy consumption and reduced traffic efficiency. Therefore, a technical solution that supports multi-degree-of-freedom trajectory planning without lane constraints and enables smooth steering control with continuous and differentiable curvature is urgently needed to comprehensively address the challenges of existing technologies. Summary of the Invention

[0004] The purpose of the present invention is to provide a lane-free intersection cooperative traffic control method based on Euler spirals, which can generate lane-free trajectories for intelligent connected vehicles, enabling them to flexibly turn within the intersection to avoid collisions, thereby reducing energy consumption and improving intersection safety.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0006] 1. A coordinated traffic control method for an intersection without lane constraints based on Euler spirals includes the following steps:

[0007] R1: When the intelligent connected vehicle reaches the preset registration distance d from the geometric center point O of the intersection reg When , the intelligent connected vehicle is added to the ordered registration set L;

[0008] R2: Sequentially extract the intelligent connected vehicle i from L obtained in step R1, and establish an optimizable trajectory without lane constraints from the intersection entrance stop line to the target exit lane. The optimizable trajectory without lane constraints consists of an Euler spiral.

[0009] R3: Acquire the perception data of the intelligent connected vehicle i, construct the state space S containing the vehicle state and environment information, define the action space A containing the trajectory control parameters, construct the reward function space R containing the trajectory quality index, and generate the state-action-reward tuple e;

[0010] R4: Input the tuple e into the decision neural network and the evaluation neural network. The former inputs the state and outputs the expected action to control the parameters of the Euler spiral, while the latter evaluates the value of the current state after executing the expected action. The network parameters are updated through the collaborative training of the two networks.

[0011] R5 iteratively generates optimized control actions for each sub-trajectory segment, which are combined to form the complete passage trajectory of the intelligent connected vehicle i.

[0012] Wherein, step R2 is specifically as follows:

[0013] R2-1 extracts the intelligent connected vehicle i in L in sequence and obtains the coordinates of the stop line of the current lane Target exit lane coordinates

[0014] R2-2 constructs an optimized laneless trajectory, which is composed of n segments of Euler spiral sub-trajectories connected end to end. The curve expression of the j-th segment is: in represents the initial point of the intelligent connected vehicle i in the jth sub-trajectory, represents the initial steering angle of the intelligent connected vehicle i at the starting point of the jth sub-trajectory, s represents the arc length from the starting point to the current coordinate, represents the linear change law of the curvature of the j-th sub-trajectory segment, represents the initial point curvature, represents the curvature change rate parameter, represents the horizontal and vertical coordinates of the intelligent connected vehicle i at the end point of the j-th sub-trajectory (also the starting point of the j+1-th sub-trajectory), is the total arc length of the j-th sub-trajectory;

[0015] Wherein, step R3 is specifically as follows:

[0016] R3-1 When the intelligent connected vehicle i reaches the starting point of the jth sub-trajectory segment, it obtains the initial point of the current sub-trajectory segment and the initial point curvature Current speed of intelligent connected vehicle i The end coordinates T of each sub-trajectory segment of all intelligent connected vehicles before i in the ordered registration set constitute the state

[0017] R3-2 Intelligent Connected Vehicle i Execution Action The arc length from the starting point is The Euler spiral reaches the end point of the j-th sub-trajectory segment (the starting point of the j+1-th sub-trajectory) and changes the speed linearly according to the constant acceleration. in represents the velocity change when reaching the end point of the j-th sub-trajectory segment;

[0018] When R3-3 intelligent connected vehicle i reaches the starting point of the j+1th sub-trajectory segment, it obtains the starting point of the current sub-trajectory segment and the initial point curvature Current speed of intelligent connected vehicle i The end coordinates T of each sub-trajectory segment of all intelligent connected vehicles before i in the ordered registration set constitute the state

[0019] R3-4 calculates the reward for intelligent connected vehicle i from the starting point to the end point of the jth sub-trajectory segment in represents the length between the end point of the j-th sub-trajectory segment and the target lane, represents the navigation time of intelligent connected vehicle i in the jth sub-trajectory segment, Collision penalty for intelligent connected vehicle i with other vehicles in sub-trajectory segment j, [α, β, η] are weight parameters;

[0020] R3-5 obtains sample data through steps R3-1 to R3-4 And input into step R4.

[0021] Wherein, step R4 is specifically as follows:

[0022] R4-1 will Input into the decision neural network and output the decision action Then Input is fed into the evaluation neural network and the output is in state Next action The value of The decision neural network and the evaluation neural network are composed of several hidden layers and an output layer. Each hidden layer contains multiple neurons, and the output of each neuron can be expressed as n represents the number of neurons, [w1,w2...,w n ] represents the weight parameter, [b1,b2...,b n ] represents the bias term, [x1,x2...,x n ] represents the output of the neurons in the previous hidden layer, f(·) represents the activation function, and the output layer of the decision neural network is composed of The output layer of the evaluation neural network is composed of ReLU(z)=max(0,z);

[0023] R4-2 will Input is fed into the evaluation neural network and the output is in state Next action The value of

[0024] R4-3 Calculate the loss value The neural network weight parameters are updated and evaluated by the gradient descent method, where g(·) represents the mean square error function and γ represents the discount rate;

[0025] R4-4 will Input into the decision neural network and output the decision action Then Input is fed into the evaluation neural network and the output is in state Next action The value of

[0026] R4-5 Calculate the decision loss value The decision neural network weight parameters are updated by gradient descent method.

[0027] The Euler spiral-based coordinated traffic control method for intersections without lane constraints provided in the above technical solution has the following advantages compared to the existing technology:

[0028] 1. This invention utilizes Euler spirals to construct multiple optimizable sub-trajectories, whose curvature varies continuously according to a linear law. This allows for a smooth trajectory for intelligent connected vehicles within intersections, avoiding vehicle steering jitter caused by sudden changes in curvature and significantly improving passenger comfort. Furthermore, by combining multiple optimizable sub-trajectories, this approach breaks through the traditional fixed lane model, allowing intelligent connected vehicles to flexibly plan their paths within intersections and achieve collision avoidance through accelerated steering, fully utilizing intersection space resources and improving energy efficiency and driving safety.

[0029] 2. Based on the collaborative training mechanism of decision network and evaluation network, the vehicle sub-trajectory parameters are realized and speed change Dynamic optimization is implemented. The decision network uses a tanh activation function to constrain the action output range, while the evaluation network utilizes ReLU to handle nonlinear value assessment. Collaborative gradient updates between the two networks ensure rapid policy convergence. Furthermore, through the input of training samples in the form of state-action-reward tuples, intelligent connected vehicles can adapt to random traffic flows and interference, achieving multi-objective optimization for traffic safety and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 Flowchart of a method for cooperative traffic control at an intersection without lane constraints based on Euler spirals according to an embodiment of the present invention;

[0031] Figure 2 An example map of lane-free trajectories that can be optimized as a benchmark of the present invention;

[0032] Figure 3 This is a diagram of the collaborative training mechanism of the decision network and the evaluation network of the present invention; DETAILED DESCRIPTION

[0033] In order to make the purpose and advantages of the present invention more clearly understood, the present invention is described in detail below with reference to the following examples. It should be understood that the following text is only used to describe one or more specific embodiments of the present invention and does not strictly limit the scope of protection of the present invention.

[0034] Next, combine the Figures 1 to 3 , the present invention is further analyzed and explained.

[0035] like Figure 1 As shown in the embodiment of the present application, an embodiment of a method for coordinated traffic control at an intersection without lane constraints based on an Euler spiral is provided, and the specific steps are as follows:

[0036] Step 1: When the intelligent connected vehicle reaches the preset registration distance d from the geometric center point O of the intersection reg When , the intelligent connected vehicle is added to the ordered registration set L;

[0037] Step 2: Sequentially extract the intelligent connected vehicle i from L obtained in step 1, and establish an optimizable trajectory without lane constraints from the intersection entrance stop line to the target exit lane, wherein the optimizable trajectory without lane constraints is composed of an Euler spiral.

[0038] 2.1 Extract the intelligent connected vehicle i in L in order and obtain the coordinates of the stop line of the current lane Target exit lane coordinates

[0039] 2.2 Construct an optimizable lane-free trajectory, which is composed of n segments of Euler spiral sub-trajectories connected end to end, such as Figure 2 As shown, the expression of the j-th segment curve is: in represents the initial point of the intelligent connected vehicle i in the jth sub-trajectory, represents the initial steering angle of the intelligent connected vehicle i at the starting point of the jth sub-trajectory, s represents the arc length from the starting point to the current coordinate, represents the linear change law of the curvature of the j-th sub-trajectory segment, represents the initial point curvature, represents the curvature change rate parameter, represents the horizontal and vertical coordinates of the intelligent connected vehicle i at the end point of the j-th sub-trajectory (also the starting point of the j+1-th sub-trajectory), is the total arc length of the j-th sub-trajectory;

[0040] Step 3: Obtain the perception data of the intelligent connected vehicle i, construct the state space S containing the vehicle state and environment information, define the action space A containing the trajectory control parameters, construct the reward function space R containing the trajectory quality index, and generate the state-action-reward tuple e;

[0041] 3.1 When the intelligent connected vehicle i reaches the starting point of the jth sub-trajectory segment, the initial point of the current sub-trajectory segment is obtained and the initial point curvature Current speed of intelligent connected vehicle i The end coordinates T of each sub-trajectory segment of all intelligent connected vehicles before i in the ordered registration set constitute the state

[0042] 3.2 Intelligent Connected Vehicle i Execution Action The arc length from the starting point is The Euler spiral reaches the end point of the j-th sub-trajectory segment (the starting point of the j+1-th sub-trajectory) and changes the speed linearly according to the constant acceleration. in represents the velocity change when reaching the end point of the j-th sub-trajectory segment;

[0043] 3.3 When the intelligent connected vehicle i reaches the starting point of the j+1th sub-trajectory segment, it obtains the starting point of the current sub-trajectory segment and the initial point curvature Current speed of intelligent connected vehicle i The end coordinates T of each sub-trajectory segment of all intelligent connected vehicles before i in the ordered registration set constitute the state

[0044] 3.4 Calculate the reward for intelligent connected vehicle i from the starting point to the end point of the jth sub-trajectory segment in represents the length between the end point of the j-th sub-trajectory segment and the target lane, represents the navigation time of intelligent connected vehicle i in the jth sub-trajectory segment, Collision penalty for intelligent connected vehicle i with other vehicles in sub-trajectory segment j, [α, β, η] are weight parameters;

[0045] 3.5 Obtain sampling data through steps 3.1 to 3.4 And enter it in step 4.

[0046] Step 4: Input the tuple e into the decision neural network and the evaluation neural network. The former inputs the state and outputs the expected action to control the parameters of the Euler spiral, while the latter evaluates the value of the current state after executing the expected action. The network parameters are updated through the collaborative training of the two networks, as shown in the following example: Figure 3 As shown;

[0047] 4.1 Input into the decision neural network and output the decision action Then Input is fed into the evaluation neural network and the output is in state Next action The value of The decision neural network and the evaluation neural network are composed of several hidden layers and an output layer. Each hidden layer contains multiple neurons, and the output of each neuron can be expressed as n represents the number of neurons, [w1,w2...,w n ] represents the weight parameter, [b1,b2...,b n ] represents the bias term, [x1,x2...,x n ] represents the output of the neurons in the previous hidden layer, f(·) represents the activation function, and the output layer of the decision neural network is composed of The output layer of the evaluation neural network is composed of ReLU(z)=max(0,z);

[0048] 4.2 Input is fed into the evaluation neural network and the output is in state Next action The value of

[0049] 4.3 Calculating Loss The neural network weight parameters are updated and evaluated by the gradient descent method, where g(·) represents the mean square error function and γ represents the discount rate;

[0050] 4.4 will Input into the decision neural network and output the decision action Then Input is fed into the evaluation neural network and the output is in state Next action The value of

[0051] 4.5 Calculating decision loss The decision neural network weight parameters are updated by gradient descent method.

[0052] Step 5: Iteratively generate the optimized control actions for each sub-trajectory segment and combine them to form the complete passage trajectory of the intelligent connected vehicle i.

[0053] The above describes the embodiments of the present invention in detail with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. After knowing the contents described in the present invention, ordinary technicians in this technical field can make several equivalent changes and substitutions without departing from the principles of the present invention. These equivalent changes and substitutions should also be regarded as falling within the scope of protection of the present invention.

Claims

1. A coordinated traffic control method for intersections without lane constraints based on Euler spirals, characterized in that: The following steps are involved: R1 When the intelligent connected vehicle reaches the preset registration distance d from the geometric center point O of the intersection reg When , the intelligent connected vehicle is added to the ordered registration set L; R2 sequentially extracts the intelligent connected vehicle i from L obtained in step R1 and establishes an optimizable trajectory without lane constraints from the intersection entrance stop line to the target exit lane. The optimizable trajectory without lane constraints is composed of several Euler spiral sub-trajectory segments connected end to end; R3 obtains the perception data of the intelligent connected vehicle i, constructs the state space S containing the vehicle state and environment information, constructs the action space A containing the trajectory control parameters, constructs the reward function space R containing the trajectory quality index, and generates the state-action-reward tuple e; R4 inputs the tuple e into the decision neural network and the evaluation neural network. The former inputs the state and outputs the expected action to control the parameters of the Euler spiral, while the latter evaluates the value of the current state after executing the expected action. The network parameters are updated through the collaborative training of the two networks. R5 iteratively generates optimized control actions for each sub-trajectory segment, which are combined to form the complete passage trajectory of the intelligent connected vehicle i.

2. The Euler spiral-based coordinated traffic control method for intersections without lane constraints according to claim 1, characterized in that: Step R2 is specifically as follows: R2-1 extracts the intelligent connected vehicle i in L in sequence and obtains the coordinates of the stop line of the current lane Target exit lane coordinates R2-2 constructs an optimized laneless trajectory, which is composed of n segments of Euler spiral sub-trajectories connected end to end. The curve expression of the j-th segment is: in represents the initial point of the intelligent connected vehicle i in the jth sub-trajectory, represents the initial steering angle of the intelligent connected vehicle i at the starting point of the jth sub-trajectory, s represents the arc length from the starting point to the current coordinate, represents the linear change law of the curvature of the j-th sub-trajectory segment, represents the initial point curvature, represents the curvature change rate parameter, represents the horizontal and vertical coordinates of the intelligent connected vehicle i at the end point of the j-th sub-trajectory (also the starting point of the j+1-th sub-trajectory), is the total arc length of the j-th sub-trajectory.

3. The Euler spiral-based coordinated traffic control method for intersections without lane constraints according to claim 1, characterized in that: Step R3 is specifically as follows: R3-1 When the intelligent connected vehicle i reaches the starting point of the jth sub-trajectory segment, it obtains the initial point of the current sub-trajectory segment and the initial point curvature Current speed of intelligent connected vehicle i The end coordinates T of each sub-trajectory segment of all intelligent connected vehicles before i in the ordered registration set constitute the state R3-2 Intelligent Connected Vehicle i Execution Action The arc length from the starting point is The Euler spiral reaches the end point of the j-th sub-trajectory segment (the starting point of the j+1-th sub-trajectory) and changes the speed linearly according to the constant acceleration. in represents the velocity change when reaching the end point of the j-th sub-trajectory segment; When R3-3 intelligent connected vehicle i reaches the starting point of the j+1th sub-trajectory segment, it obtains the starting point of the current sub-trajectory segment and the initial point curvature Current speed of intelligent connected vehicle i The end coordinates T of each sub-trajectory segment of all intelligent connected vehicles before i in the ordered registration set constitute the state R3-4 calculates the reward for intelligent connected vehicle i from the starting point to the end point of the jth sub-trajectory segment in represents the length between the end point of the j-th sub-trajectory segment and the target lane, represents the navigation time of intelligent connected vehicle i in the jth sub-trajectory segment, Collision penalty for intelligent connected vehicle i with other vehicles in sub-trajectory segment j, [α, β, η] are weight parameters; R3-5 obtains sample data through steps R3-1 to R3-4 And input into step R4.

4. The Euler spiral-based coordinated traffic control method for intersections without lane constraints according to claim 1, characterized in that: Step R4 is specifically as follows: R4-1 will Input into the decision neural network and output the decision action Then Input is fed into the evaluation neural network and the output is in state Next action The value of The decision neural network and the evaluation neural network are composed of several hidden layers and an output layer. Each hidden layer contains multiple neurons, and the output of each neuron can be expressed as n represents the number of neurons, [w1,w2...,w n ] represents the weight parameter, [b1,b2...,b n ] represents the bias term, [x1,x2...,x n ] represents the output of the neurons in the previous hidden layer, f(·) represents the activation function, and the output layer of the decision neural network is composed of The output layer of the evaluation neural network is composed of ReLU(z)=max(0,z); R4-2 will Input is fed into the evaluation neural network and the output is in state Next action The value of R4-3 Calculate the loss value The neural network weight parameters are updated and evaluated by the gradient descent method, where g(·) represents the mean square error function and γ represents the discount rate; R4-4 will Input into the decision neural network and output the decision action Then Input is fed into the evaluation neural network and the output is in state Next action The value of R4-5 Calculate the decision loss value The decision neural network weight parameters are updated by gradient descent method.

Images