A method for optimizing vehicle merging order based on probabilistic motion prediction of adjacent vehicles

Abstract

This invention discloses a vehicle merging order optimization method based on the probabilistic motion prediction of adjacent vehicles. The method includes: numbering all main road vehicles within the merging control area; receiving position and speed information of main road vehicles within the merging control area from the roadside controller for ramp vehicles; constructing a vehicle driving behavior classifier using a Gaussian mixture model and solving for the probability of adjacent vehicles belonging to each driving behavior category; predicting the trajectory of adjacent vehicles using a probabilistic motion prediction method considering driving behavior; designing a cumulative energy consumption cost function to construct a ramp vehicle merging order optimization problem and solving for the optimal merging order of ramp vehicles. This invention combines a Gaussian mixture model and an intelligent driver model, based on the probabilistic motion prediction of adjacent vehicles, to construct a ramp vehicle merging order optimization problem with the goal of minimizing energy consumption, and solves for the optimal merging order of ramp vehicles, thus improving the energy efficiency of ramp vehicles during the merging process.

Description

Technical Field

[0001] This invention belongs to the field of autonomous vehicle technology, specifically relating to a method for optimizing vehicle merging order based on probabilistic motion prediction of adjacent vehicles. Background Technology

[0002] With my country's rapid economic development, the number of motor vehicles has been increasing, leading to frequent road traffic accidents that seriously threaten people's lives and property. Among these accidents, merging areas at ramps, which present road bottlenecks, are particularly prone to accidents. In these areas, a large number of vehicles from ramps are forced to merge into the main road, creating complex interplay between ramp and main road traffic flows, which greatly increases the risk of accidents. Autonomous driving technology is considered a potential solution to this problem. Autonomous vehicles can effectively perceive and predict the behavior of surrounding vehicles, making safe and efficient decisions and improving vehicle safety.

[0003] Currently, some studies have proposed effective solutions to the merging problem of vehicles on ramps within the merging area. Chinese invention patent application CN202210271057.X, entitled "Cooperative Merging Control Method for Road Merging Areas Based on Autonomous Vehicle Queues," proposes a cooperative platooning merging control method for autonomous vehicles in ramp merging areas. This method utilizes swarm intelligence algorithms to optimize the optimal order of vehicle groups passing through the merging point and employs a car-following model (CACC) to control the longitudinal movement of vehicles, thereby improving road capacity. Chinese invention patent application CN 202211509503.2, entitled "A Cooperative Control Method for Vehicle Ramps Considering Energy Consumption of Multiple Vehicle Types," proposes a cooperative merging control method considering the energy consumption of heterogeneous vehicle types. This method considers the differences in mass and air resistance characteristics of different vehicle types, establishes energy consumption functions for heterogeneous vehicle types, and uses a pruning strategy to solve for the optimal merging order, reducing energy consumption when vehicles merge at ramp entrances. Although the existing methods described above optimize the order in which vehicles merge into the area via ramps, the optimization process ideally assumes that all vehicles in the merging area are controlled by a centralized controller and drive in the optimal order. It does not consider the motion prediction of other vehicles in the merging area under non-connected environments, which leads to the optimized order not being optimal and needs further improvement. Summary of the Invention

[0004] To address the shortcomings of the existing technologies, the present invention aims to provide a vehicle merging order optimization method based on the probabilistic motion prediction of adjacent vehicles. This method employs a Gaussian mixture model to construct a neighboring vehicle driving behavior recognizer; combines it with an intelligent driver model to perform probabilistic motion prediction of adjacent vehicles; designs a cumulative energy consumption cost function; constructs a merging order optimization problem for ramp vehicles; solves for the optimal merging order of ramp vehicles; and improves the energy efficiency and economy of ramp vehicles in the ramp merging area.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] The present invention provides a method for optimizing vehicle merging order based on probabilistic motion prediction of adjacent vehicles, comprising the following steps:

[0007] Step 1: When vehicle i from the ramp enters the merging control area, it establishes communication with the roadside controller at the merging entrance; all main road vehicles within the merging control area are numbered in ascending order of their entry into the merging control area as {1,2,...,N}. m}, N m The number of vehicles merging into the main road within the control area; Ramp vehicle i receives the position and speed information of vehicles merging into the main road within the control area from the roadside controller;

[0008] Step 2: Construct a vehicle driving behavior classifier using a Gaussian mixture model and solve for the probability that adjacent vehicles belong to each driving behavior category;

[0009] Step 3: Predict the trajectories of adjacent vehicles using a probabilistic motion prediction method that takes into account driving behavior;

[0010] Step 4: Design the cumulative energy consumption cost function, construct the merging order optimization problem of ramp vehicles, and solve for the optimal merging order of ramp vehicles.

[0011] Furthermore, in step 1, the merging control area refers to the area at a distance S from the merging entrance. p Within the region, the length S p It is determined by the maximum communication distance of the roadside controller.

[0012] Furthermore, in step 2, adjacent vehicles refer to all main road vehicles merging into the control area.

[0013] Furthermore, in step 2, a Gaussian mixture model is used to construct a vehicle driving behavior classifier. Specifically, the vehicle driving behavior classifier based on the Gaussian mixture model is:

[0014]

[0015] In the formula, p(x) is the probability density function; the variable x = [v, Δv] p ,Δs p ] T v is the vehicle speed; Δv p Δs represents the relative speed between the vehicle and the vehicle in front. p α represents the relative position of the vehicle to the vehicle in front; c represents the driving behavior category, c∈{1,2,3}; c μ c and ∑ cThese are the model mixing coefficients, mean vector, and covariance matrix corresponding to driving behavior category c, respectively.

[0016] The parameters of the Gaussian mixture model, including the model mixing coefficient α, are obtained by using the expectation-maximization algorithm. c Mean vector μ c The sum of the covariance matrix ∑ c .

[0017] Furthermore, in step 2, the probability of adjacent vehicles belonging to different driving behaviors is specifically calculated as follows: based on the variable x of adjacent vehicle n at the current time t0. n (t0), the probability that the adjacent vehicle belongs to type c driving behavior is obtained by solving:

[0018]

[0019] In the formula, p(c|x n (t0)) represents the variable x of the adjacent vehicle n at time t0. n At time (t0), the probability that the adjacent vehicle belongs to driving behavior c; p(x n (t0)|μ c ,Σ c ) is the mean vector μ c The sum of the covariance matrix ∑ c When using a Gaussian mixture model, the variable for the neighboring vehicle n at time t0 is x. n The probability of (t0).

[0020] Furthermore, the adjacent vehicle probabilistic motion prediction method considering driving behavior in step 3 specifically involves the following: the vehicle's kinematic model is:

[0021]

[0022] In the formula, s, v, and a represent the vehicle's position, velocity, and acceleration, respectively.

[0023] The intelligent driver model is as follows:

[0024]

[0025] In the formula, v represents the expected acceleration corresponding to driving behavior of type C; n Let Δs be the speed of adjacent vehicle n; n Let v be the relative position of an adjacent vehicle n with respect to the vehicle in front of it; p,n Let n be the speed of the vehicle ahead of adjacent vehicle n; and These are the maximum and minimum accelerations corresponding to Class C driving behavior, respectively. and These are the expected speed, minimum stopping distance, and expected headway for Class C driving behavior, respectively.

[0026] Based on the intelligent driver model, the predicted acceleration of neighboring vehicle n for:

[0027]

[0028] In the formula, the predicted acceleration of adjacent vehicle n is used. Substituting these values ​​into the kinematic model, we obtain the predicted positions and velocities of the adjacent vehicles n.

[0029] Furthermore, in step 4, the merging order of ramp vehicle i is represented by variable g, which is defined as ramp vehicle i entering the merging point after the main road vehicle numbered g.

[0030] Furthermore, the cumulative energy consumption cost function in step 4 is:

[0031]

[0032] In the formula, J E For the cost of accumulated energy consumption; T p For prediction in the time domain; a i (t) is the acceleration of vehicle i on the ramp, which can be solved by the following formula:

[0033]

[0034] In the formula, s i and v i These represent the position and speed of vehicle i on the ramp; s0 and T are respectively. h These are the minimum stopping distance and the expected headway for vehicle i on the ramp, respectively. Main road vehicles g in T p The predicted position at time t0; t0 is the current time.

[0035] Furthermore, the optimization problem of the merging order of vehicles on the ramp in step 4 is specifically as follows:

[0036]

[0037] In the formula, N m Given the number of vehicles merging into the main road within the control area, solve the merging order optimization problem to obtain the optimal merging order g for vehicle i on the ramp. * .

[0038] The beneficial effects of this invention are:

[0039] This invention considers the individual driving behaviors of vehicles on the main road within the merging area of ​​a ramp. Combining a Gaussian mixture model and an intelligent driver model, it constructs a problem for optimizing the merging order of ramp vehicles with the goal of minimizing energy consumption based on the probabilistic motion prediction of adjacent vehicles. The optimal merging order of ramp vehicles is then solved, improving the energy efficiency and economy of ramp vehicles during the merging process. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the merged control area.

[0041] Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0042] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to embodiments and accompanying drawings. The content mentioned in the embodiments is not intended to limit the present invention.

[0043] Reference Figures 1-2 As shown, the present invention provides a method for optimizing vehicle merging order based on probabilistic motion prediction of adjacent vehicles, comprising the following steps:

[0044] Step 1: When vehicle i from the ramp enters the merging control area, it establishes communication with the roadside controller at the merging entrance; all main road vehicles within the merging control area are numbered in ascending order of their entry into the merging control area as {1,2,...,N}. m}, N m The number of vehicles merging into the main road within the control area; Ramp vehicle i receives the position and speed information of vehicles merging into the main road within the control area from the roadside controller;

[0045] The inflow control area refers to the area that is S distance from the inflow point. p Within the region, the length S p It is determined by the maximum communication distance of the roadside controller.

[0046] Step 2: Construct a vehicle driving behavior classifier using a Gaussian mixture model and solve for the probability that adjacent vehicles belong to each driving behavior category;

[0047] In step 2, adjacent vehicles refer to all main road vehicles merging into the control area.

[0048] Specifically, in step 2, a Gaussian mixture model is used to construct a vehicle driving behavior classifier. Specifically, the vehicle driving behavior classifier based on the Gaussian mixture model is:

[0049]

[0050] In the formula, p(x) is the probability density function; the variable x = [v, Δv] p ,Δsp ] T v is the vehicle speed; Δv p Δs represents the relative speed between the vehicle and the vehicle in front. p α represents the relative position of the vehicle to the vehicle in front; c represents the driving behavior category, c∈{1,2,3}; c μ c and ∑ c These are the model mixing coefficients, mean vector, and covariance matrix corresponding to driving behavior category c, respectively.

[0051] The parameters of the Gaussian mixture model, including the model mixing coefficients α, are obtained using the Expectation-Maximization (EM) algorithm. c Mean vector μ c The sum of the covariance matrix ∑ c .

[0052] The specific steps in step 2 to calculate the probability that adjacent vehicles belong to different driving behaviors are as follows: based on the variable x of adjacent vehicle n at the current time t0. n (t0), the probability that the adjacent vehicle belongs to type c driving behavior is obtained by solving:

[0053]

[0054] In the formula, p(c|x n (t0)) represents the variable x of the adjacent vehicle n at time t0. n At time (t0), the probability that the adjacent vehicle belongs to driving behavior c; p(x n (t0)|μ c ,Σ c ) is the mean vector μ c The sum of the covariance matrix ∑ c When using a Gaussian mixture model, the variable for the neighboring vehicle n at time t0 is x. n The probability of (t0).

[0055] Step 3: Based on the probability of adjacent vehicles belonging to each driving behavior category, the adjacent vehicle probability motion prediction method considering driving behavior is used to predict the motion trajectory of adjacent vehicles.

[0056] Specifically, the adjacent vehicle probabilistic motion prediction method considering driving behavior in step 3 is as follows: The vehicle's kinematic model is:

[0057]

[0058] In the formula, s, v, and a represent the vehicle's position, velocity, and acceleration, respectively.

[0059] The intelligent driver model is as follows:

[0060]

[0061] In the formula, v represents the expected acceleration corresponding to driving behavior of type C; n Let Δs be the speed of adjacent vehicle n; n Let v be the relative position of an adjacent vehicle n with respect to the vehicle in front of it; p,n Let n be the speed of the vehicle ahead of adjacent vehicle n; and These are the maximum and minimum accelerations corresponding to Class C driving behavior, respectively. and These are the expected speed, minimum stopping distance, and expected headway for Class C driving behavior, respectively.

[0062] Based on the intelligent driver model, the predicted acceleration of neighboring vehicle n for:

[0063]

[0064] In the formula, the predicted acceleration of adjacent vehicle n is used. Substituting these values ​​into the kinematic model, we obtain the predicted positions and velocities of the adjacent vehicles n.

[0065] Step 4: Design the cumulative energy consumption cost function, construct the merging order optimization problem of ramp vehicles, and solve for the optimal merging order of ramp vehicles;

[0066] In step 4, the merging order of ramp vehicle i is represented by variable g, which is defined as ramp vehicle i entering the merging point after the main road vehicle numbered g.

[0067] The cumulative energy consumption cost function is:

[0068]

[0069] In the formula, J E For the cost of accumulated energy consumption; T p For prediction in the time domain; a i (t) is the acceleration of vehicle i on the ramp, which can be solved by the following formula:

[0070]

[0071] In the formula, s i and v i These represent the position and speed of vehicle i on the ramp; s0 and T are respectively. h These are the minimum stopping distance and the expected headway for vehicle i on the ramp, respectively. Main road vehicles g in T p The predicted position at time t0; t0 is the current time.

[0072] The specific problem of optimizing the merging order of vehicles on ramps is as follows:

[0073]

[0074] In the formula, N m Given the number of vehicles merging into the main road within the control area, solve the merging order optimization problem to obtain the optimal merging order g for vehicle i on the ramp. * .

[0075] This invention has many specific applications. The above description is only a preferred embodiment of this invention. It should be noted that for those skilled in the art, several improvements can be made without departing from the principle of this invention, and these improvements should also be considered within the scope of protection of this invention.

Claims

1. A method for optimizing vehicle merging order based on probabilistic motion prediction of adjacent vehicles, characterized in that, The steps are as follows: Step 1: When a vehicle on the ramp enters the merging control area, it establishes communication with the roadside controller at the merging entrance; all vehicles on the main road within the merging control area are numbered; the vehicle on the ramp receives the position and speed information of the vehicles on the main road within the merging control area from the roadside controller; Step 2: Construct a vehicle driving behavior classifier using a Gaussian mixture model and solve for the probability that adjacent vehicles belong to each driving behavior category; Step 3: Predict the trajectories of adjacent vehicles using a probabilistic motion prediction method that takes into account driving behavior; Step 4: Design the cumulative energy consumption cost function, construct the merging order optimization problem of ramp vehicles, and solve for the optimal merging order of ramp vehicles; In step 2, a Gaussian mixture model is used to construct a vehicle driving behavior classifier. Specifically, the vehicle driving behavior classifier based on the Gaussian mixture model is as follows: In the formula, p(x) is the probability density function; the variable x = [v, Δv] p ,Δs p ] T v is the vehicle speed; Δv p Δs represents the relative speed between the vehicle and the vehicle in front. p α represents the relative position of the vehicle to the vehicle in front; c represents the driving behavior category, c∈{1,2,3}; c μ c and ∑ c These are the model mixing coefficients, mean vector, and covariance matrix corresponding to driving behavior category c, respectively. The parameters of the Gaussian mixture model, including the model mixing coefficient α, are obtained by using the expectation-maximization algorithm. c Mean vector μ c The sum of the covariance matrix ∑ c ; The specific steps in step 2 to calculate the probability that adjacent vehicles belong to different driving behaviors are as follows: based on the variable x of adjacent vehicle n at the current time t0. n (t0), the probability that the adjacent vehicle belongs to type c driving behavior is obtained by solving: In the formula, p(c|x n (t0)) represents the variable x of the adjacent vehicle n at time t0. n At time (t0), the probability that the adjacent vehicle belongs to driving behavior c; p(x n (t0)|μ c ,Σ c ) is using the mean vector μ c The sum of the covariance matrix ∑ c When using a Gaussian mixture model, the variable for the neighboring vehicle n at time t0 is x. n The probability of (t0); The adjacent vehicle probabilistic motion prediction method considering driving behavior in step 3 is as follows: The vehicle's kinematic model is: In the formula, s, v, and a represent the vehicle's position, velocity, and acceleration, respectively. The intelligent driver model is as follows: In the formula, v represents the expected acceleration corresponding to driving behavior of type C; n Let Δs be the speed of adjacent vehicle n; n Let v be the relative position of an adjacent vehicle n with respect to the vehicle in front of it; p,n Let n be the speed of the vehicle ahead of adjacent vehicle n; and These are the maximum and minimum accelerations corresponding to Class C driving behavior, respectively. and These are the expected speed, minimum stopping distance, and expected headway for Class C driving behavior, respectively. Based on the intelligent driver model, the predicted acceleration of neighboring vehicle n for: In the formula, the predicted acceleration of adjacent vehicle n is used. Substituting these values ​​into the kinematic model, we obtain the predicted positions and velocities of the adjacent vehicles n.

2. The vehicle merging order optimization method based on adjacent vehicle probabilistic motion prediction according to claim 1, characterized in that, In step 1, the merging control area refers to the distance from the merging entrance being S. p Within the region, the length S p It is determined by the maximum communication distance of the roadside controller.

3. The vehicle merging order optimization method based on adjacent vehicle probabilistic motion prediction according to claim 1, characterized in that, In step 2, adjacent vehicles refer to all main road vehicles merging into the control area.

4. The vehicle merging order optimization method based on adjacent vehicle probabilistic motion prediction according to claim 1, characterized in that, In step 4, the merging order of ramp vehicle i is represented by variable g, which is defined as ramp vehicle i entering the merging point after the main road vehicle numbered g.

5. The vehicle merging order optimization method based on adjacent vehicle probabilistic motion prediction according to claim 1, characterized in that, The cumulative energy consumption cost function in step 4 is: In the formula, J E For the cost of accumulated energy consumption; T p For prediction in the time domain; a i (t) is the acceleration of vehicle i on the ramp, which can be solved by the following formula: In the formula, s i and v i These represent the position and speed of vehicle i on the ramp; s0 and T are respectively. h These are the minimum stopping distance and the expected headway for vehicle i on the ramp, respectively. Main road vehicles g in T p The predicted position at time t0; t0 is the current time.

6. The vehicle merging order optimization method based on adjacent vehicle probabilistic motion prediction according to claim 5, characterized in that, The specific problem of optimizing the merging order of vehicles on the ramp in step 4 is as follows: In the formula, N m Given the number of vehicles merging into the main road within the control area, solve the merging order optimization problem to obtain the optimal merging order g for vehicle i on the ramp. * .

Images