Method, system, and storage medium for lane change decision before exiting a highway

Abstract

The application discloses a lane-changing decision-making method and system before a vehicle drives off a highway and a storage medium, and relates to the field of intelligent traffic management and control. The method of the application determines a termination distance probability distribution model and a termination distance classification boundary value by constructing a highway vehicle driving behavior model, simulating driving behavior of a vehicle driving off the highway under conditions of different state parameters and state variables, training a success rate evaluation model to accurately predict a success rate and uncertainty of the vehicle successfully driving off, determining a success rate evaluation model corresponding to the highway vehicle driving behavior model, constructing a model library, and selecting a suitable highway vehicle driving behavior model in the model library according to a to-be-decided road condition and in combination with a preset success rate and confidence, and predicting a starting distance and a termination distance by using the highway vehicle driving behavior model to determine a starting position of lane-changing before the vehicle drives off the highway, thereby guaranteeing safety of the vehicle driving off the highway and stability of a traffic flow.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent traffic control, and particularly relates to a lane changing decision method and system for a vehicle before exiting a highway and a storage medium. BACKGROUND

[0002] At present, intelligent networked vehicle "vehicle-road cloud integration" has become the key to promoting the industrialization and application of intelligent networked vehicles. Intelligent networked vehicle "vehicle-road cloud integration" uses advanced wireless communication and new generation Internet technologies to realize dynamic real-time information interaction between vehicles and roads in all directions, and carries out vehicle coordination and safety and road active control on the basis of full-time and space dynamic traffic information collection and fusion, fully realizes the effective cooperation of people, vehicles and roads, ensures traffic safety and improves traffic efficiency. By combining it with the current booming automatic driving technology, it not only promotes the rapid landing of automatic driving technology, but also provides services for automatic driving vehicles and manned vehicles, effectively solves the traffic control problem in mixed driving scenarios, is conducive to promoting comprehensive traffic management, and is an important development direction of the current intelligent transportation system.

[0003] At present, for the problem of how far in advance a vehicle changes lanes to the outside at the exit of a multi-lane highway, a human-driven vehicle mainly relies on the experience of the driver to determine, and most intelligent assisted driving vehicles change lanes through the construction of a lane changing model. The lane changing model is preset with a lane changing length in the training, calibration, testing and verification stages, but the existing technology rarely involves vehicle type, road type and traffic flow conditions in determining the lane changing strategy for a vehicle before exiting a highway, and lacks a decision-making method for dynamic adaptive decision-making for vehicle type, road type and traffic flow conditions, especially for the scenario where the innermost lane of the main road is an automatic driving exclusive lane.

[0004] Therefore, it is urgent to propose a lane changing decision method, system and storage medium for a vehicle before exiting a highway, so that the target vehicle can exit the exclusive lane to the target exit ramp without causing traffic flow interference area to expand due to early lane changing, or missing the exit due to late lane changing, effectively ensuring the driving safety of the vehicle before and after exiting the highway. SUMMARY

[0005] The present application aims to solve the above problems, and provides a lane changing decision method, system and storage medium before a vehicle drives off a highway, solving the problem that it is difficult to scientifically determine the lane changing starting position when a target vehicle starts to prepare to change lanes from the innermost automatic driving lane to the outermost exit ramp to drive off the highway. By combining traffic simulation methods with machine learning methods, the termination distance probability distribution model and the success rate evaluation model under different state parameter and state variable conditions are obtained by simulating the highway vehicle driving behavior model, a model library is constructed, and the accurate determination of the lane changing starting position of the target vehicle driving off the highway is realized, thereby providing a basis for ensuring the safe driving of the vehicle during the lane changing process on the highway.

[0006] To achieve the above object, the present application adopts the following technical solutions:

[0007] A lane changing decision method before a vehicle drives off a highway is used to determine the starting position of the vehicle before driving off the highway, and includes the following steps:

[0008] According to the road to be decided, the state parameters in the road to be decided are determined, the traffic flow state of the road to be decided is obtained in real time, the target success rate evaluation model is selected from the pre-constructed model library by combining the preset success rate and confidence of the target vehicle successfully driving off the exit ramp under the current state parameter;

[0009] The target success rate evaluation model selected is used to predict the lane changing success rate of the target vehicle driving off the highway from different starting positions in the road to be decided, and the starting position of the target vehicle before driving off the highway is determined;

[0010] The construction method of the model library includes:

[0011] A highway vehicle driving behavior model is constructed, Q sets of state parameters are preset, state variables are set for each set of state parameters, a plurality of simulation tests are performed by using the highway vehicle driving behavior model, the success rate evaluation model corresponding to the highway vehicle driving behavior model under each state parameter is obtained, and the model library is constructed based on the success rate evaluation model corresponding to each state parameter;

[0012] For each set of state parameters in the Q sets of state parameters, the method for determining the success rate evaluation model includes:

[0013] The state variables are set under selected state parameters, simulation results are obtained by using the expressway vehicle driving behavior model based on the multiple sets of state variables, a termination distance probability distribution model and a termination distance classification boundary value are determined according to the simulation results of the state variables, a success rate and an uncertainty of the target vehicle successfully driving off the exit ramp are calculated, and a success rate evaluation model corresponding to the current expressway vehicle driving behavior model state parameter is determined.

[0014] The expressway vehicle driving behavior model is provided with an expressway main road model and a vehicle model; the expressway main road model is provided with multiple lanes, wherein the innermost lane is a special lane and the outermost lane is an exit ramp; the vehicle model includes a target vehicle and multiple background vehicles, wherein the target vehicle travels on the special lane, the background vehicles are other vehicles in the expressway main road model except the target vehicle, and the target vehicle is applied with a car-road dynamic real-time information interaction vehicle terminal, and the background vehicles are not applied with the car-road dynamic real-time information interaction vehicle terminal;

[0015] The state parameters include a target vehicle type, a total number of lanes , a maximum speed limit and a minimum speed limit of each lane on the main road;

[0016] The state variables include a traffic flow state and a target vehicle lane-changing starting distance; the traffic flow state includes a single-lane average flow and a single-lane average speed in the simulation process; the starting distance is a longitudinal distance between a lane-changing position of the target vehicle on the innermost lane and a position of a guide area tip of the exit ramp, and the termination distance is a longitudinal distance between a lane-changing completion position of the target vehicle on the outermost lane and the position of the guide area tip of the exit ramp.

[0017] For each of the Q sets of state parameters, the determination method of the success rate evaluation model includes:

[0018] Step a, constructing an expressway vehicle driving behavior model and setting state parameters of the expressway vehicle driving behavior model;

[0019] Step b, setting multiple sets of state variables and a number of state variable combinations , determining state variables , and of each set of state parameters, wherein is a single-lane average flow, is a single-lane average speed, is a lane-changing starting distance of the target vehicle;

[0020] Step c, for the selected state variables, use the highway vehicle driving behavior model to perform... In each simulation test, the target vehicle's process of changing lanes from the dedicated lane and leaving the exit ramp during the driving of all vehicles on the main highway was simulated to obtain... The simulation results include a termination distance and a departure result, wherein the departure result is represented by a Boolean value, with a Boolean value of 1 if the target vehicle successfully departs and a Boolean value of 0 if the target vehicle fails to depart.

[0021] Step d: Based on the results obtained from multiple simulation experiments using the current state variables... For each termination distance, a probability density estimation method is used to determine the probability distribution model of the termination distance. ;

[0022] Step e: Based on the simulation results obtained from multiple simulation experiments with the current state parameters, the departure result corresponding to the termination distance is used as a Boolean value label to obtain... Using a set of termination distance samples, a pre-defined classifier is trained. The classifier predicts whether a target vehicle has successfully exited the exit ramp based on the termination distance, obtaining the critical termination distance value between successful and unsuccessful exits, and thus determining the termination distance classification boundary value. Uncertainty of classification boundaries ;

[0023] Step f, based on the termination distance probability distribution model in S1.4 and the termination distance classification boundary value in S1.5. Determine the success rate and uncertainty of the target vehicle successfully leaving the exit ramp under the current state variables;

[0024] Step g: Determine whether the current number of simulations exceeds the preset number of state variable combinations. If the preset number of state variable combinations has been exceeded If the condition is met, proceed to step h; otherwise, select the next set of state variables and return to steps c to f to continue the simulation experiment until the preset number of state variable combinations is exceeded. ,get A simulation data sample, wherein the simulation data sample is represented as ,in, For the first The state variables of this simulation experiment For the first The success rate of the simulation test is used to proceed to step h;

[0025] Step h: Determine the success rate evaluation model corresponding to the current highway vehicle driving behavior model state parameters. .

[0026] In step f, the success rate of the target vehicle successfully leaving the exit ramp under the current state parameter is:

[0027] ;

[0028] In the formula, For the first The success rate of the target vehicle successfully leaving the exit ramp in the simulation test; To terminate the distance probability distribution model; The total number of lanes. Lane numbering, where ; The maximum speed limit for the lane. The minimum speed limit for the lane; For the first The termination distance obtained from the simulation experiment For the first The average flow rate per lane in this simulation test For the first The average speed of a single lane in this simulation test For the first The initial distance of the target vehicle in the second simulation test for lane changing. For the first A simulation test of highway vehicle driving behavior model;

[0029] The uncertainty corresponding to the success rate of the target vehicle successfully leaving the exit ramp under the current state variable is:

[0030] ;

[0031] In the formula, For the first The uncertainty of the target vehicle successfully leaving the exit ramp in the simulation test.

[0032] In step h, a success rate assessment model is constructed based on a Gaussian process regression model. The success rate assessment model is trained using simulation data samples, with state variables as input features and success rate as the target variable output by the success rate assessment model.

[0033] In the success rate evaluation model training process, the state variable of the simulation data sample is input into the success rate evaluation model, the success rate estimation value is calculated according to the state variable by using the success rate evaluation model, the accuracy rate calculated by the success rate evaluation model is obtained in combination with the success rate corresponding to the simulation data sample, the accuracy rate calculated by the success rate evaluation model is compared with the preset precision, if the accuracy rate calculated by the success rate evaluation model does not reach the preset precision, the hyperparameters of the success rate evaluation model are optimized, the success rate evaluation model is continuously trained, otherwise, the trained success rate evaluation model is obtained.

[0034] The success rate evaluation model is:

[0035] ;

[0036] The success rate evaluation model is:

[0037] The kernel function is:

[0038] ;

[0039] In the formula, , The state variable in the simulation data sample, is the success rate prior distribution; is the mean function; is the kernel function; is the variance of the signal; is the length scale parameter, used to control the smoothness of the radial basis kernel function;

[0040] The hyperparameters of the success rate evaluation model are optimized by using a log-likelihood function, and the log-likelihood function is:

[0041] ;

[0042] In the formula, is the likelihood function; is the set of hyperparameters to be optimized; is the transpose matrix; is the covariance matrix calculated by the kernel function; is the variance of the noise; y is the success rate sample;

[0043] The log-likelihood function is optimized based on the gradient descent method, and after optimization, the mean and the variance of the new input feature are predicted as:

[0044] ;

[0045] ​​ ;

[0046] In the formula, is a unit matrix.

[0047] The embodiment of the application also provides a lane-changing decision-making system before a vehicle drives off a highway, comprising a model selection module, an output module and a model library component module.

[0048] The model selection module is configured to determine state parameters in a to-be-decided road according to a to-be-decided road condition, acquire a traffic flow state of the to-be-decided road in real time, and select a target success rate evaluation model from a pre-constructed model library in combination with a preset success rate of a target vehicle successfully driving off an exit ramp and a confidence level under a current state parameter.

[0049] The output module is configured to predict a starting distance and a terminal distance of the target vehicle successfully driving off the exit ramp in the to-be-decided road by using the selected target success rate evaluation model, and determine a starting position of the target vehicle changing lanes before driving off the highway.

[0050] The model selection module is in communication connection with the model library component module, and the model library component module is configured to:

[0051] construct a highway vehicle driving behavior model, preset Q groups of state parameters, set state variables for each group of state parameters, perform a plurality of simulation tests by using the highway vehicle driving behavior model, obtain a success rate evaluation model corresponding to the highway vehicle driving behavior model constructed according to each state parameter, and construct the model library based on the success rate evaluation model corresponding to each state parameter.

[0052] For each group of the Q groups of state parameters, the determination method of the success rate evaluation model comprises:

[0053] a plurality of groups of state variables are set under the selected state parameter, a plurality of simulation tests are performed by using the highway vehicle driving behavior model based on the plurality of groups of state variables to acquire simulation results, a terminal distance probability distribution model and a terminal distance classification boundary value are determined according to the simulation results of each state variable, a success rate and an uncertainty of the target vehicle successfully driving off the exit ramp are calculated, and a success rate evaluation model corresponding to a current highway vehicle driving behavior model state parameter is determined.

[0054] The embodiment of the application also provides a computer readable storage medium, which stores computer program instructions, and the program instructions are executed by a vehicle lane-changing decision-making system before driving off a highway to realize the steps of the above method.

[0055] The application has the following beneficial technical effects:

[0056] The present application provides a lane-changing decision-making method, system and storage medium for a vehicle before driving off a highway, simulates lane-changing driving-off behaviors of a target vehicle on the highway under different state parameters and state variable conditions through simulation, constructs a model library, and combines real-time traffic states, which is applicable to various traffic scenes, realizes rapid decision-making on a lane-changing starting position of the vehicle before driving off the highway, solves the problem that the lane-changing starting position of the vehicle when driving off the highway needs to rely on presetting of artificial experience, lacks scientific analysis, and is difficult to quickly adaptively adjust the lane-changing decision-making scheme according to real-time road conditions, and can provide lane-changing starting position decision-making support for a networked vehicle driving on the highway in real time, guide the lane-changing driving-off of the networked vehicle, and effectively improve the safety of vehicle driving and the stability of traffic flow on the highway. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 An exemplary flowchart of the lane-changing decision-making method for a vehicle before driving off a highway according to the present application is shown.

[0058] Figure 2 An exemplary flowchart of a construction method of a success rate evaluation model according to the present application is shown.

[0059] Figure 3 An exemplary schematic diagram of a highway vehicle driving behavior model according to the present embodiment is shown.

[0060] Figure 4 An exemplary schematic diagram of a lane-changing decision-making system for a vehicle before driving off a highway according to the present application is shown.

[0061] Figure 5 An exemplary schematic diagram of another lane-changing decision-making system for a vehicle before driving off a highway according to the present application is shown. DETAILED DESCRIPTION

[0062] The present application will be further described in detail below in combination with the drawings and embodiments.

[0063] Embodiment 1

[0064] The present embodiment provides a lane-changing decision-making method for a vehicle before driving off a highway, which is used to determine a lane-changing starting position of the vehicle before driving off the highway.

[0065] Figure 1 An exemplary flowchart of the lane-changing decision-making method for a vehicle before driving off a highway according to the present embodiment is shown.

[0066] As Figure 1 shown, the lane-changing decision-making method for a vehicle before driving off a highway of the present embodiment includes steps S101-S102.

[0067] In step S101, according to the road condition to be decided, the state parameters in the road to be decided are determined, the traffic flow state of the road to be decided is acquired in real time, the target success rate evaluation model is selected from the pre-constructed model library in combination with the preset success rate and confidence of the target vehicle successfully driving off the exit ramp under the current state parameters. The state parameters can be used to measure the objective environmental state that affects the lane change of the vehicle, and the state parameters can include a plurality of parameter combinations. Under different parameter combinations, the driving behavior of the vehicle is different, and the success rate and confidence of the target vehicle successfully driving off the exit ramp are different. By acquiring the real-time traffic flow state, in combination with the preset success rate and confidence of the target vehicle successfully driving off the exit ramp under the current state parameters, the success rate evaluation model that matches the current state parameters, traffic flow state and success rate expectation can be selected in the model library, and the accuracy of the lane change decision is improved.

[0068] In step S102, the target success rate evaluation model is used to predict the lane change success rate of the target vehicle driving from different starting positions to the exit ramp to leave the highway in the road to be decided, and the starting position of the target vehicle before changing lanes to leave the highway is determined. For example, the lane change success rates of starting positions A, B and C are predicted to be 90%, 80% and 70% respectively, and the starting position A with the highest lane change success rate can be selected as the starting position of the lane change decision.

[0069] The model library can be pre-constructed. By simulating the driving state of the vehicle under various state parameter combinations and performing simulation experiments, the model library of the evaluation model that can predict the lane change success rate under various traffic conditions is constructed to realize accurate judgment of the lane change decision of the vehicle under the real-time traffic environment.

[0070] In the embodiment of the application, the specific construction method of the model library can include: presetting Q groups of state parameters, setting a highway vehicle driving behavior model for each group of state parameters, performing a plurality of simulation experiments by using the highway vehicle driving behavior model, obtaining a success rate evaluation model corresponding to the highway vehicle driving behavior model constructed according to each state parameter, and constructing the model library based on the success rate evaluation model corresponding to each state parameter. It should be understood that for Q groups of state parameters, the model library can include Q success rate evaluation models.

[0071] Further, for each group of the Q groups of state parameters, the method for determining the success rate evaluation model includes:

[0072] The multiple sets of state variables are set under selected state parameters, multiple simulation tests are performed based on the multiple sets of state variables by using the expressway vehicle driving behavior model to obtain simulation results, a termination distance probability distribution model and a termination distance classification boundary value are determined according to the simulation results of each state variable, a success rate and an uncertainty of the target vehicle successfully driving off the exit ramp are calculated, and a success rate evaluation model corresponding to the current expressway vehicle driving behavior model state parameter is determined.

[0073] Figure 2 A specific example of a construction method of a success rate evaluation model according to an example of the present application is illustrated, in which the expressway vehicle driving behavior model is constructed based on a traffic simulation software, such as SUMO software. Multiple sets of state variables of the expressway vehicle driving behavior model can be set under selected sets of state variables, multiple simulation tests are performed by using the expressway vehicle driving behavior model, simulation results of each simulation test are obtained, including a termination distance and a Boolean value of the target vehicle. A termination distance probability distribution model and a termination distance classification boundary value are determined according to all simulation results under each state variable, a success rate and an uncertainty of the target vehicle successfully driving off the exit ramp corresponding to each state variable are calculated, and a success rate evaluation model corresponding to the current expressway vehicle driving behavior model state parameter is determined.

[0074] Specifically, the following sub-steps can be included:

[0075] Step a, constructing an expressway vehicle driving behavior model based on a traffic simulation software including SUMO, and setting state parameters of the expressway vehicle driving behavior model.

[0076] The expressway vehicle driving behavior model includes an expressway main road model and a vehicle model, as shown in Figure 3 The expressway main road model is provided with multiple lanes, in which the innermost lane is a special lane and the outermost lane is an exit ramp; the vehicle model includes a target vehicle and multiple background vehicles, in which the target vehicle travels on the special lane, the background vehicles are other vehicles in the expressway main road model except the target vehicle, and the target vehicle is applied with a car-road dynamic real-time information interaction vehicle terminal, and the background vehicles are not applied with the car-road dynamic real-time information interaction vehicle terminal.

[0077] The state parameters include a target vehicle type, a total number of lanes, a maximum speed limit and a minimum speed limit of each lane on the main road.

[0078] In this embodiment, the target vehicle type includes but is not limited to a connected manual driving vehicle, a connected automatic driving vehicle, or a vehicle fleet consisting of a connected manual driving vehicle as the first vehicle and connected automatic driving vehicles as the rest of the vehicles. In the example of the present application, the total number of lanes is set to 4, and the maximum and minimum speed limits of each lane in the main road are , units are , the upstream length of the simulation section in the highway main road model is set to 4 km, the downstream length is set to 200 m, the length of the extended lane is set to 100 m, and the length of the ramp is set to 200 m. In the highway main road model, the left 1st lane is set as the target vehicle exclusive lane, and the background vehicles are not allowed to enter. The target vehicle is also not allowed to change out before leaving the highway at the lane change starting position.

[0079] The highway main road model can select the lane change model LC2013 and the lane change model DK2008 in the traffic simulation software SUMO. In this example, the default vehicle lane change model LC2013 in the traffic simulation software SUMO is selected for simulation. During the simulation process, when the target vehicle reaches the lane change starting position, the target vehicle is allowed to leave the inner exclusive lane, and the autonomous lane change module is activated at this time. The lane change decision of this model has four kinds. Since the target vehicle needs to be forced to exit from the front ramp exit, strategic lane changing is performed. The speed changes of the target vehicle and the vehicles in front and behind it are calculated in each simulation step to facilitate the successful execution of the expected lane change action.

[0080] Set state variables, including traffic flow state and target vehicle lane change starting distance; the traffic flow state includes single lane average flow and single lane average speed ; the starting distance is the longitudinal distance between the target vehicle at the most inner lane change starting position and the exit ramp guide area tip position, and the ending distance is the longitudinal distance between the target vehicle at the most outer lane change completion position and the exit ramp guide area tip position.

[0081] In this embodiment, the value range of the single lane average flow is [0, 1800] with units of vehicles / hour, the value range of the single lane average speed is [0, 120] with units of km / h, and the value range of the target vehicle lane change starting distance is [200, 3000] with units of m. The single lane average flow is divided into single lane average flow values according to the value range, and the single lane average speed is divided into one single-lane average speed value, and the target vehicle lane-changing starting distance value is uniformly divided according to a value range, and a total of target vehicle lane-changing starting distance values are formed group state variables, In the embodiment, the state variable group is .

[0082] Step b, setting a plurality of groups of state variables and a number of state variable combinations , determining the state variable of each group of state variables 、 and wherein, is the single-lane average flow, is the single-lane average speed, is the lane-changing starting distance of the target vehicle.

[0083] Step c, for the selected state variable, a preset total number of simulation experiments, for example, 100 simulation experiments, are performed using the highway vehicle driving behavior model. In each simulation experiment, the process in which the target vehicle changes lanes from the special lane to drive off the exit ramp during the driving process of all vehicles on the main road of the highway is simulated, and 100 simulation results are obtained. The simulation results include the termination distance and the driving-off result, wherein the driving-off result is represented by a Boolean value. If the target vehicle successfully drives off, the Boolean value is 1, and if the target vehicle fails to drive off, the Boolean value is 0.

[0084] Step d, according to the 100 termination distances obtained by the plurality of simulation experiments of the current state variable, a probability density estimation method is used to determine the termination distance probability distribution model .

[0085] In the embodiment, the termination distance obtained by the 100 simulation experiments , a non-parametric kernel density estimation termination distance probability distribution model is used. This method does not need to assume the probability distribution form expressed by a few parameters, but directly estimates the probability density function from the data, and can maximize the approximation of the real probability density distribution in an infinite-dimensional parameter space.

[0086] Since the kernel density estimation discretizes the infinite-dimensional parameter space into a finite dimension, that is, discretizes the infinite-dimensional parameter space into a plurality of intervals, and then determines the probability distribution values of the sample points in each interval according to the frequency of the sample points in each interval.

[0087] When the histogram distribution model is used as the kernel function, the center point and interval length of each interval are used as the hyperparameters of the kernel function to determine the numerical expression of the specific discrete dimension. If the interval length is h, the probability density of any point can be determined by the sample point frequency obtained by the interval center point, i.e.

[0088] ;

[0089] wherein, is an indicator function,

[0090] ;

[0091] wherein, is the termination distance probability distribution function; is the sample data point to be estimated; i is the serial number of the data point; is the i-th data point in the simulation data sample; is the interval length; is the number of simulation data samples; is a Boolean value.

[0092] It is verified that the histogram distribution model satisfies the normalization requirement.

[0093] Meanwhile, the Gaussian function can also be used as the kernel function, i.e.

[0094] ;

[0095] wherein, is the bandwidth parameter, usually referred to as the smoothing parameter, used to adjust the smoothness of the histogram distribution model, thereby affecting the accuracy of the probability density function.

[0096] Step e: according to the simulation results obtained by multiple simulations of the current state variable, the off-ramp result corresponding to the termination distance is taken as a Boolean label to obtain 100 termination distance samples, and the termination distance samples are used to train the preset classifier. The training classifier predicts whether the target vehicle successfully off-ramps the exit ramp according to the termination distance.

[0097] In this embodiment, the classifier is specifically a binary classification algorithm logistic regression learning classification model. In the training process, the termination distance ​The input features and success labels are used as target variables for the classifier. To improve the classifier's generalization ability, cross-validation is used to select the optimal model parameters. The input termination distance samples are divided and labeled as "successfully exited the exit ramp" and "unsuccessfully exited the exit ramp," respectively. The critical termination distance values ​​between successful and unsuccessful exits are then obtained, determining the termination distance classification boundary values. Uncertainty of classification boundaries .

[0098] Step f, based on the termination distance probability distribution model in S1.4 and the termination distance classification boundary value in S1.5. By using simple cumulative probability calculation, the success rate of the target vehicle successfully leaving the exit ramp under the current state variable is determined as follows:

[0099] ;

[0100] In the formula, For the first The success rate of the target vehicle successfully leaving the exit ramp in the simulation test; To terminate the distance probability distribution model; The total number of lanes. Lane numbering, , The maximum speed limit for the lane. The minimum speed limit for the lane; For the first The termination distance obtained from the simulation experiment For the first The average flow rate per lane in this simulation test For the first The average speed of a single lane in this simulation test. For the first The initial distance of the target vehicle in the second simulation test for lane changing. For the first A simulation test of highway vehicle driving behavior model.

[0101] The uncertainty corresponding to the success rate of the target vehicle successfully leaving the exit ramp under the current state variable is:

[0102] ;

[0103] In the formula, For the first The uncertainty of the target vehicle successfully leaving the exit ramp in the simulation test.

[0104] Step g: Determine whether the value of the current simulation count i exceeds the preset number of state variable combinations. If the preset number of state variable combinations has been exceeded If the condition is met, proceed to step h; otherwise, select the next set of state variables and return to steps c to f to continue the simulation experiment until the preset number of state variable combinations is exceeded. ,get A simulation data sample. The simulation data sample is represented as... ,in, For the first The state variables of this simulation experiment For the first Success rate of the simulation test.

[0105] When i > N, proceed to step h.

[0106] Step h: Determine the success rate evaluation model corresponding to the current highway vehicle driving behavior model state parameters. .

[0107] In this embodiment, a success rate evaluation model is constructed based on a Gaussian process regression model. The success rate evaluation model is trained using simulation data samples, with each group of state variables... As input features to the success rate assessment model, feature vectors are constructed. Success rate As the target variable output by the success rate assessment model .

[0108] During the training process of the success rate evaluation model, the state variables of the simulation data samples are input into the success rate evaluation model. The success rate evaluation model calculates the estimated success rate based on the state variables. Combined with the corresponding success rate of the simulation data sample input, the accuracy of the success rate evaluation model is obtained. The accuracy of the success rate evaluation model is compared with the preset precision. If the accuracy of the success rate evaluation model does not reach the preset precision, the hyperparameters of the success rate evaluation model are optimized and the success rate evaluation model is trained again. Otherwise, the trained success rate evaluation model is obtained.

[0109] When the success rate assessment model is constructed using a Gaussian process regression model, the expression is:

[0110] ;

[0111] in, To achieve a successful first-order distribution; It is a mean function, which we assume to be the zero function, i.e. ; Here, a kernel function is used to describe the correlation between input sample points, where... , are the same or different feature vectors.

[0112] The radial basis kernel function is used as the kernel function of the Gaussian process regression model, and the kernel function is:

[0113] ;

[0114] In the formula, ; is the variance of the signal; is a length scale parameter for controlling the smoothness of the radial basis kernel function.

[0115] In order to improve the fitting effect of the success rate evaluation model, the hyperparameters of the success rate evaluation model are optimized by using maximum likelihood estimation or Bayesian optimization. In the embodiment, the hyperparameters of the success rate evaluation model are optimized by using a log-likelihood function, and the log-likelihood function is:

[0116] ;

[0117] In the formula, is a likelihood function; is a set of hyperparameters to be optimized; is a transpose matrix; is a covariance matrix calculated by a function; is the variance of the noise.

[0118] The log-likelihood function is optimized based on the gradient descent method, and the goal is to maximize the log-likelihood function to obtain the best hyperparameter value. When the success rate evaluation model is trained using simulation data samples, the success rate evaluation model can not only predict the success rate through learning, but also estimate the uncertainty of the success rate prediction value based on the correlation between the sample data. The optimized success rate evaluation model can predict the mean and the variance of the new input feature as follows:

[0119] ;

[0120] ;

[0121] In the formula, is an identity matrix, and k is a covariance matrix calculated by substituting the known simulation sample into the kernel function.

[0122] In an embodiment of the present application, according to the real-time acquired road single-lane average flow and single-lane average speed , the single-lane average flow​ and single lane average speed The success rate evaluation model corresponding to the highway vehicle driving behavior model is input, and a one-dimensional slice corresponding to the success rate evaluation model is used According to the 2sigma principle or the 3sigma principle, the lower limit of the reliable range of the one-dimensional slice is obtained in combination with the preset confidence requirement When the 2sigma principle is selected, the confidence is selected as 95%, that is When the 3sigma principle is selected, the confidence is selected as 99%, that is .

[0123] Thus, the starting distance satisfying is determined , that is, the final decision output starting distance result, that is, the best starting position.

[0124] Embodiment 2

[0125] The embodiments of the application also provide a vehicle lane changing decision system before driving off the highway.

[0126] Figure 4 An exemplary schematic diagram of a vehicle lane changing decision system before driving off the highway according to the application is shown.

[0127] As shown in Figure 4 , the vehicle lane changing decision system 400 before driving off the highway includes a model selection module 401, an output module 402 and a model library component module 403.

[0128] The model selection module 401 is configured to determine the state parameters in the road to be decided according to the road condition to be decided, to obtain the traffic flow state of the road to be decided in real time, to select the target success rate evaluation model from the pre-constructed model library in combination with the preset success rate of the target vehicle successfully driving off the exit ramp and the confidence under the current state parameters.

[0129] The output module 402 is configured to predict the starting distance and the ending distance of the target vehicle successfully driving off the exit ramp in the road to be decided by using the selected target success rate evaluation model, and to determine the starting position of the target vehicle lane changing before driving off the highway.

[0130] The model library component module 403 is configured to: construct a highway vehicle driving behavior model, preset a Q group of state parameters, set state variables for each group of state parameters, perform multiple simulation tests using the highway vehicle driving behavior model, obtain a success rate evaluation model corresponding to the highway vehicle driving behavior model constructed according to each state parameter, and construct the model library based on the success rate evaluation model corresponding to each state parameter; wherein for each group of the Q group of state parameters, the determination method of the success rate evaluation model includes: setting multiple groups of state variables under the selected state parameter, performing multiple simulation tests using the highway vehicle driving behavior model based on the multiple groups of state variables to obtain simulation results, determining a termination distance probability distribution model and a termination distance classification boundary value according to the simulation results of each state variable, calculating the success rate and the uncertainty of the target vehicle successfully driving off the exit ramp, and determining the success rate evaluation model corresponding to the state parameter of the current highway vehicle driving behavior model.

[0131] The model selection module 401 is in communication connection with the model library component module 403.

[0132] Embodiment 3

[0133] Another vehicle lane-changing decision system before driving off the highway is proposed in this embodiment, which is used to implement the vehicle lane-changing decision method before driving off the highway described in embodiment 1.

[0134] Figure 5 An exemplary schematic diagram of another vehicle lane-changing decision system before driving off the highway according to the present application is shown.

[0135] As Figure 5 described, the another vehicle lane-changing decision system before driving off the highway includes a simulation module, a termination distance prediction module, a success rate estimation module, a success rate evaluation model establishment module, a road state real-time acquisition module, and a decision module.

[0136] The simulation module is internally preset with a highway vehicle driving behavior model, which is used to simulate the process of the target vehicle driving off the exit ramp using the highway vehicle driving behavior model, and to obtain and store the simulation results under different state parameters and state variables.

[0137] The simulation module is connected with the termination distance prediction module and the success rate estimation module, respectively, and is used to transmit the simulation results to the termination distance prediction module and the success rate estimation module.

[0138] The termination distance prediction module is connected with the success rate estimation model, and is used to determine the termination distance probability distribution model according to the simulation results obtained from the simulation module.

[0139] The success rate estimation module is connected with the success rate evaluation model establishment module, and is configured to determine the success rate and uncertainty of the target vehicle successfully driving off the exit ramp according to the simulation result obtained from the simulation module and the termination distance probability distribution model determined by the termination distance prediction module.

[0140] The success rate evaluation model establishment module is connected with the decision module, and is configured to train the success rate evaluation model by using the simulation result and the success rate corresponding to the simulation result, optimize the success rate evaluation model corresponding to the highway vehicle driving behavior model constructed by using each state parameter, and construct a model library of the success rate evaluation model.

[0141] The road state real-time acquisition module is connected with the success rate evaluation model establishment module, and is configured to acquire the state parameters of the highway to be decided in real time, and transmit the type of the target vehicle, the maximum speed limit and the minimum speed limit of each lane of the main road of the highway to be decided to the decision module.

[0142] The decision module is configured to make a decision scheme for the lane-changing start position before the vehicle drives off the highway, select a suitable highway vehicle driving behavior model in the model library according to the state parameters of the highway to be decided transmitted by the road state real-time acquisition module and in combination with the preset requirements of the success rate and uncertainty of the target vehicle successfully driving off the exit ramp, predict the start distance and the termination distance of the target vehicle successfully driving off the exit ramp in the highway to be decided by using the highway vehicle driving behavior model, and determine the lane-changing start position of the target vehicle before driving off the highway.

[0143] Embodiment 4

[0144] In this embodiment, a computer readable storage medium is provided, and computer program instructions are stored on the computer readable storage medium. The computer program instructions are executed by the vehicle lane-changing decision system before driving off the highway described in Embodiment 2 or 3, and are configured to implement the specific steps of the vehicle lane-changing decision method before driving off the highway described in Embodiment 1.

[0145] Of course, the above description is not a limitation of the present application, and the present application is not limited to the above examples. Changes, modifications, additions or replacements made by those skilled in the art within the essential scope of the present application should also be within the protection scope of the present application.

Claims

1. A method of making a decision on lane change before a vehicle exits a highway, characterized by, A method for determining a starting position of a lane change before a vehicle exits a highway, comprising the following steps: According to the road condition to be decided, the state parameters in the road to be decided are determined, the traffic flow state of the road to be decided is obtained in real time, the target success rate evaluation model is selected from the pre-constructed model library combined with the preset success rate and confidence of the target vehicle successfully leaving the exit ramp under the current state parameters; The success rate of the target vehicle starting from different starting positions in the road to be decided to the exit ramp to exit the highway is predicted by using the selected target success rate evaluation model, and the starting position of the lane change before the target vehicle exits the highway is determined; The method for constructing the model library comprises: A highway vehicle driving behavior model is constructed, Q sets of state parameters are preset, state variables are set for each set of state parameters, and a plurality of simulation tests are performed by using the highway vehicle driving behavior model to obtain a success rate evaluation model corresponding to the highway vehicle driving behavior model under each state parameter, and the model library is constructed based on the success rate evaluation model corresponding to each state parameter; For each set of state parameters in the Q sets of state parameters, the method for determining the success rate evaluation model comprises: A plurality of sets of state variables are set under the selected state parameters, a plurality of simulation tests are performed based on the plurality of sets of state variables by using the highway vehicle driving behavior model to obtain simulation results, a termination distance probability distribution model and a termination distance classification boundary value are determined according to the simulation results of each state variable, a success rate and an uncertainty of the target vehicle successfully leaving the exit ramp are calculated, and a success rate evaluation model corresponding to the current state parameter of the highway vehicle driving behavior model is determined.

2. The method of claim 1, wherein The highway vehicle driving behavior model is provided with a highway main road model and a vehicle model; the highway main road model is provided with a plurality of lanes, wherein the innermost lane is a special lane and the outermost lane is an exit ramp; the vehicle model comprises a target vehicle and a plurality of background vehicles, wherein the target vehicle travels on the special lane, the background vehicles are other vehicles in the highway main road model except the target vehicle, and a car-road dynamic real-time information interaction vehicle terminal is applied to the target vehicle and no car-road dynamic real-time information interaction vehicle terminal is applied to the background vehicles; The state variable includes target vehicle type, total number of lanes , maximum and minimum speed limits of each lane in the main road The state variables include a traffic flow state and a target vehicle lane-changing start distance; the traffic flow state includes a single-lane average flow in a simulation process and a single-lane average speed ; the start distance is a longitudinal distance of the target vehicle from a position at which lane changing is started in an innermost lane to a position at a tip of a ramp guide area, and the end distance is a longitudinal distance of the target vehicle from a position at which lane changing is completed in an outermost lane to the position at the tip of the ramp guide area.

3. The method of claim 2, wherein For each set of state parameters in the Q sets of state parameters, the determination method of the success rate evaluation model comprises: Step a, constructing a highway vehicle driving behavior model, and setting a state parameter of the highway vehicle driving behavior model; Step b, set multiple groups of state variables and the number of state variable combinations , determine the state variable of each group of state variables 、 and , wherein, is the average flow of a single lane, is the average speed of a single lane, is the lane-changing starting distance of the target vehicle; Step c, using the highway vehicle driving behavior model for the selected state variable a secondary simulation test, in each simulation test process, the process of the target vehicle changing lanes from the special lane to the exit ramp is simulated by simulating the driving process of all vehicles on the main road of the highway, and a simulation result is obtained simulation results, the simulation results include a termination distance and a departure result, wherein the departure result is represented by a Boolean value, and the Boolean value is 1 if the target vehicle departs successfully, and the Boolean value is 0 if the target vehicle departs unsuccessfully; Step d, according to the current state variable multiple simulation test obtained by a termination distance, the probability density estimation method is used to determine the termination distance probability distribution model ; Step e, according to the simulation results of the current state variable multiple simulation test, the off result corresponding to the termination distance is taken as a Boolean value label, and a termination distance sample is obtained , a preset classifier is trained by using the termination distance sample, the classifier is trained to predict whether the target vehicle successfully drives off the exit ramp according to the termination distance, a termination distance critical value between the successful exit ramp and the unsuccessful exit ramp is obtained, and the termination distance classification boundary value and the uncertainty of the classification boundary ; Step f, determine the success rate and uncertainty of the target vehicle successfully driving off the exit ramp under the current state variable according to the termination distance probability distribution model in S1.4 and the termination distance classification boundary value in S1.5 , determine the success rate and uncertainty of the target vehicle successfully driving off the exit ramp under the current state variable; Step g, judging whether the current simulation number exceeds the preset state variable combination number , if the preset state variable combination number is exceeded , then entering step h, otherwise, selecting the next group of state variables and returning to steps c-f to continue the simulation test until the preset state variable combination number is exceeded , obtaining simulation data samples, which are expressed as , wherein, is the state variable of the simulation test, is the success rate of the simulation test, and entering step h; Step h, determining a success rate evaluation model corresponding to the current highway vehicle driving behavior model state parameter ; wherein, is the total number of lanes, is the lane number, wherein, ; is the maximum speed limit of the lane, is the minimum speed limit of the lane, is the highway vehicle driving behavior model of the nth simulation test.

4. The method of claim 3, wherein In step f, the success rate of the target vehicle successfully leaving the exit ramp under the current state parameter is: ； In the formula, For the first The success rate of the target vehicle successfully leaving the exit ramp in the simulation test; To terminate the distance probability distribution model; The total number of lanes. Lane numbering, where ; The maximum speed limit for the lane. The minimum speed limit for the lane; For the first The termination distance obtained from the simulation experiment For the first The average flow rate per lane in this simulation test For the first The average speed of a single lane in this simulation test. For the first The initial distance of the target vehicle in the second simulation test for lane changing. For the first A simulation test of highway vehicle driving behavior model; The uncertainty corresponding to the success rate of the target vehicle successfully leaving the exit ramp under the current state variable is: ； In the formula, is the uncertainty of the target vehicle successfully driving off the exit ramp in the nth simulation test.

5. The method of claim 4, wherein, In step h, the success rate evaluation model is constructed based on a Gaussian process regression model, wherein the success rate evaluation model is trained by using simulation data samples, the state variable is used as the input feature of the success rate evaluation model, and the success rate is used as the target variable output by the success rate evaluation model. In the success rate evaluation model training process, the state variable of the simulation data sample is input into the success rate evaluation model, the success rate evaluation model is used to calculate the success rate estimate value according to the state variable, the accuracy rate calculated by the success rate evaluation model is obtained in combination with the success rate corresponding to the simulation data sample, the accuracy rate calculated by the success rate evaluation model is compared with the preset precision, if the accuracy rate calculated by the success rate evaluation model does not reach the preset precision, the hyperparameters of the success rate evaluation model are optimized, the success rate evaluation model is continuously trained, otherwise, the trained success rate evaluation model is obtained.

6. The method of claim 5, wherein The success rate evaluation model is: ； wherein, Kernel function is: ； wherein , are state variables in the simulated data samples, is a success rate prior distribution; is a mean function; is a kernel function; is a variance of the signal; is a length scale parameter for controlling the smoothness of the radial basis kernel function; The hyperparameters of the success rate evaluation model are optimized by using a log-likelihood function, and the log-likelihood function is: ； wherein is a likelihood function; is a set of hyperparameters to be optimized; is a transpose matrix; is a covariance matrix calculated by a kernel function; is a variance of noise, y is a success rate sample; Based on gradient descent method to optimize the log-likelihood function, the optimized new input features , the predicted mean and variance of the new input features are: ； ； In the formula, is the identity matrix.

7. A system for lane change decision before exiting a highway for a vehicle, characterized by, The system comprises a model selection module, an output module and a model library component module. The model selection module is configured to determine the state variable in the road to be decided according to the road condition to be decided, to obtain the traffic flow state of the road to be decided in real time, to select the target success rate evaluation model from the pre-constructed model library in combination with the preset success rate and confidence of the target vehicle successfully driving off the exit ramp under the current state variable, The output module is configured to predict the starting distance and ending distance of the target vehicle successfully driving off the exit ramp in the road to be decided by using the selected target success rate evaluation model, and to determine the starting position of the target vehicle changing lanes before driving out of the highway, The model selection module and the model library component module are communicatively connected, and the model library component module is configured to: construct a highway vehicle driving behavior model, preset Q groups of state variables, set state variables for each group of state variables, perform multiple simulation tests by using the highway vehicle driving behavior model, obtain the success rate evaluation model corresponding to the highway vehicle driving behavior model constructed according to each state variable, and construct the model library based on the success rate evaluation model corresponding to each state variable. For each group of the Q groups of state variables, the determination method of the success rate evaluation model comprises: a plurality of groups of state variables are set under the selected state variable, simulation results are obtained by using the highway vehicle driving behavior model based on the plurality of groups of state variables, a termination distance probability distribution model and a termination distance classification boundary value are determined according to the simulation results of each state variable, the success rate and the uncertainty of the target vehicle successfully driving off the exit ramp are calculated, and the success rate evaluation model corresponding to the current highway vehicle driving behavior model state variable is determined.

8. A computer-readable storage medium having stored thereon computer program instructions, wherein, The program instructions are executed by the vehicle lane changing decision-making system before driving out of the highway to realize the steps of the method in any one of claims 1-6.

Images