Autonomous lane changing method and system for vehicle, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a vehicle autonomous lane changing method and system, electronic equipment and a storage medium, and belongs to the technical field of vehicle intelligent control. According to the method, driving styles of a vehicle and surrounding environment vehicles are considered, driving styles of a target vehicle owner and surrounding vehicle owners are implicitly represented by historical track data of a target vehicle and the environment vehicles through a hidden layer of a lane changing decision model, and lane changing decision index characteristics influencing lane changing behaviors are considered at the same time, so that the driving styles of the target vehicle owner and the surrounding vehicle owners are represented. The driving style features and the lane changing decision index features are input into the full connection layer of the lane changing decision model, the full connection layer combines input data to carry out lane changing decision, the lane selection accuracy can be remarkably improved, and meanwhile compared with a driving style explicit definition lane selection mode, the mode of embedding the driving style hidden features has the advantage that the lane selection efficiency is improved. Personalized driving styles can be reflected, the flexibility of autonomous lane changing decision making is improved, and the intelligent level of automatic driving is improved.

Description

Technical Field

[0001] The present application relates to the field of vehicle intelligent control technology, and in particular to a vehicle autonomous lane changing method, system, electronic device and storage medium. Background Art

[0002] The research on decision-making strategies of autonomous vehicles is one of the key research directions in the current field of autonomous driving. Intelligent traffic decision-making systems can make real-time judgments on current driving scenarios and complete driving tasks more safely and efficiently.

[0003] At present, the decision-making methods of autonomous driving mainly include rule-based and learning-based methods. Among them, the rule-based method is to formulate strategies based on mathematical models and current scenarios. It is usually simple in structure and highly interpretable. However, the decision-making strategies formulated by rule-based methods lack flexibility and are usually unable to handle complex and changing environments. At the same time, they may not be able to cope with emergencies outside the scope of rule design. The learning-based method to formulate autonomous driving decision-making strategies refers to the use of data and learning algorithm methods to train models to obtain appropriate decision-making strategies, but it usually has the characteristics of poor interpretability and high data dependence. In the process of autonomous lane changing of vehicles, general indicators are often used to make decisions on lane changing decisions, while human drivers deal with sudden lane changing situations based on their driving experience and driving preferences. Decisions made based on concrete indicators alone cannot truly restore human driving intentions, resulting in the inability of autonomous lane changing of vehicles to imitate human driving intentions and difficulty in realizing intelligent autonomous driving of vehicles. Summary of the invention

[0004] The main purpose of the embodiments of the present application is to propose a vehicle autonomous lane changing method, system, electronic device and storage medium, aiming to simulate the user's personalized driving experience and driving preferences to make autonomous lane changing decisions and improve the intelligence level of autonomous driving.

[0005] To achieve the above objective, an embodiment of the present application provides a method for autonomous lane changing of a vehicle, comprising the following steps:

[0006] Acquire first historical trajectory data of the target vehicle and second historical trajectory data of the environment vehicle;

[0007] Determining, through a hidden layer of the lane change decision model, a first driving style feature of the target vehicle and a second driving style feature of the environment vehicle according to the first historical trajectory data and the second historical trajectory data;

[0008] Determining a lane-changing decision indicator feature that affects the lane-changing behavior according to the first historical trajectory data;

[0009] The first driving style feature, the second driving style feature, the lane change decision index feature and the current driving data are input into a fully connected layer of a lane change decision model to obtain a lane change decision result, wherein the lane change decision result is used to determine whether to change lanes and the lane change direction.

[0010] In some embodiments, determining the first driving style feature of the target vehicle and the second driving style feature of the environment vehicle through the hidden layer of the lane change decision model according to the first historical trajectory data and the second historical trajectory data includes the following steps:

[0011] Constructing a trajectory feature matrix of a target vehicle and a trajectory feature matrix of an environment vehicle respectively according to the first historical trajectory data and the second historical trajectory data;

[0012] The trajectory feature matrix of the target vehicle and the trajectory feature matrix of the environment vehicle are respectively input into the hidden layer of the lane change decision model to obtain a first driving style feature of the target vehicle and a second driving style feature of the environment vehicle, wherein the hidden layer includes a convolutional neural network and a bidirectional gated recurrent unit connected in sequence.

[0013] In some embodiments, the trajectory feature matrix of the target vehicle or the environment vehicle is obtained by the following steps:

[0014] Segmenting the vehicle trajectory according to a preset time step to obtain a plurality of trajectory segments, wherein the vehicle trajectory is the first historical trajectory data or the second historical trajectory data;

[0015] Calculating multiple variable parameters of each trajectory segment to obtain a feature vector of the trajectory segment, wherein the multiple parameter variables include lateral speed, longitudinal speed, total speed, lateral acceleration, longitudinal acceleration, total acceleration, time interval and driving distance;

[0016] The feature vectors of a plurality of trajectory segments of the vehicle trajectory are combined to obtain a trajectory feature matrix of the vehicle trajectory.

[0017] In some embodiments, determining the lane changing decision indicator characteristics affecting the lane changing behavior according to the first historical trajectory data includes the following steps:

[0018] determining a lane change decision point in the first historical trajectory data by analyzing fluctuation characteristics of the first historical trajectory data;

[0019] The lane-changing decision index characteristics of the target vehicle are determined according to the historical driving data of the lane-changing decision point.

[0020] In some embodiments, the vehicle autonomous lane changing method further includes the following steps:

[0021] When the lane-changing decision result represents a lane-changing, determining a candidate trajectory cluster according to the lane-changing direction in the lane-changing decision result, wherein the candidate trajectory cluster includes a plurality of candidate lane-changing trajectories;

[0022] Based on the expected lane-changing driving index, extracting the first trajectory feature of each of the candidate lane-changing trajectories and the second trajectory feature of the actual lane-changing trajectory, and determining the actual trajectory data set according to the second trajectory features of all the actual lane-changing trajectories;

[0023] A target lane-changing trajectory is determined according to the first trajectory feature and the real trajectory data set by using a maximum entropy inverse reinforcement learning algorithm.

[0024] In some embodiments, determining a candidate trajectory cluster according to the lane changing direction in the lane changing decision result comprises the following steps:

[0025] Determine the target lane according to the lane change direction, and construct a time-continuous function of the trajectory with respect to the lateral direction and the longitudinal direction, wherein the time-continuous function is a polynomial function;

[0026] Determining a lateral trajectory terminal interval according to the lateral area of ​​the target lane, and determining a longitudinal trajectory terminal interval according to the longitudinal lane change distance configuration;

[0027] Determining the boundary conditions of the time continuous function according to the lateral trajectory terminal interval, the longitudinal trajectory terminal interval and the current state data of the target vehicle;

[0028] A polynomial trajectory planning algorithm is used to determine a candidate trajectory cluster according to the time continuous function and the boundary conditions.

[0029] In some embodiments, the determining the target lane change trajectory according to the first trajectory feature and the real trajectory data set by using the maximum entropy inverse reinforcement learning algorithm includes the following steps:

[0030] Taking maximizing the objective function as the optimization goal, determining a feature weight vector of each candidate lane changing trajectory according to the first trajectory feature of each candidate lane changing trajectory and the real trajectory data set;

[0031] Determining a reward function of the candidate lane-changing trajectory according to a feature weight vector of the candidate lane-changing trajectory;

[0032] Determine a target lane-changing trajectory by selecting according to the reward function of each of the candidate lane-changing trajectories;

[0033] The optimization objective is expressed as follows:

[0034]

[0035] in, represents the objective function, represents the real trajectory dataset, P(ζ|θ) represents the probability of the candidate lane-changing trajectory ζ, and θ represents the feature weight vector to be optimized.

[0036] To achieve the above objectives, another aspect of the embodiment of the present application provides a vehicle autonomous lane changing system, comprising:

[0037] The first module is used to obtain the first historical trajectory data of the target vehicle and the second historical trajectory data of the environment vehicle;

[0038] A second module is used to determine a first driving style feature of the target vehicle and a second driving style feature of the environment vehicle according to the first historical trajectory data and the second historical trajectory data through a hidden layer of the lane change decision model;

[0039] A third module is used to determine a lane-changing decision indicator feature that affects the lane-changing behavior according to the first historical trajectory data;

[0040] The fourth module is used to input the first driving style feature, the second driving style feature, the lane change decision index feature and the current driving data into the fully connected layer of the lane change decision model to obtain a lane change decision result, wherein the lane change decision result is used to determine whether to change lanes and the lane change direction.

[0041] To achieve the above-mentioned purpose, another aspect of an embodiment of the present application proposes an electronic device, which includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory, and the program implements the method described in the above embodiment when executed by the processor.

[0042] To achieve the above-mentioned purpose, another aspect of an embodiment of the present application proposes a storage medium, which is a computer-readable storage medium used for computer-readable storage, and the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the method described in the above embodiment.

[0043] The vehicle autonomous lane changing method, system, electronic device and storage medium proposed in the present application take into account the driving styles of the vehicle itself and the surrounding vehicles, and implicitly represent the driving styles of the target vehicle owner and the surrounding vehicle owners through the hidden layer of the lane changing decision model through the historical trajectory data of the target vehicle and the surrounding vehicles. At the same time, the lane changing decision indicator characteristics that affect the lane changing behavior are considered, and the driving style characteristics and the lane changing decision indicator characteristics are input into the fully connected layer of the lane changing decision model. The fully connected layer combines the input data to make lane changing decisions, which can significantly improve the lane selection accuracy. At the same time, compared with the lane selection method that explicitly defines the driving style, the method of embedding the hidden characteristics of the driving style can reflect the personalized driving style, improve the flexibility of autonomous lane changing decisions, and improve the intelligence level of autonomous driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a flow chart of a vehicle autonomous lane changing method provided in an embodiment of the present application;

[0045] Figure 2 is a schematic diagram of a lane change decision point provided in an embodiment of the present application;

[0046] Figure 3 is a schematic diagram of a lane change decision process provided by an embodiment of the present application;

[0047] Figure 4 is a schematic diagram of the lane change trajectory planning process provided by an embodiment of the present application;

[0048] Figure 5 is a schematic diagram of a lane change trajectory cluster provided in an embodiment of the present application;

[0049] Figure 6 It is a schematic diagram of the application architecture of the vehicle autonomous lane changing method provided in an embodiment of the present application;

[0050] Figure 7 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0052] It should be noted that, although the functional modules are divided in the system and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the system or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0054] The vehicle autonomous lane changing method, system, electronic device and storage medium provided in the embodiments of the present application are specifically described through the following embodiments. First, the vehicle autonomous lane changing method in the embodiments of the present application is described.

[0055] The vehicle autonomous lane changing method provided in the embodiment of the present application relates to the field of vehicle intelligent control technology. The vehicle autonomous lane changing method provided in the embodiment of the present application can be applied to a terminal, can also be applied to a server side, and can also be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a vehicle-mounted terminal, etc.; the server side can be configured as an independent physical server, or a server cluster or a distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the vehicle autonomous lane changing method, etc., but is not limited to the above forms.

[0056] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0057] Figure 1 is an optional flowchart of the vehicle autonomous lane changing method provided in an embodiment of the present application. Figure 1 The method may include but is not limited to steps S101 to S104.

[0058] Step S101, obtaining first historical trajectory data of the target vehicle and second historical trajectory data of the environment vehicle;

[0059] Step S102, determining a first driving style feature of the target vehicle and a second driving style feature of the environment vehicle according to the first historical trajectory data and the second historical trajectory data through a hidden layer of the lane change decision model;

[0060] Step S103, determining a lane-changing decision indicator feature that affects the lane-changing behavior according to the first historical trajectory data;

[0061] Step S104, inputting the first driving style feature, the second driving style feature, the lane change decision index feature and the current driving data into the fully connected layer of the lane change decision model to obtain a lane change decision result, wherein the lane change decision result is used to determine whether to change lanes and the lane change direction.

[0062] In steps S101 to S104 shown in the embodiment of the present application, the driving styles of the vehicle itself and the surrounding vehicles are taken into consideration, and the historical trajectory data of the target vehicle and the surrounding vehicles are passed through the hidden layer of the lane change decision model to implicitly represent the driving styles of the target vehicle owner and the surrounding vehicle owners. At the same time, the lane change decision indicator characteristics that affect the lane change behavior are considered, and the driving style characteristics and the lane change decision indicator characteristics are input into the fully connected layer of the lane change decision model. The fully connected layer combines the input data to make lane change decisions, which can significantly improve the lane selection accuracy. At the same time, compared with the lane selection method that explicitly defines the driving style, the method of embedding the hidden characteristics of the driving style can reflect the personalized driving style, improve the flexibility of autonomous lane change decisions, and improve the intelligence level of autonomous driving.

[0063] In step S101 of some embodiments, the target vehicle is the object of autonomous lane change decision and planning, and the environmental vehicles are the surrounding vehicles of the target vehicle, which can be the adjacent vehicles of the target vehicle or the vehicles within a certain area. The first historical trajectory data refers to the driving data of the historical complete trajectory of the target vehicle, including the speed, acceleration, steering angle and environmental data (surrounding traffic conditions, road conditions, etc.) of each trajectory point. The second historical trajectory data refers to the driving data of the historical complete trajectory of the environmental vehicle, including the speed, acceleration, steering angle and environmental data (surrounding traffic conditions, road conditions, etc.) of each trajectory point.

[0064] In step S102 of some embodiments, the lane change decision model of this embodiment mainly includes an input layer, a hidden layer and a fully connected layer, and the hidden layer of the lane change decision model is a neural network based on a combination of a convolutional neural network and a bidirectional gated recurrent unit, which is used to extract driving style features implicit in trajectory data. The lane change decision model of this embodiment is obtained by training a large amount of sample data, and the sample data includes trajectory data of a specified vehicle and its surrounding vehicles, lane change decision index features of the specified vehicle, and real lane change decisions.

[0065] In some embodiments, step S102 may include but is not limited to steps S201 to S202:

[0066] Step S201, constructing a trajectory feature matrix of a target vehicle and a trajectory feature matrix of an environment vehicle respectively according to the first historical trajectory data and the second historical trajectory data;

[0067] Step S202, respectively inputting the trajectory feature matrix of the target vehicle and the trajectory feature matrix of the environment vehicle into the hidden layer of the lane change decision model to obtain a first driving style feature of the target vehicle and a second driving style feature of the environment vehicle, wherein the hidden layer includes a convolutional neural network and a bidirectional gated recurrent unit connected in sequence.

[0068] In some embodiments of step S201 , the first historical trajectory data and the second historical trajectory data need to be processed into an input format that conforms to the hidden layer, that is, the trajectory data are processed into a trajectory feature matrix.

[0069] Specifically, the trajectory feature matrix of the target vehicle or the environment vehicle is obtained by the following steps:

[0070] Step S301, segmenting the vehicle trajectory according to a preset time step to obtain a plurality of trajectory segments, wherein the vehicle trajectory is the first historical trajectory data or the second historical trajectory data;

[0071] Step S302, calculating multiple variable parameters of each trajectory segment to obtain a feature vector of the trajectory segment, wherein the multiple parameter variables include lateral speed, longitudinal speed, total speed, lateral acceleration, longitudinal acceleration, total acceleration, time interval and driving distance;

[0072] Step S303 , combining the feature vectors of the plurality of trajectory segments of the vehicle trajectory to obtain a trajectory feature matrix of the vehicle trajectory.

[0073] For example, the vehicle trajectory of duration T (ie, the first historical trajectory data or the second historical trajectory data) is divided into n time steps T = {t1, t2, ..., t n} and discretize each trajectory segment based on the driving behavior characteristic variables of the vehicle trajectory.

[0074] The variables used to describe mainly include lateral velocity, longitudinal velocity, velocity, lateral acceleration, longitudinal acceleration, acceleration, time interval, collision time and distance, which are specifically expressed as follows:

[0075] V={V1,V2,...,V n};

[0076] Therefore, a vehicle trajectory can be represented as a 9×n trajectory feature matrix B k , specifically expressed as follows:

[0077]

[0078] Among them, b ij represents the time step t i The recorded characteristic variable V j The value of .

[0079] In some embodiments of step S201, after obtaining the trajectory feature matrix of the target vehicle and the trajectory feature matrix of the environment vehicle, a neural network composed of a convolutional neural network and a bidirectional gated recurrent unit is used to extract the corresponding driving style features of the target vehicle and the environment vehicle from the trajectory feature matrix, wherein the convolutional neural network is used to extract the local features of the driving style in the trajectory data through its convolution layer, and the bidirectional gated recurrent unit is used to mine the temporal dependency of the driving style features. In this embodiment, considering that it is difficult to obtain personal information of the driver such as driving age and gender, the sensitivity of the neural network combined with the convolutional neural network and the bidirectional gated recurrent unit to the time series is utilized to extract the features of the vehicle historical trajectory data described by the multivariate time series, so as to realize the implicit representation of the driving style of the target vehicle and the environment vehicle.

[0080] In step S103 of some embodiments, in the lane change decision process, the driver is not only influenced by his own driving experience and driving habits, but also mainly influenced by the current driving scenario to choose the best lane, which is mainly to obtain a higher speed, vehicle distance, etc., or to avoid accidents. The main traffic factors affecting lane selection are summarized as safety, necessity and profitability. The lane change decision indicators that affect lane change behavior may include but are not limited to the longitudinal distance between the target vehicle and the environment vehicle, lane change safety, speed gain, lane change necessity, etc. The index value of each index is calculated based on the first historical trajectory data of the target vehicle, so as to determine the personalized lane change decision index characteristics of the target vehicle.

[0081] Specifically, step S103 may include but is not limited to steps S401 to S402:

[0082] Step S401, determining a lane change decision point in the first historical trajectory data by analyzing the fluctuation characteristics of the first historical trajectory data;

[0083] Step S402, determining a lane-changing decision index characteristic of the target vehicle according to historical driving data of the lane-changing decision point.

[0084] In this embodiment, please refer to Figure 2 By analyzing the fluctuation characteristics of the vehicle trajectory, the lane change start and end points are automatically identified based on the historical trajectory data of the vehicle, and the lane change trajectory segments are automatically extracted. At the same time, the lane change decision point is calculated based on the extracted lane change trajectory segments. The vehicle trajectory data is subjected to discrete wavelet transform by wavelet transform to obtain detail coefficients of multiple scales. Since the human reaction time is 0.7s to 0.9s, this embodiment can select 1s as the reaction time before the human driver makes a decision, that is, the trajectory point corresponding to 1s before the lane change start point is used as the lane change decision point.

[0085] The index values ​​of each lane changing decision index of the target vehicle are calculated based on the historical driving data (speed, acceleration, speed of the surrounding vehicle, etc.) of the lane changing decision point, and the lane changing decision index characteristics of the target vehicle are determined based on the index values ​​of the lane changing decision index. The lane changing decision index characteristics characterize the importance of the lane changing decision index on the lane changing decision of the target vehicle. There is a certain mapping relationship between the index value of the lane changing decision index and the lane changing decision index characteristics. For example, if the calculated longitudinal distance index value at the lane changing decision point is smaller, it indicates that the driver is not sensitive to the longitudinal distance when making a lane changing decision, then the lane changing decision index characteristics of the longitudinal distance are smaller, that is, the longitudinal distance has little effect on the lane changing decision.

[0086] This embodiment improves the accuracy of the lane changing decision result output by the model by summarizing the lane changing characteristic indicators that have a significant impact on the lane changing behavior of the target vehicle.

[0087] In order to calculate the factors affecting the driver's lane selection, the following variables are used to describe the speed of the own vehicle and surrounding vehicles and the longitudinal distance between the own vehicle and surrounding vehicles, as follows:

[0088] v i , d i , i∈{E, P, PL, PR, FL, FR, L, R};

[0089] For example, eight vehicles are considered, including the target vehicle E and seven environment vehicles (P, PL, PR, FL, FR, L, R). Among them, (P, PL, PR, FL, FR, L, R) represent the target vehicle, the front vehicle in the target lane, the front vehicle in the adjacent left lane, the front vehicle in the adjacent right lane, the rear vehicle in the adjacent left lane, the rear vehicle in the adjacent right lane, the parallel vehicle in the adjacent left lane, and the parallel vehicle in the adjacent right lane. iand d i are the speed of the surrounding vehicle i and the longitudinal distance between the vehicle E and vehicle i. If the surrounding vehicle i does not exist, then v i and d i will be set to 0.

[0090] Lane changing safety safety Refers to the collision risk of the vehicle after changing to the target lane, which is one of the important factors to be considered when selecting the target lane. If there are adjacent vehicles in the target lane or the vehicle in front of the target lane does not leave enough overtaking distance, the vehicle has a greater risk of collision. However, if the distance and speed difference between the vehicle and the front and rear vehicles in the target lane are large enough, and there is no interference from adjacent vehicles in the target lane, the basic lane change conditions are met and the collision risk is low; otherwise, the vehicle will continue to follow. Therefore, lane safety can be expressed as follows:

[0091] f safety =f s (d FL , d FR , d FL , d PR , v E -v FL , v E -v FR , v E -v PL , v E -v PR , S L , S R );

[0092] Among them, S L , S R ∈{0, 1} indicates whether there are vehicles in the adjacent left lane and the adjacent right lane, 0 indicates no vehicles, and 1 indicates yes vehicles.

[0093] Drivers usually change lanes to obtain speed gain and a higher driving speed. In this embodiment, the speed gain is expressed by comparing the speed difference between the achievable speed of the left and right adjacent lanes and the speed of the vehicle, and the speed difference between the vehicle speed in the current lane and the vehicle in front. The achievable speed of the right lane is affected by the speed of the vehicle in front of the lane, so the gain V of the adjacent left lane is expressed by the following formula: ΔL and the speed gain V of the adjacent right lane ΔR Similarly, the distance gain is expressed by the distance between the vehicle in the current lane and the vehicle in front, and the distance between the vehicle in the left and right adjacent lanes. incentive It can be expressed as follows:

[0094] V ΔL =(v PL -v P )-(vE -v P )=v PL -v E ;

[0095] V ΔR =(v PR -v P )-(v E -v P )=v PR -v E ;

[0096] f incentive =f I (v P -v E , v PL -v E , v PR -v E , d PL -d P , d PR -d P );

[0097] Drivers usually consider the safety distance and the distance to the vehicle in front to determine the necessity of changing lanes. If the distance to the vehicle in front of the current lane is close to the safety distance or smaller than the safety distance, the driver tends to change lanes; usually, the driver tends to keep a larger distance than the safety distance. Therefore, the smaller the distance to the vehicle in front of the current lane is compared to the safety distance, the greater the necessity of changing lanes. Necessity of lane changef necessity It can be expressed as follows:

[0098]

[0099] Among them, t r represents the reaction time, where the driver's reaction time can be taken as 1s; μ is the friction coefficient between the vehicle and the road surface, which is usually 0.7s~0.9s, and can be conservatively taken as 0.7s here; g is the acceleration due to gravity, which is approximately 9.81m / s9.

[0100] In step S104 of some embodiments, the first driving style feature, the second driving style feature, the lane change decision index feature and the current driving data are input into the fully connected layer of the lane change decision model to obtain a lane change decision result, which is used to determine whether to change lanes and the lane change direction. The fully connected layer is represented as follows:

[0101] y=F LS ((f I , f S , f N , f DS (B k ));

[0102] Among them, f I is the motivation factor, f S For safety factors, f N is a necessary factor; DS (B k ) is the driving style characteristics of the vehicle itself and surrounding vehicles output by the hidden layer. The lane change decision model can output three types of results, namely the left adjacent lane, the right adjacent lane and the current lane.

[0103] Please refer to Figure 3 In this embodiment, the convolutional neural network is used to extract local features of driving style from the historical trajectory data of the target vehicle and the environment vehicle through its convolutional layer, and the bidirectional gated recurrent unit is used to mine the temporal dependency of driving style features, thereby obtaining the driving style features of each vehicle. The summarized personalized lane change factor features and the extracted driving style feature vector are input into the fully connected layer for further feature combination and transformation to obtain the lane change decision. The driving style of the human driver and the traffic factors that affect the driver's decision are comprehensively considered to identify the lane that the own vehicle should choose in the current scenario, thereby effectively capturing the lane change strategy of the human driver.

[0104] In some embodiments, the vehicle autonomous lane changing method of the embodiment of the present application may also include but is not limited to steps S501 to S503:

[0105] Step S501, when the lane changing decision result indicates a lane changing, determining a candidate trajectory cluster according to the lane changing direction in the lane changing decision result, wherein the candidate trajectory cluster includes a plurality of candidate lane changing trajectories;

[0106] Step S502, extracting first trajectory features of each candidate lane change trajectory and second trajectory features of the actual lane change trajectory based on the expected lane change driving index, and determining an actual trajectory data set according to the second trajectory features of all actual lane change trajectories;

[0107] Step S503: determining a target lane-changing trajectory according to the first trajectory feature and the real trajectory data set by using a maximum entropy inverse reinforcement learning algorithm.

[0108] In this embodiment, after obtaining the lane change decision result and determining that the target vehicle needs to change lanes, please refer to Figure 4In the embodiment of the present application, based on the target lane indicated by the lane change decision result output by the model, a quintic polynomial trajectory planning generation algorithm is used to generate a candidate trajectory cluster that can successfully drive from the lane change starting position of the own vehicle to the target lane. The extracted real lane change trajectory features of humans with different driving behaviors and the generated candidate lane change trajectory features are used as the input of the trajectory learning layer. The maximum entropy inverse reinforcement learning algorithm is used in the trajectory learning layer to maximize the human-like trajectory matching probability in the candidate trajectory, quantify the weight of each feature indicator, and finally obtain the reward function and the executable human-like target lane change trajectory.

[0109] In some embodiments of step S501, the step of determining a candidate trajectory cluster according to the lane changing direction in the lane changing decision result may include but is not limited to steps S601 to S604:

[0110] Step S601, determining a target lane according to the lane change direction, and constructing a time-continuous function of the trajectory with respect to the lateral and longitudinal directions, wherein the time-continuous function is a polynomial function;

[0111] Step S602, determining a lateral trajectory terminal interval according to the lateral area of ​​the target lane, and determining a longitudinal trajectory terminal interval according to the longitudinal lane change distance configuration;

[0112] Step S603, determining the boundary conditions of the time continuous function according to the lateral trajectory terminal interval, the longitudinal trajectory end interval and the current state data of the target vehicle;

[0113] Step S604: using a polynomial trajectory planning algorithm to determine a candidate trajectory cluster according to the time continuous function and the boundary conditions.

[0114] For example, a feasible trajectory cluster that enables the vehicle to successfully change to the target lane is generated using a quintic polynomial trajectory planning algorithm. The time-continuous function of the candidate trajectory cluster with respect to the lateral (x) and longitudinal (y) directions can be expressed as:

[0115] x(t)=a0+a1t+a2t 2 +a3t 3 +a4t 4 +a5t 5 ;

[0116] y(t)=b0+b1t+b2t 2 +b3t 3 +b4t 4 +b5t 5 ;

[0117] Where t represents the time variable, and {a0, ..., a5} and {b0, ..., b5} are the coefficients of the polynomial functions of the longitudinal and lateral trajectories, respectively.

[0118] The boundary conditions of the polynomial function used in trajectory planning are as follows:

[0119]

[0120] First, the initial state conditions of the vehicle need to be determined These conditions represent the initial position, velocity, and acceleration of the longitudinal (x) and lateral (y) trajectories. Then, the lateral trajectory endpoint interval y is set according to the lateral area coordinate position of the target lane. f ∈[y fmin ,y fmax ], and select different termination times t≤T to generate candidate trajectory clusters that can drive to the target lane within T time, covering as many vehicle driving trajectories as possible. For example, the candidate trajectory clusters are as follows Figure 5 The end point of the longitudinal trajectory can be determined according to the technical parameters of the road on which the vehicle is traveling; the number of generated trajectory clusters N is determined by the size of the sampling space.

[0121] In step S502 of some embodiments, based on the expected lane change driving index, the first trajectory feature of each candidate lane change trajectory and the second trajectory feature of the actual lane change trajectory are extracted, and the actual trajectory data set is determined according to the second trajectory features of all actual lane change trajectories. The expected lane change driving index may include efficiency, comfort, risk, and interactivity with surrounding vehicles, etc. The trajectory feature of the candidate lane change trajectory is formed by calculating the index value of the above-mentioned expected lane change driving index of the candidate lane change trajectory.

[0122] For each lane-changing trajectory, the vehicle driving behavior state and action at each time step are determined, and the time series is discretized. The state s of the vehicle at time step t is t The data includes the speed, position and direction of the vehicle itself and surrounding vehicles, as well as related features calculated based on these parameters. The action a at each time step t Including decisions such as speed and steering control. Therefore, the finite length lane change trajectory can be expressed as follows:

[0123] ζ=[s1, a1, s2, a2,...,s L , a L ];

[0124] Next, from the driving efficiency f I (s t ), comfort level C (s t ),risk and interactivity with surrounding vehicles I (s t ) are used to describe the actual lane changing trajectory and the candidate lane changing trajectory in four aspects, so as to extract the corresponding track features.

[0125] The driver's goal is usually to reach the destination in the shortest possible time, so the vehicle's speed v(t) at time step t is used to represent driving efficiency:

[0126] f I (s t ) = v(t);

[0127] Comfort is also one of the important factors that drivers consider when driving a vehicle. x (t), longitudinal acceleration a y (t) and acceleration rate of change jerk(t) are important indicators for measuring vehicle running comfort. Therefore, comfort can be expressed as:

[0128]

[0129] Considering the collision risk with the front and rear vehicles, the headway is considered to be a key indicator for assessing the risk of vehicle collision. Due to the nonlinear characteristics of the exponential function, it can significantly increase the sensitivity to risk in a short period of time. This nonlinear relationship can effectively reflect the changes in risk faced by vehicles in the real world. Therefore, an exponential function related to the headway is used to characterize the collision risk between the vehicle and the front and rear vehicles. and The specific formula is as follows:

[0130]

[0131] Among them, d p (t) is the distance between the preceding vehicle and the own vehicle at time step t, d R (t) is the distance between the vehicle after time step t and the vehicle itself, v E (t) is the speed of the vehicle at time step t, v R (t) is the speed of the vehicle after time step t.

[0132] Surrounding vehicles will be affected by the own vehicle, which is a basic feature of human driving behavior, mainly reflected in the deceleration reaction of surrounding vehicles. Therefore, the deceleration change a of the surrounding vehicles affected by the own vehicle at each time step t is i The sum describes the interaction between vehicles f I (s t ). The specific description is as follows:

[0133] f I (s t )=∑ i a i (t), a i (t)<0.

[0134] In some embodiments of step S503, step S503 may include but is not limited to steps S701 to S703:

[0135] Step S701, taking maximizing the objective function as the optimization goal, determining a feature weight vector of each candidate lane changing trajectory according to the first trajectory feature of each candidate lane changing trajectory and the real trajectory data set;

[0136] Step S702, determining a reward function of the candidate lane-changing trajectory according to the feature weight vector of the candidate lane-changing trajectory;

[0137] Step S703: selecting and determining a target lane-changing trajectory according to the reward function of each candidate lane-changing trajectory.

[0138] This embodiment extracts the features of the human expert demonstration trajectory (i.e., the real lane-changing trajectory) and the candidate lane-changing trajectory as the input of the maximum entropy inverse reinforcement learning algorithm to achieve the imitation of the human expert trajectory. The objective function is constructed based on the maximum entropy inverse reinforcement learning algorithm theory to improve the selection probability of the human-like candidate trajectory. Its optimization objectives are as follows:

[0139]

[0140] in, represents the set of human expert demonstration trajectories (i.e., the real trajectory dataset); P(ζ|θ) represents the probability of selecting trajectory ζ, θ Represents the weight vector for model optimization.

[0141] The probability P(ζ|θ) of selecting a candidate lane-changing trajectory ζ is described using the Boltzmann distribution associated with the trajectory reward. This distribution also conforms to the maximum entropy principle of the expected feature distribution corresponding to the expert demonstration trajectory. Specifically, the probability of trajectory ζ is proportional to the exponential of the reward associated with trajectory ζ. This method helps avoid premature convergence to a suboptimal solution. The exact formula is as follows:

[0142]

[0143] Among them, R(ζ) represents the reward function of the candidate lane-changing trajectory ζ, and Z(θ) represents the distribution function.

[0144] Considering the stability of the urban expressway driving scene, the state s t The reward function of the state s is defined as a linear structural equation t Therefore, the vehicle state s t The reward function r(s t ) can be defined as follows:

[0145] r(s t )=θ Tf(s t );

[0146] Among them, θ = [θ1, θ2,…, θ k ] is a K-dimensional weight vector; f(s t )=[f1(s t ), f2(s t ), …, f K (s t )] is used to describe the vehicle state s t The feature vector of . Therefore, the reward function R(ζ) of trajectory ζ can be expressed as:

[0147]

[0148] where f ζ represents the accumulated features of the trajectory ζ at each discrete time step.

[0149] Since a limited number of candidate trajectories are generated here based on the lane selection based on the driving style, and the partition function is approximated based on the generated limited feasible trajectories to reduce the computational space. Therefore, the probability P(ζ|θ) of trajectory ζ can be expressed as:

[0150]

[0151] Among them, ξ i represents the lateral displacement that sets the termination condition based on the decision result of the upper lane selection layer, and is a candidate generated trajectory with the same initial state as the vehicle in the current scenario, is the feature vector extracted for this trajectory, and N is the number of candidate trajectories generated.

[0152] According to some embodiments of this application, please refer to Figure 6The application architecture of the vehicle autonomous lane changing method of the embodiment of the present application may include an upper optimal lane selection layer and a lower inverse reinforcement learning trajectory planning layer. The historical trajectory data of the target vehicle and surrounding vehicles are input into the optimal lane selection layer for driving style extraction, and the personalized lane changing decision index is selected by analyzing the lane changing decision index characteristics considering the vehicle driving scenario. The driving behavior decision is made based on the lane changing decision index characteristics and the driving style characteristics to obtain the lane changing decision result. The lower inverse reinforcement learning trajectory planning layer combines the real human expert trajectory characteristics to optimize the feature weight vector of the candidate lane changing trajectory, so that the candidate lane changing trajectory can simulate the human lane changing trajectory, thereby determining the target lane changing trajectory. The upper lane selection layer of this embodiment considers the driving style of the own vehicle and surrounding vehicles, and uses an improved neural network to implicitly represent the human driving style, and then embeds the vehicle driving style into the upper lane selection model of the decision framework, which can significantly improve the lane selection accuracy. The lower-level trajectory planning learning layer obtains the superior candidate trajectory cluster in the current scenario based on the upper-level lane judgment, and imitates the characteristics of human driving behavior through maximum entropy inverse reinforcement learning, which effectively improves the accuracy of the model. At the same time, it infers the potential personalized reward function, enabling autonomous driving vehicles to better understand the motivations and preferences of vehicle driving behavior.

[0153] The embodiment of the present application also provides a vehicle autonomous lane changing system, comprising:

[0154] The first module is used to obtain the first historical trajectory data of the target vehicle and the second historical trajectory data of the environment vehicle;

[0155] A second module is used to determine a first driving style feature of the target vehicle and a second driving style feature of the environment vehicle according to the first historical trajectory data and the second historical trajectory data through a hidden layer of the lane change decision model;

[0156] A third module is used to determine a lane-changing decision indicator feature that affects the lane-changing behavior according to the first historical trajectory data;

[0157] The fourth module is used to input the first driving style feature, the second driving style feature, the lane change decision index feature and the current driving data into the fully connected layer of the lane change decision model to obtain a lane change decision result, wherein the lane change decision result is used to determine whether to change lanes and the lane change direction.

[0158] It can be understood that the contents of the above-mentioned vehicle autonomous lane changing method embodiment are all applicable to the present system embodiment, the functions specifically implemented by the present system embodiment are the same as those in the above-mentioned vehicle autonomous lane changing method embodiment, and the beneficial effects achieved are also the same as those achieved by the above-mentioned vehicle autonomous lane changing method embodiment.

[0159] The embodiment of the present application also provides an electronic device, the electronic device comprising: a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory, and the program is executed by the processor to realize the above-mentioned vehicle autonomous lane changing method. The electronic device can be any intelligent terminal including a tablet computer, a vehicle-mounted computer, etc.

[0160] See also Figure 7 , Figure 7 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0161] The processor 701 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0162] The memory 702 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 702 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 702, and the processor 701 calls and executes the vehicle autonomous lane changing method of the embodiment of this application;

[0163] Input / output interface 703, used to implement information input and output;

[0164] Communication interface 704, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.);

[0165] A bus 705 that transmits information between the various components of the device (e.g., the processor 701, the memory 702, the input / output interface 703, and the communication interface 704);

[0166] The processor 701 , the memory 702 , the input / output interface 703 and the communication interface 704 are connected to each other in communication within the device via a bus 705 .

[0167] An embodiment of the present application also provides a storage medium, which is a computer-readable storage medium used for computer-readable storage. The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the above-mentioned vehicle autonomous lane changing method.

[0168] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0169] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0170] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0171] The system embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment.

[0172] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0173] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0174] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0175] In the several embodiments provided in the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or units, which can be electrical, mechanical or other forms.

[0176] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0177] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0178] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0179] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.

Claims

1. A vehicle autonomous lane changing method, characterized in that: The following steps are involved: Acquire first historical trajectory data of the target vehicle and second historical trajectory data of the environment vehicle; Determining, through a hidden layer of the lane change decision model, a first driving style feature of the target vehicle and a second driving style feature of the environment vehicle according to the first historical trajectory data and the second historical trajectory data; Determining a lane-changing decision indicator feature that affects the lane-changing behavior according to the first historical trajectory data; The first driving style feature, the second driving style feature, the lane change decision index feature and the current driving data are input into a fully connected layer of a lane change decision model to obtain a lane change decision result, wherein the lane change decision result is used to determine whether to change lanes and the lane change direction.

2. The vehicle autonomous lane changing method according to claim 1, characterized in that: The method of determining the first driving style feature of the target vehicle and the second driving style feature of the environment vehicle according to the first historical trajectory data and the second historical trajectory data through the hidden layer of the lane change decision model comprises the following steps: Constructing a trajectory feature matrix of a target vehicle and a trajectory feature matrix of an environment vehicle respectively according to the first historical trajectory data and the second historical trajectory data; The trajectory feature matrix of the target vehicle and the trajectory feature matrix of the environment vehicle are respectively input into the hidden layer of the lane change decision model to obtain a first driving style feature of the target vehicle and a second driving style feature of the environment vehicle, wherein the hidden layer includes a convolutional neural network and a bidirectional gated recurrent unit connected in sequence.

3. The vehicle autonomous lane changing method according to claim 2, characterized in that: The trajectory feature matrix of the target vehicle or the environment vehicle is obtained by the following steps: Segmenting the vehicle trajectory according to a preset time step to obtain a plurality of trajectory segments, wherein the vehicle trajectory is the first historical trajectory data or the second historical trajectory data; Calculating multiple variable parameters of each trajectory segment to obtain a feature vector of the trajectory segment, wherein the multiple parameter variables include lateral speed, longitudinal speed, total speed, lateral acceleration, longitudinal acceleration, total acceleration, time interval and driving distance; The feature vectors of a plurality of trajectory segments of the vehicle trajectory are combined to obtain a trajectory feature matrix of the vehicle trajectory.

4. The vehicle autonomous lane changing method according to claim 1, characterized in that: The determining, according to the first historical trajectory data, the lane changing decision indicator characteristics affecting the lane changing behavior comprises the following steps: determining a lane change decision point in the first historical trajectory data by analyzing fluctuation characteristics of the first historical trajectory data; The lane-changing decision index characteristics of the target vehicle are determined according to the historical driving data of the lane-changing decision point.

5. The vehicle autonomous lane changing method according to claim 1, characterized in that: The vehicle autonomous lane changing method further comprises the following steps: When the lane-changing decision result represents a lane-changing, determining a candidate trajectory cluster according to the lane-changing direction in the lane-changing decision result, wherein the candidate trajectory cluster includes a plurality of candidate lane-changing trajectories; Based on the expected lane-changing driving index, extracting the first trajectory feature of each of the candidate lane-changing trajectories and the second trajectory feature of the actual lane-changing trajectory, and determining the actual trajectory data set according to the second trajectory features of all the actual lane-changing trajectories; A target lane-changing trajectory is determined according to the first trajectory feature and the real trajectory data set by using a maximum entropy inverse reinforcement learning algorithm.

6. The vehicle autonomous lane changing method according to claim 5, characterized in that: The step of determining a candidate trajectory cluster according to the lane changing direction in the lane changing decision result comprises the following steps: Determine the target lane according to the lane change direction, and construct a time-continuous function of the trajectory with respect to the lateral direction and the longitudinal direction, wherein the time-continuous function is a polynomial function; Determining a lateral trajectory terminal interval according to a lateral area of ​​the target lane, and determining a longitudinal trajectory terminal interval according to a longitudinal lane change distance configuration; Determining the boundary conditions of the time continuous function according to the lateral trajectory terminal interval, the longitudinal trajectory terminal interval and the current state data of the target vehicle; A polynomial trajectory planning algorithm is used to determine a candidate trajectory cluster according to the time continuous function and the boundary conditions.

7. The vehicle autonomous lane changing method according to claim 6, characterized in that: The method of determining the target lane-changing trajectory according to the first trajectory feature and the real trajectory data set by using the maximum entropy inverse reinforcement learning algorithm comprises the following steps: Taking maximizing the objective function as the optimization goal, determining a feature weight vector of each candidate lane changing trajectory according to the first trajectory feature of each candidate lane changing trajectory and the real trajectory data set; Determining a reward function of the candidate lane-changing trajectory according to a feature weight vector of the candidate lane-changing trajectory; Determine a target lane-changing trajectory by selecting according to the reward function of each of the candidate lane-changing trajectories; The optimization objective is expressed as follows: in, represents the objective function, represents the real trajectory dataset, P(ζ|θ) represents the probability of the candidate lane-changing trajectory ζ, and θ represents the feature weight vector to be optimized.

8. A vehicle autonomous lane changing system, characterized in that: include: The first module is used to obtain the first historical trajectory data of the target vehicle and the second historical trajectory data of the environment vehicle; A second module is used to determine a first driving style feature of the target vehicle and a second driving style feature of the environment vehicle according to the first historical trajectory data and the second historical trajectory data through a hidden layer of the lane change decision model; A third module is used to determine a lane-changing decision indicator feature that affects the lane-changing behavior according to the first historical trajectory data; The fourth module is used to input the first driving style feature, the second driving style feature, the lane change decision index feature and the current driving data into the fully connected layer of the lane change decision model to obtain a lane change decision result, wherein the lane change decision result is used to determine whether to change lanes and the lane change direction.

9. An electronic device, characterized in that: The electronic device includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory. When the program is executed by the processor, the steps of the method described in any one of claims 1 to 7 are realized.

10. A storage medium, the storage medium being a computer-readable storage medium, used for computer-readable storage, characterized in that: The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of any one of claims 1 to 7.

Images