CN114162145A - Automatic vehicle driving method and device and electronic equipment - Google Patents

Abstract

The application provides a vehicle automatic driving method, a vehicle automatic driving device and electronic equipment, wherein the method comprises the following steps: the method comprises the steps of firstly obtaining driving information of a target vehicle, driving information of related vehicles and lane information, then predicting driving scores through a first neural network model to obtain driving scores of a plurality of different driving strategies, and finally determining the target driving strategy according to the driving scores. According to the technology, a plurality of driving strategies are determined firstly through driving decisions, then score prediction is carried out on each driving strategy by using the neural network model, so that the finally determined target driving strategy integrates effective prediction of the running performance of a target vehicle by the neural network on the basis of a traditional decision algorithm, the target driving strategy determined through the scores predicted by the first neural network model is closer to the actual complex traffic condition, and the safety of automatic driving of the vehicle is effectively improved.

Description

Vehicle automatic driving method, device and electronic device

technical field

The present application relates to the technical field of automatic driving, and in particular, to a vehicle automatic driving method, device and electronic device.

Background technique

With the rapid development of artificial intelligence technology and urban intelligent rail transit, self-driving vehicles, namely self-driving systems, have gradually entered the daily life of the public. The self-driving system is a comprehensive system that brings together many high-tech, as a key link in environmental information acquisition and Intelligent decision control relies on a series of high-tech innovations and breakthroughs such as sensor technology, image recognition technology, electronic and computer technology and control technology.

How to make reasonable and predictable driving decisions in a dynamic and complex environment to ensure the fast and stable driving of autonomous vehicles is a very challenging research topic in the field of autonomous driving. In the existing technology, decisions are usually based on finite states Algorithms to determine driving strategies.

However, for traditional decision-making algorithms based on rules and finite states, the decision-making needs to be based on a specific driving situation, and the real traffic driving environment is a complex situation with a combination of various factors, which cannot be well matched with the simply set driving situation. Complete compliance, which leads to the fact that the driving strategy determined by the traditional decision-making algorithm cannot meet the actual complex road conditions, and there will be dangerous situations such as collisions.

SUMMARY OF THE INVENTION

In view of this, the purpose of the present application is to provide a vehicle automatic driving method, device and electronic device, so as to improve the safety of vehicle automatic driving.

In a first aspect, an embodiment of the present application provides an automatic driving method for a vehicle, the method includes: acquiring first driving information of a target vehicle at the current moment, and second driving information corresponding to a related vehicle within a first specified range from the target vehicle and the lane information of the relevant lanes within a second specified range from the target vehicle; the first neural network model is used to predict the driving score of the target vehicle traveling with a plurality of different predetermined driving strategies through the first driving information, the second driving information and the lane information ; wherein, the driving score is used to characterize the driving conditions and vehicle performance of the target vehicle traveling with a predetermined driving strategy; the first neural network model is obtained by training corresponding sample data under different traffic conditions; The score determines the target driving strategy from a number of different driving strategies, so that the target vehicle drives according to the target driving strategy at the next moment from the current moment.

Further, the above-mentioned related vehicles within a specified range from the target vehicle include at least one of the following: a first related vehicle whose distance from the target vehicle in the lane where the target vehicle is located is less than a first distance threshold and is located in front of the driving direction of the target vehicle; The second related vehicle in the right lane of the lane where the distance from the target vehicle is less than the second distance threshold and located in front of the driving direction of the target vehicle; in the lane on the right side of the lane where the target vehicle is located, the distance from the target vehicle is less than the third distance threshold and located in the target vehicle The third related vehicle behind the driving direction; the fourth related vehicle in the left lane of the target vehicle's lane with the A fifth associated vehicle whose distance to the target vehicle is less than the fifth distance threshold and is located behind the direction of travel of the target vehicle.

Further, the first travel information of the target vehicle at the current moment includes the first position and the first speed of the target vehicle; the step of obtaining the second travel information corresponding to the relevant vehicles within a specified range of the target vehicle includes: obtaining relevant The second position and second speed of the vehicle at the current moment; calculate the relative position of the second position relative to the first position, and the relative speed of the second speed relative to the first speed; determine the relative position and relative speed as the corresponding vehicle of the second driving information.

Further, the above method also includes: if the lane where the target vehicle is located is the rightmost lane in the road, setting the second driving information of the relevant vehicle in the lane on the right side of the lane where the target vehicle is located to 0; if the lane where the target vehicle is located is the road In the leftmost lane, the second driving information of the relevant vehicle in the left lane of the lane where the target vehicle is located is set to 0.

Further, the above-mentioned relevant lanes include the lane where the target vehicle is located, the right lane of the lane where the target vehicle is located, and the left lane of the lane where the target vehicle is located; the step of acquiring the lane information of the relevant lane within a second specified range from the target vehicle includes: : take the position of the vehicle at the current moment as the origin position; calculate the horizontal distance and the longitudinal distance between each position in the first preset number of continuous positions in the relevant lane and the origin position; wherein, in the first preset number of continuous positions The distance between every two adjacent positions is equal; the set consisting of the lateral distance and the longitudinal distance corresponding to each position in the first preset number of continuous positions is determined as the distance between the relevant lane within the second specified range from the target vehicle Lane information.

Further, the above-mentioned multiple driving strategies include at least any two of the following: accelerating straight ahead, maintaining speed straight ahead, decelerating straight ahead, straight-line emergency braking, uniform lane change to the left lane, and uniform lane change to the right lane.

Further, the above-mentioned first neural network model is obtained by training through the following steps: obtaining sample data; wherein, the sample data includes the speed information and position information of the target sample vehicle, and the speed information and position information of the relevant sample vehicles related to the target sample vehicle, and the lane information of the lane corresponding to the target sample vehicle; calculate the loss value of the initial neural network model corresponding to each moment in the second preset number of consecutive moments according to the sample data and the loss function; wherein, the loss function includes collision parameters, energy consumption One or more of parameters, lane change penalty parameters, and driving efficiency parameters; adjust the parameters of the initial neural network model according to the loss value, and determine the initial neural network model corresponding to the training stop condition as the first neural network model.

Further, the above method also includes: determining the prediction scores corresponding to multiple driving strategies at the next moment at the current moment in the second preset number of consecutive moments through an initial neural network model; and using the driving strategy with the largest predicted score as the next moment. The driving strategy is obtained, and the sample data of the next moment generated by the target sample vehicle driving under the driving strategy of the next moment is obtained; based on the sample data of the next moment, the initial neural network model is continued to be trained.

Further, the above-mentioned step of determining a target driving strategy from a plurality of different driving strategies according to the driving scores corresponding to the plurality of different driving strategies includes: determining a driving strategy with a driving score higher than a preset score threshold as a set of alternative driving strategies ; Determine the target driving strategy from the set of alternative driving strategies.

Further, the above-mentioned step of determining the target driving strategy from the alternative driving strategy set includes: judging whether the driving strategy with the highest driving score in the alternative driving strategy set satisfies the preset safe driving conditions; if so, determining the driving strategy is the target driving strategy; if not satisfied, delete the driving strategy from the set of alternative driving strategies.

其中，

表示方向盘的转角，θ_n与θ_f分别表示目标车道上前方第一位置与前方第二位置与自车位置形成的夹角，Δt为时间步长，k_f、k_n和k_I表示驾驶行为的常数，目标车道为目标车辆准备从目标车辆所在的当前车道切换的车道。Further, the above method also includes: if the target driving strategy is to change lanes at a constant speed to the left lane or change lanes at a constant speed to the right lane, then control the vehicle to change lanes by the following formula:

in,

Represents the steering wheel angle, θ _n and θ _f respectively represent the angle formed by the first position in front of the target lane and the second position in front and the position of the vehicle, Δt is the time step, k _f , k _n and k _I represent the driving behavior The constant of , the target lane is the lane that the target vehicle is ready to switch from the current lane where the target vehicle is located.

In a second aspect, an embodiment of the present application further provides an automatic driving device for a vehicle, the device includes: an information acquisition module configured to acquire first driving information of a target vehicle at the current moment and a distance from the target vehicle within a first specified range The second driving information corresponding to the relevant vehicle and the lane information of the relevant lane within the second specified range from the target vehicle; the prediction module is used to pass the first neural network model, the first driving information, the first The driving information and the lane information predict the driving score of the target vehicle when the target vehicle is driven by a plurality of different predetermined driving strategies; wherein, the driving score is used to characterize the driving condition of the target vehicle driving by the predetermined driving strategy and vehicle performance; the first neural network model is obtained by training corresponding sample data under different traffic conditions; a target driving strategy determination module is used for driving scores corresponding to the plurality of different driving strategies from the plurality of different driving strategies A target driving strategy is determined in the driving strategy, so that the target vehicle drives according to the target driving strategy at the next moment of the current moment.

In a third aspect, embodiments of the present application further provide an electronic device, including a processor and a memory, where the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the vehicle of the first aspect above Autopilot method.

In a fourth aspect, embodiments of the present application further provide a computer-readable storage medium, where the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are invoked and executed by the processor, the computer-executable instructions cause the processor to The vehicle automatic driving method of the above first aspect is realized.

Compared with the prior art, the present application has the following beneficial effects:

The above-mentioned vehicle automatic driving method, device, and electronic device provided by the embodiments of the present application first obtain the driving information of the target vehicle, the driving information and lane information of the relevant vehicle, and then use the first neural network model to predict the driving score, and obtain multiple Driving scores of different driving strategies, and finally determine the target driving strategy according to the driving scores. In the technology of the present application, a plurality of driving strategies are firstly determined through driving decisions, and then a neural network model is used to predict the score of each driving strategy, so that the final target driving strategy is based on the traditional decision-making algorithm. The network can effectively predict the driving performance of the target vehicle. Since the first neural network model is obtained by training sample data under a variety of different traffic conditions, the target driving strategy determined by the score predicted by the first neural network model is closer to the target. The actual complex traffic situation effectively improves the safety of vehicle automatic driving.

Additional features and advantages of the present disclosure will be set forth in the description that follows, or some may be inferred or unambiguously determined from the description, or may be learned by practicing the above-described techniques of the present disclosure.

In order to make the above-mentioned objects, features and advantages of the present disclosure more obvious and easy to understand, the preferred embodiments are exemplified below, and are described in detail as follows in conjunction with the accompanying drawings.

Description of drawings

In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the accompanying drawings that need to be used in the description of the specific embodiments or the prior art will be briefly introduced below. The drawings are some embodiments of the present application. For those of ordinary skill in the art, other drawings can also be obtained from these drawings without any creative effort.

1 is a schematic structural diagram of an electronic system according to an embodiment of the present application;

FIG. 2 is a flowchart of a method for automatic driving of a vehicle provided by an embodiment of the present application;

FIG. 3 is a flowchart of another vehicle automatic driving method provided by an embodiment of the present application;

4 is a schematic structural diagram of a first neural network model provided by an embodiment of the present application;

5 is a flowchart of a training method of a first neural network model provided by an embodiment of the present application;

FIG. 6 is a schematic structural diagram of a vehicle automatic driving device provided by an embodiment of the present application;

FIG. 7 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the present application will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, not all of them. example. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

The main purpose of autonomous driving is to reduce labor costs, increase transportation efficiency, reduce energy consumption, and avoid traffic accidents. With the development of artificial intelligence, the technology of autonomous driving of motor vehicles is also becoming more and more mature. In the early days of autonomous driving, the main focus of research and development was on a safe and stable driving experience. In recent years, how to improve driving efficiency and reduce fuel consumption on the basis of safety has attracted wider attention. How to make reasonable and predictable driving decisions in dynamic and complex environments to ensure fast and smooth driving of autonomous vehicles is still very challenging. Although traditional rule-based and finite-state decision-making algorithms have achieved success in some autonomous driving tasks, one of their biggest drawbacks is that their decision-making needs to be based on specific driving situations. This makes it difficult to generalize more complex real traffic driving environments. Based on this, embodiments of the present application provide a vehicle automatic driving method, device, and electronic device, so as to improve the safety of vehicle automatic driving.

Referring to the schematic structural diagram of the electronic system 100 shown in FIG. 1 . The electronic system can be used to implement the vehicle automatic driving method and device according to the embodiments of the present application.

As a schematic structural diagram of an electronic system shown in FIG. 1 , the electronic system 100 includes one or more processing devices 102 and one or more storage devices 104 . Optionally, the electronic system 100 may also include an input device 106, an output device 108, and one or more data acquisition devices 110, these components being interconnected by a bus system 112 and/or other form of connection mechanism (not shown). It should be noted that the components and structures of the electronic system 100 shown in FIG. 1 are only exemplary and non-limiting, and the electronic system may have some components in FIG. 1 or other components and structures as required.

The processing device 102 can be a server, an intelligent terminal, or a device that includes a central processing unit (CPU) or other forms of processing units with data processing capabilities and/or instruction execution capabilities, and can process data from other components in the electronic system 100 . Processing may also control other components in the electronic system 100 to perform vehicle autopilot functions.

Storage 104 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and/or cache memory, among others. Non-volatile memory may include, for example, read only memory (ROM), hard disk, flash memory, and the like. One or more computer program instructions may be stored on the computer-readable storage medium, and the processing device 102 may execute the program instructions to implement the client functions (implemented by the processing device) in the following embodiments of the present application and/or other desired Function. Various application programs and various data, such as various data used and/or generated by the application program, etc., may also be stored in the computer-readable storage medium.

Input device 106 may be a device used by a user to input instructions, and may include one or more of a keyboard, mouse, microphone, touch screen, and the like.

The output device 108 may output various information (eg, images or sounds) to the outside (eg, a user), and may include one or more of a display, a speaker, and the like.

The data acquisition device 110 may acquire the data to be processed and store the data to be processed in the storage device 104 for use by other components.

Exemplarily, each device in the method, device, and electronic device for implementing the automatic driving of a vehicle according to the embodiments of the present application may be integrated or distributed, such as the processing device 102 , the storage device 104 , the input device 106 , and the output device. The device 108 is integrated and arranged in one body, and the data collection device 110 is arranged at a designated location where data can be collected. When the various devices in the above electronic system are integrated, the electronic system can be implemented as a smart terminal such as a camera, a smart phone, a tablet computer, a computer, a vehicle-mounted terminal, and the like.

FIG. 2 is a flowchart of a vehicle automatic driving method provided by an embodiment of the present application. As shown in FIG. 2 , the method specifically includes the following steps:

S202: Acquire first driving information of the target vehicle at the current moment, second driving information corresponding to a relevant vehicle within a first specified range from the target vehicle, and lane information of a relevant lane within a second specified range from the target vehicle;

The target vehicle is a vehicle controlled by the method provided in this embodiment of the application, and the first driving information may include information such as speed and position of the target vehicle. The relevant vehicle is a vehicle whose distance from the target vehicle is within the first specified range, for example, the preceding vehicle and the rear vehicle in the same lane of the target vehicle, the preceding vehicle and the rear vehicle in the adjacent lane of the target vehicle, and the like. The second travel information includes speed and position information of the relevant vehicle. The lane information includes information such as the relative distance between the lane and the lane where the target vehicle is located.

S204: Predict the driving score of the target vehicle traveling with multiple different predetermined driving strategies by using the first neural network model, the first driving information, the second driving information and the lane information; wherein, the driving score is used to characterize the target vehicle with the predetermined driving strategy. The driving conditions and vehicle performance for driving; the first neural network model is obtained by training the corresponding sample data under different traffic conditions;

The first neural network model is a driving score prediction model, which can process the input first driving information, second driving information and lane information to obtain a plurality of driving scores, each driving score corresponds to a predetermined driving strategy, for example, a predetermined driving strategy. Assuming five driving strategies, the driving scores are 10, 25, 80, 45, and 96, respectively. Then, according to the driving scores, it can be determined which driving strategy the target vehicle adopts to drive at the next moment. The training method of the first neural network model will be described in detail below, and will not be repeated here.

S206: Determine a target driving strategy from the plurality of different predetermined driving strategies according to the driving scores corresponding to the plurality of different predetermined driving strategies, so that the target vehicle drives according to the target driving strategy at the next moment of the current moment.

The above-mentioned vehicle automatic driving method provided by the embodiment of the present application first obtains the driving information of the target vehicle, the driving information of the relevant vehicle and the lane information, and then uses the first neural network model to predict the driving score, and obtains driving with multiple different driving strategies. Score, and finally determine the target driving strategy according to the driving score. In the technology of the present application, a plurality of driving strategies are firstly determined through driving decisions, and then a neural network model is used to predict the score of each driving strategy, so that the final target driving strategy is based on the traditional decision-making algorithm. The network can effectively predict the driving performance of the target vehicle. Since the first neural network model is obtained by training sample data under a variety of different traffic conditions, the target driving strategy determined by the score predicted by the first neural network model is closer to the target. The actual complex traffic situation effectively improves the safety of vehicle automatic driving.

In some possible implementations, the above-mentioned related vehicles within a specified range from the target vehicle include at least one of the following:

(1) A first related vehicle whose distance from the target vehicle in the lane where the target vehicle is located is less than the first distance threshold and is located in front of the target vehicle's traveling direction;

(2) a second related vehicle in the right lane of the lane where the target vehicle is located, the distance from the target vehicle is less than the second distance threshold and is located in front of the target vehicle's driving direction;

(3) a third related vehicle whose distance from the target vehicle is less than the third distance threshold and is located behind the target vehicle's driving direction in the right lane of the lane where the target vehicle is located;

(4) a fourth related vehicle in the left lane of the lane where the target vehicle is located, the distance from the target vehicle is less than the fourth distance threshold and is located in front of the target vehicle's driving direction;

(5) A fifth related vehicle whose distance from the target vehicle is less than the fifth distance threshold and located behind the target vehicle's traveling direction in the right lane of the lane where the target vehicle is located.

FIG. 3 is a flowchart of another vehicle automatic driving method provided by the embodiment of the present application. The method focuses on describing the specific implementation of how to obtain the second driving information corresponding to the relevant vehicle. As shown in FIG. 3 , the method specifically includes: The following steps:

S302: Acquire first driving information of the target vehicle at the current moment and lane information of a relevant lane within a second specified range from the target vehicle;

The first travel information of the target vehicle at the current moment includes the first position and the first speed of the target vehicle.

S304: Obtain the second position and second speed of the relevant vehicle at the current moment;

S306: Calculate the relative position of the second position with respect to the first position, and the relative velocity of the second speed with respect to the first speed;

S308: Determine the relative position and relative speed as the second travel information corresponding to the relevant vehicle.

In some examples, if the lane where the target vehicle is located is the rightmost lane in the road, setting the second travel information of the relevant vehicle located in the right lane of the lane where the target vehicle is located to 0;

In other examples, if the lane where the target vehicle is located is the leftmost lane in the road, the second travel information of the relevant vehicle located in the left lane of the lane where the target vehicle is located is set to 0.

The above-mentioned right lane of the rightmost lane and the left lane of the leftmost lane are equivalent to the lanes beyond the boundary, which means that the considered target lane does not exist in the real road. For example, when we are driving in the rightmost lane, we need to consider the information about the obstacles in the own lane and the 3 lanes on the left and right. In this case, the corresponding lane and the left lane are real, so the information of the obstacle corresponds to the real information. But the right lane does not exist and is a virtual concept. For consistency, we assume that in this case the right lane is filled with virtual vehicles to ensure that the lane beyond the boundary is inaccessible.

S310: Predict the driving score of the target vehicle traveling with multiple different predetermined driving strategies by using the first neural network model, the first driving information, the second driving information and the lane information;

S312: Determine a target driving strategy from a plurality of different predetermined driving strategies according to the driving scores corresponding to the plurality of different predetermined driving strategies, so that the target vehicle drives according to the target driving strategy at the next moment of the current moment.

With the method provided by the above embodiment, the information of the target vehicle, related vehicles and lanes can be fully considered in the process of predicting the vehicle driving strategy, so that the driving score corresponding to the driving strategy is more in line with the actual traffic driving conditions.

In some possible implementations, the above-mentioned relevant lanes include the lane where the target vehicle is located, the right lane of the lane where the target vehicle is located, and the left lane of the lane where the target vehicle is located; based on this, the lane information of the relevant lane can be obtained by the following methods:

(1) Take the position of the vehicle at the current moment as the origin position;

(2) Calculate the lateral distance and the longitudinal distance between each position in the first preset number of continuous positions in the relevant lane and the origin position; wherein, in the first preset number of continuous positions, the distance between each two adjacent positions equal distance

(3) Determining the set of the lateral distance and the longitudinal distance corresponding to each of the first preset number of continuous positions as the lane information of the relevant lane within the second specified range from the target vehicle.

In some possible implementations, the plurality of driving strategies in the above embodiments include at least any two of the following: accelerating straight ahead, maintaining speed straight ahead, decelerating straight ahead, straight-line emergency braking, changing lanes at a constant speed to the left lane, and going straight to the right lane Change lanes evenly.

The following describes the structure of the first neural network model and the training method thereof provided by the embodiment of the present application with reference to FIG. 4 . As shown in Figure 4, the input of the first neural network model consists of three parts: target vehicle features, related vehicle features, and lane structure features. The target vehicle features and related vehicle features both output feature vectors with a size of 128 through the multi-layer perceptron. Each lane structure feature is outputted as a feature vector of size 1024 through a series of one-dimensional convolutions. After integrating the relevant vehicle feature vector with the lane feature vector, through a fully connected neural network, an output value of size 6 is finally obtained, corresponding to the driving score of each driving strategy, as shown in Figure 4. The driving score of a driving strategy indicates the quality of the driving strategy. In practical applications, the driving strategy with the largest driving score should be selected as the optimal strategy choice at the current moment. decision-making scheme to consider the feasibility of the optimal strategy.

In the training process of the first neural network model, in order to simulate the traffic flow of the expressway in reality, the expressway used in the experiment and test is 4 rows of one-way driving lanes. The surrounding vehicles are randomly selected from the preset 18 types of vehicles (length range: 2~18m, width range: 1.6~3m), and the maximum number is 50 vehicles.

To ensure safe driving behavior, the driving model of surrounding vehicles adopts the longitudinal intelligent driver (IDM) following model, and the lateral adopts the minimization of total braking caused by lane change (MOBIL) adaptive cruise controller.

Among them, the expression of the longitudinal IDM cruise controller is

Among them, v represents the speed of the vehicle, a represents the maximum expected acceleration, b represents the expected speed reduction rate, s ₀ is the minimum interval between the two vehicles, s represents the actual interval between the two vehicles, and Δv represents the distance between the two vehicles The speed difference of , _vset is the desired speed, and _Tset is the desired time interval.

The expression of the lateral MOBIL lane change decision controller is:

和

分别表示变道任务执行后自车的加速度，变道目标车道后续车辆的加速度以及自车后续跟随车辆的加速度。b_safe表示车辆速度最大的减小速率。Δa_th表示加速度转变阈值。当(3)与(4)同时满足时，车辆则向目标车道执行变道任务。Among them, a _c , an and a _o respectively represent the acceleration of the own vehicle, the acceleration of the following vehicle in the lane _- change target lane, and the acceleration of the following vehicle of the own vehicle.

and

Respectively represent the acceleration of the ego vehicle after the lane change task is executed, the acceleration of the following vehicle in the lane change target lane, and the acceleration of the ego vehicle following the vehicle. b _safe represents the maximum reduction rate of vehicle speed. Δa _th represents the acceleration transition threshold. When (3) and (4) are satisfied at the same time, the vehicle performs the lane change task to the target lane.

On the basis of the above model structure, the embodiment of the present application also provides a training method for the first neural network model. The method shown in 5 is used to train the model, and the method can specifically include the following steps:

S502: Obtain sample data; wherein the sample data includes speed information and position information of the target sample vehicle, speed information and position information of the relevant sample vehicle related to the target sample vehicle, and lane information of the lane corresponding to the target sample vehicle;

S504: Calculate the loss value of the initial neural network model corresponding to each of the second preset number of consecutive moments according to the sample data and the loss function; wherein the loss function includes a collision parameter, an energy consumption parameter, a lane change penalty parameter, and a driving efficiency one or more of the parameters;

S506: Adjust the parameters of the initial neural network model according to the loss value, and determine the corresponding initial neural network model when the training stop condition is satisfied as the first neural network model.

In order to enable the ego car to drive efficiently and safely on the highway while taking into account fuel consumption, the reward selected by the first neural network model includes the following aspects:

①.Whether there is a collision. If the ego vehicle collides with other vehicles or is out of the freeway drivable lane, a penalty of r _safe = -100 is imposed and the mission ends in failure. r _safe = 0 if no collision occurs.

②. Driving efficiency. For the execution of each action, the speed of the ego vehicle directly determines the speed of driving. Therefore, the choice of reward is proportional to the current speed

Among them, v is the speed of the vehicle, and v _des =30m/s represents the desired speed.

③. Lane change penalty. Although changing lanes is a prerequisite for overtaking, changing lanes too frequently is not good driving behavior. To limit the frequency of lane changes, each lane change decision is penalized with r _change = -1.

④. Fuel consumption. After each macro action is executed, by recording the strength curve of the speed, combined with the engine fuel consumption model of the company's truck, the fuel consumption p (L/km) during the execution of the action can be obtained. For the purpose of reducing fuel consumption, the corresponding report is r _oil = -p/100.

Combining all the above return definitions, the total return expression is

R=r _safe +r _v +r _change +r _oil (6)

The first neural network model of the embodiment of the present application adopts the DQN algorithm, and uses the maximum value of the action in the state at the next moment to update the action value at the current moment. The loss function is defined as:

L=(R+γ*argmax _a′ Q(s′,a′)-Q(s,a)) ² (7)

Among them, R is the return, γ is the loss factor, and Q(s, a) is the value of the action a selected in the state s. The gradient descent algorithm is used to optimize it, and the update of the training parameter w satisfies the

Among them, α is the learning rate, s represents a state quantity, which is the observation state corresponding to Table 1, and a represents the action, which is six discrete macroscopic decision-making behaviors.

Further, in some examples, after the prediction score of the next moment is determined, the driving mode of the sample vehicle at the next moment can also be determined according to the following method:

(1) determining the respective prediction scores corresponding to multiple driving strategies at the next moment at the current moment in the second preset number of consecutive moments through the initial neural network model;

(2) Taking the driving strategy with the largest predicted score as the driving strategy at the next moment, and obtaining the sample data of the next moment generated by the target sample vehicle driving under the driving strategy at the next moment;

(3) Continue to train the initial neural network model based on the sample data at the next moment.

At each moment in the DQN algorithm, each driving strategy is equivalent to a state, represented by s, and the different action values Q(s, a) in each state s are approximated by the first neural network model, so For a given "next moment" in the training sample, we can get the Q values corresponding to 6 actions, and then select the largest one as the iterative target updated under the "current moment". Although these values are initially given randomly, through constant iteration, it will eventually converge.

Only a fixed action value is updated each iteration, depending on the information recorded in the training samples. For example, in the training process, a sample records that the action a1 is selected in the state s, and the action value corresponding to this a1 may not be the largest. Different actions can gradually converge through iteration. There is a certain probability of selecting different actions during the training process, which is called exploration, because we do not know the size of the real value corresponding to each action at the beginning, and we need to go through a certain trial before we can evaluate it.

When the training is completed, the driving strategy is to select the 6 action values at the current moment, and then select the action corresponding to the maximum value as the driving strategy.

When the deep reinforcement learning agent is trained and used, first, the corresponding feature information is extracted through the driving characteristics of the relevant vehicle obtained by the perception module of the relevant vehicle and the road structure information given by the map positioning module. Then each action value at the current moment state is obtained through the neural network. The selection of the optimal action follows the following rules: First, sort according to the size of the action value, and give priority to the action corresponding to the maximum value. The state machine then checks whether the action can be successfully executed. If possible, execute the action, and if the transition condition of the state machine is not satisfied, select the action corresponding to the remaining one with the largest action value. The verification of the state machine is repeated until the condition is satisfied. If all actions are not satisfied, trigger the return instruction.

In some possible implementations, after determining the driving scores of a plurality of different driving strategies, a target driving strategy may be selected from the following methods: a driving strategy with a driving score higher than a preset score threshold is determined as a set of alternative driving strategies ; Determine the target driving strategy from the set of alternative driving strategies.

Specifically, it can be first determined whether the driving strategy with the highest driving score in the set of alternative driving strategies satisfies the preset safe driving conditions; if so, the driving strategy is determined as the target driving strategy; if not, the driving strategy is taken from the backup Deleted from the selected driving strategy collection.

It should be noted that the safe driving conditions in the embodiments of the present application can be judged by means of a state machine, and the state machine can be realized by using an existing decision-making and judgment algorithm. Make specific restrictions. For example, when our decision model gives an indication of a lane change, the detection of the state machine is triggered. The state machine judges whether the current distance from the rear vehicle meets the safety distance length, and whether the relative speed with the preceding vehicle will collide with a series of conditions. When all conditions are met, the decision instruction is considered executable. The state machine currently used in the present invention is relatively simple, and as a preliminary test, it is mainly used to verify the overall security of the decision.

In some possible implementations, if it is determined that the target driving strategy is changing lanes at a constant speed to the left lane or changing lanes at a constant speed to the right lane, the vehicle is controlled to change lanes by the following formula:

表示方向盘的转角，θ_n与θ_f分别表示目标车道上前方第一位置与前方第二位置与自车位置形成的夹角，Δt为时间步长，k_f、k_n和k_I表示驾驶行为的常数，目标车道为目标车辆准备从目标车辆所在的当前车道切换的车道。in,

Represents the steering wheel angle, θ _n and θ _f respectively represent the angle formed by the first position in front of the target lane and the second position in front and the position of the vehicle, Δt is the time step, k _f , k _n and k _I represent the driving behavior The constant of , the target lane is the lane that the target vehicle is ready to switch from the current lane where the target vehicle is located.

For ease of understanding, the embodiment of the present application also provides a vehicle automatic driving method in a practical application scenario, and the method specifically includes the following steps:

Step 1: Obtain the first speed v1 and first position d1 of the target vehicle at the current moment;

Step 2: Determine the relevant vehicle;

Relevant vehicles include: within 100m of the perception range of the target vehicle, the two closest vehicles in front of the current lane where the target vehicle is located, the two closest vehicles in front/rear in the left lane, and the two closest vehicles in front/rear in the right lane. 2 cars, 10 cars in total.

Step 3: Determine the relevant lane;

The relevant lanes include: the lane where the target vehicle is located, the right lane of the lane where the target vehicle is located, and the left lane of the lane where the target vehicle is located;

Step 4: Determine the feature 1 of the target vehicle and the feature 2 of the related vehicle;

The feature 1 of the target vehicle is the speed at the current moment and the distance from the center line of the lane.

The feature 2 of each vehicle in the related vehicles is that each vehicle obtains the position and speed at the current moment through radar and positioning information, and converts it into a relative position and relative speed relative to the own vehicle.

Step 5: Determine the feature 3 corresponding to the lane;

Feature 3 is that each lane is spaced at a distance of 1m, and the relative positions of the trajectory points of 80 center lines in front of 80 meters and the horizontal and vertical directions of the target vehicle are selected. The sizes of the above features are shown in Table 1.

Table 1

It should be noted that if there is no vehicle within a range of 100m in a certain direction, the corresponding relative position and speed information is filled with constant 0. For a lane beyond the boundary, the lane is considered to be full of vehicles and inaccessible. Therefore, the lateral relative position of the vehicle in the lane is the lane width, the longitudinal relative position is 0, and the relative speed is 0.

Step 6: Determine the driving strategy set

The driving strategy in this embodiment of the present application includes: accelerating straight ahead, maintaining speed straight ahead, decelerating straight ahead, straight line emergency braking, changing lanes to the left lane at a constant speed, and changing lanes to the right lane at a constant speed. Among them, the underlying control of linear acceleration and deceleration is realized by changing the control parameters of the IDM adaptive cruise algorithm in formulas (1) and (2), as shown in Table 2.

Table 2

longitudinal acceleration Longitudinal deceleration a 2.0m/s2 0.6m/s2 b 3.0m/s2 1.0m/s2 s₀ 0m/s 5m/s v_set 30m/s 19m/s T_set 1s 2s

The embodiment of the present application adopts a simplified 2-point control model to complete the control of the lane change. The specific formula is as follows:

表示方向盘的转角，θ_n与θ_f分别表示目标车道上前方50米与100米参考点与自车位置形成的夹角，Δt＝0.05为时间步长，k_f＝20，k_n＝10，k_I＝6表示驾驶行为的常数。On the basis of maintaining a constant speed, the lane changing operation is realized by changing the steering angle. in,

represents the steering wheel angle, θ _n and θ _f respectively represent the angle formed by the reference point 50 meters and 100 meters ahead on the target lane and the position of the vehicle, Δt=0.05 is the time step, k _f =20, k _n =10, k _I =6 represents a constant for driving behavior.

Step 7: Determine a driving strategy with a driving score higher than a preset score threshold as a set of alternative driving strategies;

Step 8: judging whether the driving strategy with the highest driving score in the alternative driving strategy set satisfies the preset safe driving conditions;

Step 9: If satisfied, determine the driving strategy as the target driving strategy;

Step 10: The target driving strategy is to change lanes at a constant speed to the left lane or change lanes at a constant speed to the right lane, and the target vehicle drives according to the target driving strategy at the next moment.

Specifically, the vehicle is controlled to change lanes by the following formula:

表示方向盘的转角，θ_n与θ_f分别表示目标车道上前方第一位置与前方第二位置与自车位置形成的夹角，Δt为时间步长，k_f、k_n和k_I表示驾驶行为的常数，目标车道为目标车辆准备从目标车辆所在的当前车道切换的车道。in,

Represents the steering wheel angle, θ _n and θ _f respectively represent the angle formed by the first position in front of the target lane and the second position in front and the position of the vehicle, Δt is the time step, k _f , k _n and k _I represent the driving behavior The constant of , the target lane is the lane that the target vehicle is ready to switch from the current lane where the target vehicle is located.

Based on the foregoing method embodiments, an embodiment of the present application further provides a vehicle automatic driving device, as shown in FIG. 6 , the device includes:

The information acquisition module 602 is used to acquire the first driving information of the target vehicle at the current moment, the second driving information corresponding to the relevant vehicle within the first specified range from the target vehicle, and the relevant lane within the second specified range from the target vehicle lane information;

The prediction module 604 is used for predicting the driving score of the target vehicle traveling with a plurality of different predetermined driving strategies through the first neural network model, the first driving information, the second driving information and the lane information; wherein, the driving score is used to characterize the target vehicle Driving conditions and vehicle performance for driving with a predetermined driving strategy; the first neural network model is obtained by training corresponding sample data under different traffic conditions;

The target driving strategy determination module 606 is configured to determine the target driving strategy from the plurality of different driving strategies according to the driving scores corresponding to the plurality of different driving strategies, so that the target vehicle drives according to the target driving strategy at the next moment of the current moment.

The above-mentioned vehicle automatic driving device provided by the embodiment of the present application first obtains the driving information of the target vehicle, the driving information of the relevant vehicle and the lane information, and then uses the first neural network model to predict the driving score, so as to obtain a plurality of driving strategies with different driving strategies. Score, and finally determine the target driving strategy according to the driving score. In the technology of the present application, a plurality of driving strategies are firstly determined through driving decisions, and then a neural network model is used to predict the score of each driving strategy, so that the final target driving strategy is based on the traditional decision-making algorithm. The network can effectively predict the driving performance of the target vehicle. Since the first neural network model is obtained by training sample data under a variety of different traffic conditions, the target driving strategy determined by the score predicted by the first neural network model is closer to the target. The actual complex traffic situation effectively improves the safety of vehicle automatic driving.

The above-mentioned related vehicles within a specified range from the target vehicle include at least one of the following: a first related vehicle whose distance from the target vehicle in the lane where the target vehicle is located is less than a first distance threshold and is located in front of the driving direction of the target vehicle; A second related vehicle in the side lane whose distance from the target vehicle is less than the second distance threshold and is located in front of the direction of travel of the target vehicle; in the lane on the right side of the lane where the target vehicle is located, the distance from the target vehicle is less than the third distance threshold and is located behind the direction of travel of the target vehicle the third related vehicle; the fourth related vehicle whose distance from the target vehicle in the left lane of the target vehicle’s lane is less than the fourth distance threshold and is located in front of the target vehicle’s driving direction; the distance from the target vehicle in the right lane of the target vehicle’s lane A fifth associated vehicle that is less than the fifth distance threshold and is located behind the target vehicle in the direction of travel.

The first driving information of the above-mentioned target vehicle at the current moment includes the first position and the first speed of the target vehicle; the above-mentioned information obtaining module 602 is also used to: obtain the second position and the second speed of the relevant vehicle at the current moment; calculate the second The relative position of the position with respect to the first position, and the relative speed of the second speed with respect to the first speed; the relative position and the relative speed are determined as the second travel information corresponding to the relevant vehicle.

The above device is also used for: if the lane where the target vehicle is located is the rightmost lane on the road, set the second driving information of the relevant vehicle in the right lane of the lane where the target vehicle is located to 0; if the lane where the target vehicle is located is the most right lane on the road Left lane, set the second driving information of related vehicles in the left lane of the lane where the target vehicle is located to 0.

The above-mentioned relevant lanes include the lane where the target vehicle is located, the right side lane of the lane where the target vehicle is located, and the left side lane of the lane where the target vehicle is located; the above-mentioned information acquisition module 602 is also used to: take the position of the vehicle at the current moment as the origin position; calculate the relevant lanes The horizontal distance and the vertical distance between each position in the first preset number of continuous positions and the origin position; wherein, the distance between every two adjacent positions in the first preset number of continuous positions is equal; A set consisting of a lateral distance and a longitudinal distance corresponding to each of the preset number of consecutive positions is determined as lane information of a relevant lane within a second specified range from the target vehicle.

The above multiple driving strategies include at least any two of the following: accelerating straight ahead, maintaining speed straight ahead, decelerating straight ahead, emergency braking in a straight line, changing lanes at a constant speed to the left lane, and changing lanes at a constant speed to the right lane.

The above-mentioned first neural network model is obtained through the following steps of training: obtaining sample data; wherein the sample data includes speed information and position information of the target sample vehicle, speed information and position information of the relevant sample vehicles related to the target sample vehicle, and the target sample vehicle. Lane information of the lane corresponding to the vehicle; calculate the loss value of the initial neural network model corresponding to each of the second preset number of consecutive moments according to the sample data and the loss function; wherein, the loss function includes collision parameters, energy consumption parameters, variable One or more of the road penalty parameter and the driving efficiency parameter; adjust the parameters of the initial neural network model according to the loss value, and determine the corresponding initial neural network model when the training stop condition is satisfied as the first neural network model.

The above device is also used for: determining the prediction scores corresponding to multiple driving strategies at the next moment at the current moment in the second preset number of consecutive moments through an initial neural network model; and using the driving strategy with the largest predicted score as the driving strategy at the next moment , and obtain the sample data of the next moment generated by the target sample vehicle driving under the driving strategy of the next moment; based on the sample data of the next moment, continue to train the initial neural network model.

The above-mentioned target driving strategy determination module 606 is further configured to: determine a driving strategy with a driving score higher than a preset score threshold as a set of alternative driving strategies; and determine a target driving strategy from the set of alternative driving strategies.

The above process of determining a target driving strategy from a set of alternative driving strategies includes: judging whether the driving strategy with the highest driving score in the set of alternative driving strategies satisfies a preset safe driving condition; if so, determining the driving strategy as the target driving strategy; if not satisfied, delete the driving strategy from the set of alternative driving strategies.

其中，

表示方向盘的转角，θ_n与θ_f分别表示目标车道上前方第一位置与前方第二位置与自车位置形成的夹角，Δt为时间步长，k_f、k_n和k_I表示驾驶行为的常数，目标车道为目标车辆准备从目标车辆所在的当前车道切换的车道。The above device is also used for: if the target driving strategy is to change lanes at a constant speed to the left lane or change lanes at a constant speed to the right lane, then control the vehicle to change lanes through the following formula:

in,

Represents the steering wheel angle, θ _n and θ _f respectively represent the angle formed by the first position in front of the target lane and the second position in front and the position of the vehicle, Δt is the time step, k _f , k _n and k _I represent the driving behavior The constant of , the target lane is the lane that the target vehicle is ready to switch from the current lane where the target vehicle is located.

The implementation principle and the technical effects of the vehicle automatic driving device provided by the embodiments of the present application are the same as those of the foregoing method embodiments. For the sake of brief description, for the parts not mentioned in the embodiments of the foregoing device, reference may be made to the foregoing vehicle automatic driving method implementation. corresponding content in the example.

In order to verify the feasibility of the first neural network model provided by the embodiment of the present application, the embodiment of the present application adopts the CARLA automatic driving simulation simulator to establish a driving environment of a highway. Among them, the initial speed distribution of the surrounding vehicles is 60km/h～120km/h, the initial speed of the own vehicle is 60km/h, and the maximum speed is 110km/h. The purpose is whether the model of the present invention can drive more quickly on the highway on the basis of safety, and can reduce fuel consumption at the same time.

The structure of 1000 sets of test road conditions shows that using the traditional rule-based car following model, the average time per kilometer is 44.76s and the fuel consumption is 0.211 liters. By adopting the vehicle automatic driving method provided by the embodiment of the present application, the self-vehicle can reasonably choose the timing of changing lanes and overtaking during the driving process, and no collision accident occurs. The average time spent was reduced by 21.27%, and the fuel consumption was reduced by 2.3%.

An embodiment of the present application also provides an electronic device, as shown in FIG. 7 , which is a schematic structural diagram of the electronic device, wherein the electronic device includes a processor 1501 and a memory 1502 , and the memory 1502 stores data that can be used by the processor 1501 Executed computer-executable instructions, the processor 1501 executes the computer-executable instructions to implement the above-mentioned vehicle automatic driving method.

In the embodiment shown in FIG. 7 , the electronic device further includes a bus 1503 and a communication interface 1504 , wherein the processor 1501 , the communication interface 1504 and the memory 1502 are connected through the bus 1503 .

The memory 1502 may include a high-speed random access memory (RAM, Random Access Memory), and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory. The communication connection between the network element of the system and at least one other network element is implemented through at least one communication interface 1504 (which may be wired or wireless), which may use the Internet, a wide area network, a local network, a metropolitan area network, and the like. The bus 1503 may be an ISA (IndustryStandard Architecture, industry standard architecture) bus, a PCI (Peripheral Component Interconnect, peripheral component interconnect standard) bus, or an EISA (Extended Industry Standard Architecture, extended industry standard architecture) bus, or the like. The bus 1503 can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one bidirectional arrow is used in FIG. 7, but it does not mean that there is only one bus or one type of bus.

The processor 1501 may be an integrated circuit chip with signal processing capability. In the implementation process, each step of the above-mentioned method may be completed by an integrated logic circuit of hardware in the processor 1501 or an instruction in the form of software. The above-mentioned processor 1501 may be a general-purpose processor, including a central processing unit (CPU for short), a network processor (NP for short), etc.; it may also be a digital signal processor (Digital Signal Processor, DSP for short) , Application Specific Integrated Circuit (ASIC for short), Field-Programmable Gate Array (FPGA for short) or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in conjunction with the embodiments of the present application may be directly embodied as executed by a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may be located in random access memory, flash memory, read-only memory, programmable read-only memory or electrically erasable programmable memory, registers and other storage media mature in the art. The storage medium is located in the memory, and the processor 1501 reads the information in the memory, and completes the steps of the vehicle automatic driving method in the foregoing embodiment in combination with its hardware.

Embodiments of the present application further provide a computer-readable storage medium, where computer-executable instructions are stored in the computer-readable storage medium, and when the computer-executable instructions are invoked and executed by a processor, the computer-executable instructions cause the processor to For the implementation of the above-mentioned vehicle automatic driving method, the specific implementation can refer to the foregoing method embodiments, which will not be repeated here.

The computer program product of the vehicle automatic driving method, device, and electronic device provided by the embodiments of the present application includes a computer-readable storage medium storing program codes, and the instructions included in the program codes can be used to execute the methods described in the foregoing method embodiments. The specific implementation can refer to the method embodiment, which is not repeated here.

The relative steps, numerical expressions and numerical values of the components and steps set forth in these embodiments do not limit the scope of the present application unless specifically stated otherwise.

The functions, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a processor-executable non-volatile computer-readable storage medium. Based on this understanding, the technical solution of the present application can be embodied in the form of a software product in essence, or the part that contributes to the prior art or the part of the technical solution. The computer software product is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes .

In the description of this application, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. The indicated orientation or positional relationship is based on the orientation or positional relationship shown in the accompanying drawings, which is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the indicated device or element must have a specific orientation or a specific orientation. construction and operation, and therefore should not be construed as limitations on this application. Furthermore, the terms "first", "second", and "third" are used for descriptive purposes only and should not be construed to indicate or imply relative importance.

Finally, it should be noted that the above-mentioned embodiments are only specific implementations of the present application, and are used to illustrate the technical solutions of the present application, rather than limit them. The embodiments describe the application in detail, and those of ordinary skill in the art should understand that: any person skilled in the art can still modify the technical solutions described in the foregoing embodiments within the technical scope disclosed in the application. Or can easily think of changes, or equivalently replace some of the technical features; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions in the embodiments of the application, and should be covered in this application. within the scope of protection. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims (14)

1. a vehicle automatic driving method, is characterized in that, described method comprises: Obtain the first driving information of the target vehicle at the current moment, the second driving information corresponding to the relevant vehicle within the first specified range from the target vehicle, and the lane information of the relevant lane within the second specified range from the target vehicle ; The driving score of the target vehicle when the target vehicle travels with multiple different predetermined driving strategies is predicted by using the first neural network model, the first driving information, the second driving information and the lane information; wherein, the driving score is determined by is used to characterize the driving conditions and vehicle performance of the target vehicle traveling with the predetermined driving strategy; the first neural network model is obtained by training corresponding sample data under different traffic conditions; A target driving strategy is determined from the plurality of different driving strategies according to the driving scores corresponding to the plurality of different driving strategies, so that the target vehicle drives according to the target driving strategy at the next moment of the current moment. 2 . The method according to claim 1 , wherein the relevant vehicles within a specified range from the target vehicle include at least one of the following: 2 . a first related vehicle in the lane where the target vehicle is located and whose distance from the target vehicle is less than a first distance threshold and is located in front of the driving direction of the target vehicle; a second related vehicle in the right lane of the lane where the target vehicle is located, the distance from the target vehicle being less than a second distance threshold and ahead of the direction of travel of the target vehicle; A third related vehicle whose distance from the target vehicle is less than a third distance threshold and located behind the target vehicle in the direction of travel in the right lane of the lane where the target vehicle is located; a fourth related vehicle whose distance from the target vehicle is less than a fourth distance threshold in the left lane of the lane where the target vehicle is located and is located in front of the direction of travel of the target vehicle; A fifth related vehicle whose distance from the target vehicle is less than a fifth distance threshold and located behind the target vehicle in the direction of travel in the right lane of the lane where the target vehicle is located. 3. The method according to claim 1, wherein the first driving information of the target vehicle at the current moment comprises a first position and a first speed of the target vehicle; The step of acquiring the second driving information corresponding to the relevant vehicle within a specified range of the target vehicle includes: obtaining the second position and second speed of the relevant vehicle at the current moment; calculating the relative position of the second position with respect to the first position, and the relative velocity of the second velocity with respect to the first velocity; The relative position and the relative speed are determined as second travel information corresponding to the relevant vehicle. 4. The method according to claim 3, wherein the method further comprises: If the lane where the target vehicle is located is the rightmost lane in the road, set the second driving information of the relevant vehicle in the lane on the right side of the lane where the target vehicle is located to 0; If the lane where the target vehicle is located is the leftmost lane in the road, the second travel information of the relevant vehicle located in the left lane of the lane where the target vehicle is located is set to 0. 5. The method according to claim 1, wherein the relevant lanes comprise a lane where the target vehicle is located, a right lane of the lane where the target vehicle is located, and a left lane of the lane where the target vehicle is located; The step of acquiring lane information of a relevant lane within a second specified range from the target vehicle includes: Taking the position of the vehicle at the current moment as the origin position; Calculate the lateral distance and the longitudinal distance between each position in the first preset number of consecutive positions in the relevant lane and the origin position; wherein, in the first preset number of consecutive positions, every two adjacent positions the distance between them is equal; The set formed by the lateral distance and the longitudinal distance corresponding to each of the first preset number of consecutive positions is determined as lane information of a relevant lane within a second specified range from the target vehicle. 6. The method of claim 1, wherein the plurality of different driving strategies include at least any two of the following: Accelerate straight, maintain speed straight, slow down straight, emergency braking in a straight line, uniform lane change to the left lane, and uniform lane change to the right lane. 7. The method according to claim 1, wherein the first neural network model is obtained by training through the following steps: Obtain sample data; wherein, the sample data includes speed information and position information of a target sample vehicle, speed information and position information of a relevant sample vehicle related to the target sample vehicle, and a lane of a lane corresponding to the target sample vehicle information; Calculate the loss value of the initial neural network model corresponding to each of the second preset number of consecutive moments according to the sample data and the loss function; wherein the loss function includes collision parameters, energy consumption parameters, lane change penalty parameters, and driving efficiency one or more of the parameters; The parameters of the initial neural network model are adjusted according to the loss value, and the corresponding initial neural network model when the training stop condition is satisfied is determined as the first neural network model. 8. The method according to claim 7, wherein the method further comprises: Determine, by using the initial neural network model, the prediction scores corresponding to the plurality of different driving strategies at the next moment of the current moment in the second preset number of consecutive moments; Taking the driving strategy with the largest predicted score as the driving strategy at the next moment, and acquiring the sample data of the next moment generated by the target sample vehicle running with the driving strategy at the next moment; Based on the sample data at the next moment, continue to train the initial neural network model. 9. The method according to claim 1, wherein the step of determining a target driving strategy from the plurality of different driving strategies according to the driving scores corresponding to the plurality of different driving strategies respectively comprises: determining the driving strategy with the driving score higher than the preset score threshold as a set of alternative driving strategies; A target driving strategy is determined from the set of candidate driving strategies. 10. The method according to claim 9, wherein the step of determining a target driving strategy from the set of alternative driving strategies comprises: Judging whether the driving strategy with the highest driving score in the candidate driving strategy set satisfies the preset safe driving condition; If satisfied, determine the driving strategy as the target driving strategy; If not, the driving strategy is deleted from the set of alternative driving strategies. 11. The method according to any one of claims 1-10, wherein the method further comprises: If the target driving strategy is changing lanes at a constant speed to the left lane or changing lanes at a constant speed to the right lane, the vehicle is controlled to change lanes by the following formula:

in,

Represents the steering wheel angle, θ _n and θ _f respectively represent the angle formed by the first position in front of the target lane and the second position in front and the position of the vehicle, Δt is the time step, k _f , k _n and k _I represent the driving behavior The target lane is the lane where the target vehicle is prepared to switch from the lane where the target vehicle is located. 12. A vehicle automatic driving device, characterized in that the device comprises: The information acquisition module is used to acquire the first driving information of the target vehicle at the current moment, the second driving information corresponding to the relevant vehicles within the first specified range from the target vehicle, and the second specified range from the target vehicle. Lane information of the relevant lane; a prediction module, configured to predict the driving score of the target vehicle traveling with a plurality of different predetermined driving strategies by using the first neural network model, the first driving information, the second driving information and the lane information; wherein, The driving score is used to characterize the driving conditions and vehicle performance of the target vehicle traveling with the predetermined driving strategy; the first neural network model is obtained by training corresponding sample data under different traffic conditions; A target driving strategy determination module, configured to determine a target driving strategy from the plurality of different driving strategies according to the driving scores corresponding to the plurality of different driving strategies, so that the target vehicle is at the next moment of the current moment Drive according to the target driving strategy. 13. An electronic device, comprising a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, the processor executing the computer-executable instructions to implement the claims The method of any one of 1-11. 14. A computer-readable storage medium, characterized in that the computer-readable storage medium stores computer-executable instructions that, when invoked and executed by a processor, cause the processor to Implementing the method of any of claims 1-11.

Images