Vehicle trajectory control method, device and storage medium

Abstract

The present application provides a vehicle trajectory control method, device, and storage medium. The vehicle trajectory control method includes: obtaining the real-time driving status of a target vehicle; determining the target risk level of the target vehicle based on the real-time driving status; determining a trajectory control model corresponding to the target risk level, wherein the trajectory control models are multiple and different trajectory control models correspond to different risk levels; and determining driving parameters output by the trajectory control model based on the real-time driving status, so as to control the driving status of the target vehicle according to the driving parameters. The above method can select the corresponding trajectory control model according to different risk levels to control the vehicle's driving status, thereby ensuring that the vehicle maintains both safety and efficiency during operation.

Description

Technical Field

[0001] The present application relates to the field of vehicle trajectory control technology, and in particular to a vehicle trajectory control method, device, and storage medium. Background Art

[0002] With the advancement of artificial intelligence, communications technology, and intelligent transportation systems, autonomous vehicles are gradually entering our horizons. According to relevant statistics, as many as 120 million autonomous vehicles will be sold worldwide by 2030, with China accounting for approximately one-third of these vehicles. This indicates that the era of hybrid human-machine driving is approaching, posing new challenges to traffic safety. To ensure the safe operation of autonomous vehicles on the road, some technology companies typically focus solely on safety, conducting driving tests on closed or open roads, quantifying risk scenarios, and developing corresponding control methods. However, these tests are often too simplistic and fail to fully cover the complex and ever-changing real-world traffic scenarios. Furthermore, overemphasizing the safety of individual vehicles while ignoring the driving behavior of surrounding vehicles generally compromises traffic efficiency, increases energy consumption, and thus increases travel costs.

[0003] Existing technologies are insufficient in extracting and depicting complex vehicle-following scenarios, and have not established alternative traffic accident indicators that can accurately quantify the risk level of the scenarios. Therefore, when faced with a traffic environment with complex and changing risk levels, existing technologies lack methods that can adopt optimal driving conditions at different risk levels. Summary of the Invention

[0004] In response to the above-mentioned technical problems existing in the prior art, the present application provides a vehicle trajectory control method, device and storage medium.

[0005] An embodiment of the present application provides a vehicle trajectory control method, including steps S101 to S104.

[0006] Step S101: Acquire the real-time driving status of the target vehicle.

[0007] Step S102: determining the target risk level of the target vehicle based on the real-time driving status.

[0008] Step S103: determining a trajectory control model corresponding to the target risk level according to the target risk level, wherein there are multiple trajectory control models, and different trajectory control models correspond to different risk levels.

[0009] Step S104: determining a driving parameter output by the trajectory control model based on the real-time driving state, so as to control the driving state of the target vehicle according to the driving parameter.

[0010] In some embodiments, determining the driving parameters output by the trajectory control model based on the real-time driving state specifically includes:

[0011] determining, via the trajectory control model, a risk level of an adjacent lane adjacent to the target vehicle;

[0012] Comparing the risk level of the adjacent lane with the target risk level to obtain the comparison result;

[0013] The driving parameters of the target vehicle, including a lateral driving parameter, are determined based on the comparison result.

[0014] In some embodiments, the trajectory control model is obtained by the following method:

[0015] Acquire a database, the database comprising at least trajectory data of a plurality of vehicles in a time series;

[0016] Classifying the trajectory data in the database into risk levels to obtain different risk level groups;

[0017] The vehicle following data included in the different risk level groups are calculated based on a machine learning method to obtain trajectory control models corresponding to different risk levels.

[0018] In some embodiments, the risk classification of the trajectory data in the database to obtain different risk level groups specifically includes:

[0019] Building a vehicle following scenario library comprising a plurality of sub-scenarios based on the database, each of the sub-scenario comprising at least vehicle following data of a vehicle following pair within a preset time;

[0020] Determine the risk level of each of the sub-scenarios in the vehicle following scenario library, and divide each of the sub-scenarios into different risk level groups based on the risk level.

[0021] In some embodiments, determining the target risk level of the target vehicle based on the real-time driving status specifically includes:

[0022] The target risk level of the target vehicle is calculated using a collision conflict safety index through a machine learning method; wherein the collision conflict safety index is used to assess the collision risk during vehicle driving, and the collision conflict safety index is configured to be calculated using the following formula:

[0023]

[0024] in, is the collision safety index; x r is the relative position of the front and rear vehicles in the following direction; dp is the braking distance of the following vehicle to center the vehicle in front; d f is the braking distance of the following vehicle to the center of the following vehicle; v p (t) is the speed of the vehicle in front of the following vehicle at time t; v f (t) is the speed of the following vehicle at time t; t reaction is the reaction time of the driver of the following car; p is the acceleration of the following car centering the front car; a f It is the acceleration of the following car relative to the car behind.

[0025] In some embodiments, calculating the target risk level of the target vehicle using a collision safety index through a machine learning method specifically includes:

[0026] The target risk level of the target vehicle is calculated using the following formula via machine learning:

[0027]

[0028] Among them, w IN and w M There are two types of parameters related to the risk of the vehicle, w IN +w M =1; It indicates the value of the standard safety index when the front vehicle decelerates at normal braking acceleration and the rear vehicle decelerates at maximum braking acceleration; It is the value of the standard safety indicator when both the front and rear vehicles in a following vehicle are decelerating at the maximum braking deceleration.

[0029] In some embodiments, the vehicle trajectory control method further includes:

[0030] The reward function in the machine learning method is calculated using a first boundary speed and a second boundary speed; wherein the first boundary speed is calculated using the following formula:

[0031]

[0032] The second boundary velocity is calculated using the following formula:

[0033]

[0034] Among them, x vr =x r -1.5m, m is meter; a partialbraking represents the vehicle acceleration under partial braking conditions; a fullbraking Indicates vehicle acceleration under full braking, with partial braking providing a comfortable braking force and full braking providing maximum braking force.

[0035] In some embodiments, the use of the first boundary speed and the second boundary speed to calculate the reward function in the machine learning method specifically includes:

[0036] The safety reward function included in the reward function is calculated based on the first boundary speed and the second boundary speed using the following formula:

[0037]

[0038] An embodiment of the present application further provides a vehicle trajectory control device, comprising an acquisition module and a determination module. The acquisition module is configured to acquire the real-time driving status of a target vehicle. The determination module is configured to determine the target risk level of the target vehicle based on the real-time driving status; and, based on the target risk level, determine a trajectory control model corresponding to the target risk level, wherein there are multiple trajectory control models, and different trajectory control models correspond to different risk levels; and determine driving parameters output by the trajectory control model based on the real-time driving status, so as to control the driving status of the target vehicle according to the driving parameters.

[0039] An embodiment of the present application further provides a storage medium storing a computer program, which implements the vehicle trajectory control method when executed by a processor.

[0040] Compared with the prior art, the beneficial effect of the embodiments of the present application is that: the present application obtains the real-time driving status of the target vehicle, and determines the current target risk level of the target vehicle based on the actual driving status, and then determines the corresponding trajectory control model according to the target risk level after determining the target risk level. Therefore, the corresponding trajectory control model can be selected according to different risk levels to control the vehicle driving status, thereby ensuring that the vehicle takes into account both safety and efficiency during operation. Especially for traffic environments with complex and changeable risk levels, different trajectory control models can be used at different risk levels to output driving parameters for controlling the driving status of the target vehicle, thereby achieving the purpose of enabling the target vehicle to travel in the optimal driving state. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In the drawings, which are not necessarily drawn to scale, like reference numerals may describe similar components in different views. The drawings illustrate various embodiments generally by way of example and not limitation, and together with the description and claims, serve to illustrate the disclosed embodiments. Where appropriate, like reference numerals are used throughout the drawings to refer to like or similar parts. Such embodiments are illustrative and are not intended to be exhaustive or exclusive of the embodiments of the present apparatus or method.

[0042] Figure 1This is a first flow chart of the vehicle trajectory control method according to an embodiment of the present application;

[0043] Figure 2 This is a second flow chart of the vehicle trajectory control method according to an embodiment of the present application;

[0044] Figure 3 This is a third flow chart of the vehicle trajectory control method according to an embodiment of the present application;

[0045] Figure 4 Schematic diagram of constructing a vehicle-following scenario library using a sliding window method in an embodiment of the present application;

[0046] Figure 5 A schematic diagram of a curve of the safety reward function according to an embodiment of the present application;

[0047] Figure 6 This is a structural block diagram of the vehicle trajectory control device according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] It should be understood that various modifications may be made to the embodiments of the invention herein. Therefore, the above description should not be considered as limiting, but merely as an example of an embodiment. Other modifications within the scope and spirit of the present application will occur to those skilled in the art.

[0049] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the present application and, together with the general description of the present application given above and the detailed description of the embodiments given below, serve to explain the principles of the present application.

[0050] These and other characteristics of the present application will become apparent from the following description of a preferred form of embodiment given as a non-limiting example with reference to the accompanying drawings.

[0051] It should also be understood that although the present application has been described with reference to certain specific examples, those skilled in the art will readily be able to implement many other equivalent forms of the present application.

[0052] The above and other aspects, features and advantages of the present application will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings.

[0053] Specific embodiments of the present application will be described hereinafter with reference to the accompanying drawings; however, it should be understood that the embodiments described are merely examples of the present application and may be implemented in a variety of ways. Well-known and / or repetitive functions and structures are not described in detail to avoid obscuring the present application with unnecessary or redundant details. Therefore, the specific structural and functional details described herein are not intended to be limiting, but rather to serve as a basis and representative basis for the claims to teach those skilled in the art to variously utilize the present application with substantially any suitable detailed structure.

[0054] This specification may use the phrases "in one embodiment," "in another embodiment," "in yet another embodiment," or "in other embodiments," which may all refer to one or more of the same or different embodiments according to the present application.

[0055] An embodiment of the present application provides a vehicle trajectory control method. The vehicle trajectory control method can be applied to a vehicle trajectory control device. The vehicle trajectory control device can be installed in a vehicle, and the vehicle can achieve automatic driving through the vehicle trajectory control device.

[0056] like Figure 1 As shown, the vehicle trajectory control method includes steps S101 to S104.

[0057] Step S101: Acquire the real-time driving status of the target vehicle.

[0058] Optionally, the target vehicle may be understood as a vehicle traveling on a lane, specifically any vehicle traveling on the lane, that is, the vehicle trajectory control method may be applied to any vehicle on the lane to achieve control of the driving parameters of the vehicle on the lane.

[0059] Optionally, the real-time driving status may be related to one or more of the following information: driving parameter information of the target vehicle itself, driving parameter information of vehicles surrounding the target vehicle, driving trajectory information of the target vehicle, and relative driving relationship information between the target vehicle and surrounding vehicles. The driving parameters may include at least one or more of the following: driving speed and driving acceleration; the driving trajectory may include at least one or more of the following: the driving position and driving route of the target vehicle; and the relative driving relationship information between the target vehicle and surrounding vehicles may include at least one or more of the following: relative position, relative speed, and relative acceleration between the target vehicle and surrounding vehicles.

[0060] Step S102: determining the target risk level of the target vehicle based on the real-time driving status.

[0061] Optionally, the target risk level may be understood as the risk level of the target vehicle traveling in the lane, and may specifically include the risk level of the current lane in which the target vehicle is traveling.

[0062] Optionally, the aforementioned risk levels may be divided into multiple categories based on risk level. This application does not specifically limit the number of risk levels. The following description uses the example of risk levels including at least high risk, medium risk, and low risk. The vehicle trajectory control device may determine, based on the real-time driving status of the target vehicle, the risk level of the current lane in which the target vehicle is located as a target risk level.

[0063] Step S103: determining a trajectory control model corresponding to the target risk level according to the target risk level, wherein there are multiple trajectory control models, and different trajectory control models correspond to different risk levels.

[0064] Optionally, different trajectory control models can be determined corresponding to different risk levels. For example, when the target risk level is determined to be high risk, a trajectory control module corresponding to high risk can be determined; when the target risk level is determined to be medium risk, a trajectory control module corresponding to medium risk can be determined; when the target risk level is determined to be low risk, a trajectory control module corresponding to low risk can be determined.

[0065] Optionally, the trajectory control module can be trained by a machine learning method, which will be described in detail below and will not be repeated here.

[0066] Step S104: determining a driving parameter output by the trajectory control model based on the real-time driving state, so as to control the driving state of the target vehicle according to the driving parameter.

[0067] In this way, the output data of the trajectory control model corresponding to the risk level can be used to control the driving state of the target vehicle. For example, the target vehicle can be controlled to enter a driving state that includes at least acceleration, deceleration, lane changing, etc. according to the driving parameters output by the trajectory control model.

[0068] Optionally, the above-mentioned driving parameters may include at least driving speed and driving acceleration, and the driving acceleration may include at least lateral acceleration and / or longitudinal acceleration, wherein, if the lateral acceleration is 0, it indicates that the target vehicle does not change lanes; if the lateral acceleration is greater than 0, it indicates that the target vehicle changes lanes.

[0069] The present application obtains the real-time driving status of the target vehicle, and determines the current target risk level of the target vehicle based on the actual driving status, and then determines the corresponding trajectory control model according to the target risk level after determining the target risk level. Therefore, the corresponding trajectory control model can be selected according to different risk levels to control the vehicle's driving status, thereby ensuring that the vehicle takes into account both safety and efficiency during operation. Especially for traffic environments with complex and changing risk levels, different trajectory control models can be used at different risk levels to output driving parameters for controlling the driving status of the target vehicle, thereby achieving the purpose of enabling the target vehicle to travel in the optimal driving state.

[0070] In some embodiments, as Figure 2 As shown, step S104 of determining the driving parameters output by the trajectory control model based on the real-time driving state specifically includes steps S201 to S203.

[0071] Step S201: Determine the risk level of an adjacent lane adjacent to the target vehicle via the trajectory control model.

[0072] Step S202: Compare the risk level of the adjacent lane with the target risk level to obtain the comparison result.

[0073] Step S203: Determine the lateral driving parameter included in the driving parameters of the target vehicle based on the comparison result.

[0074] In this way, by comparing the risk level of the adjacent lanes with the target risk level, it can be determined whether to control the target vehicle to perform a lane change operation, thereby improving the safety and efficiency of the vehicle during operation.

[0075] Optionally, the above-mentioned lateral driving parameters may include at least lateral acceleration and may also include lateral displacement distance.

[0076] Optionally, the risk level of the adjacent lane adjacent to the target vehicle can be understood as the larger risk level between the risk level of the target vehicle and the risk level of the vehicle behind the target vehicle after the target vehicle is controlled to change lanes, that is, the higher risk level between the risk level of the target vehicle and the risk level of the vehicle behind the target vehicle is the risk level of the adjacent lane.

[0077] Optionally, after comparing the risk level of the adjacent lane with the target risk level to obtain the comparison result, the lateral driving parameter determined by the comparison result may indicate whether the target vehicle performs a lane change operation.

[0078] For example, when it is determined that the risk level of an adjacent lane adjacent to the target vehicle is lower than the risk level of the current lane in which the target vehicle is located, the driving parameters output by the trajectory control model may include a lateral acceleration, which is greater than 0, and the target vehicle will perform a lane change operation based on the lateral acceleration.

[0079] Optionally, if the driving parameters of the target vehicle include lateral driving parameters determined based on the comparison results, and the target vehicle performs a lane change operation, the target vehicle can perform the lane change operation using a polynomial curve model. Of course, other mathematical models can also be used to perform the lane change operation. This application does not make specific limitations on this.

[0080] Optionally, after executing step S203, steps S201 to S203 may be executed again after the next time step. The duration of the time step may be set based on experience or determined based on the real-time driving state of the target vehicle, and this application does not make any specific restrictions on this.

[0081] In some embodiments, as Figure 3 As shown, the trajectory control model steps S301 to S303 are obtained using the following method.

[0082] Step S301: Acquire a database, where the database at least includes trajectory data of multiple vehicles in a time series.

[0083] Step S302: dividing the trajectory data in the database into risk levels to obtain groups of different risk levels.

[0084] Step S303: Calculating the vehicle following data included in the different risk level groups based on a machine learning method to obtain trajectory control models corresponding to different risk levels.

[0085] In this way, the model can be trained using a large amount of vehicle trajectory data in the database, so that the trajectory control model trained by machine learning methods can ensure that the vehicle drives safely, efficiently and energy-efficiently in driving scenarios with complex and variable risk levels.

[0086] Optionally, the database may include at least an NGSIM (Next Generation Simulation) dataset, which is US highway driving data and includes the driving conditions of all vehicles on roads such as US101 and I-80 over a period of time. Of course, the database may also include driving datasets containing trajectory data of multiple vehicles within other time series. This application does not impose specific limitations on this. The database may contain a large amount of data and fully encompass vehicle trajectories under a wider range of road conditions.

[0087] Optionally, after obtaining the database, in order to avoid the problem of inaccurate data in the database, abnormal data in the database can be removed, and the trajectory curve of each vehicle can be smoothed using a moving average algorithm so that the vehicle speed and acceleration in the trajectory curve can be within a normal range.

[0088] Optionally, trajectory data of different vehicles may have different risk levels. After determining the risk level of each trajectory data, trajectory data of the same risk level may be grouped into the same risk level group. For example, after determining that the risk levels of multiple trajectory data are all low risk, the multiple trajectory data may be grouped into the low risk level group.

[0089] Optionally, the above-mentioned machine learning method may be calculated using the DDPG algorithm, and of course other deep reinforcement learning algorithms may also be used, which is not specifically limited in this application. By training the model with the vehicle-following data included in different risk level groups, trajectory control models corresponding to different risk levels can be obtained. Therefore, when controlling the automatic driving of the target vehicle, the trajectory control model adapted to the target vehicle can be used to control the driving state of the target vehicle.

[0090] Optionally, the trajectory data may include trajectory data for a large number of vehicles traveling on a lane. Adjacent vehicles on a lane may form a following vehicle pair. The trajectory data includes data for multiple following vehicle pairs traveling on the lane, i.e., following vehicle data. The following vehicle data includes at least driving data for the leading vehicle and the following vehicle.

[0091] In some embodiments, step S302 classifies the trajectory data in the database into risk levels to obtain different risk level groups, specifically including:

[0092] Building a vehicle following scenario library comprising a plurality of sub-scenarios based on the database, each of the sub-scenario comprising at least vehicle following data of a vehicle following pair within a preset time;

[0093] Determine the risk level of each of the sub-scenarios in the vehicle following scenario library, and divide each of the sub-scenarios into different risk level groups based on the risk level.

[0094] In this way, the risk level of each sub-scenario can be determined by the following vehicle data of the following vehicle pairs contained in each sub-scenario, and the model can be trained based on the machine learning method and the data included in the sub-scenario of different risk levels to obtain the trajectory control model corresponding to the wind turbines with different risks.

[0095] Alternatively, as Figure 4 As shown, a sliding window method can be used to construct a following scenario library. The specific method is as follows: For a following pair with a length of time T = [t0, t1], starting from the trajectory start time t0, a continuous trajectory of length Δ is selected, that is, the trajectory of the following pair for the time period [t0, t0+Δ]. Starting from t0+δ, the trajectory data of the subsequent sliding time window is obtained, that is, the continuous vehicle trajectory of the following pair within the time period [t0+δ, t0+δ+Δ]. Then, using the same method, the time window is continuously sliding until the tail time of the time window exceeds t1. For example, δ = 3s and Δ = 5s.

[0096] In some embodiments, determining the target risk level of the target vehicle based on the real-time driving status in step S102 specifically includes:

[0097] The target risk level of the target vehicle is calculated using a collision conflict safety index through a machine learning method; wherein the collision conflict safety index is used to evaluate the collision risk during vehicle driving, and the collision conflict safety index is configured to be calculated using the following formula 1.1.

[0098]

[0099] in, is the collision safety index; x ris the relative position of the front and rear vehicles in the following direction; d p is the braking distance of the following vehicle to center the vehicle in front; d f is the braking distance of the following vehicle to the center of the following vehicle; v p (t) is the speed of the vehicle in front of the following vehicle at time t; v f (t) is the speed of the following vehicle at time t; t reaction is the reaction time of the driver of the following car; p is the acceleration of the following car centering the front car; a f It is the acceleration of the following car relative to the car behind.

[0100] In this way, the potential risk of collision between the front and rear vehicles in the following car can be quantified by the above formula 1.1, which enables an accurate judgment of the risk of collision between the front and rear vehicles in the following car.

[0101] Optionally, t reaction The SCR value range is (0,1). A larger SCR value indicates a higher risk of collision between the preceding and following vehicles.

[0102] Optionally, the risk level of the vehicle at time t can be calculated using the following formula 1.2.

[0103]

[0104] Among them, RISK(t) is the risk level of the vehicle at time t, which can be understood as a linear combination of two collision safety indicators, where Indicates the value of the collision conflict safety index when the leading vehicle decelerates at normal braking acceleration and the following vehicle decelerates at maximum braking acceleration in a following vehicle collision. It is the value of the collision conflict safety index when the leading vehicle and the following vehicle are both decelerating at the maximum braking deceleration in the following vehicle. w1 and w2 are both coefficients, which can be set based on experience. This application does not specifically limit the specific value of the coefficient.

[0105] In some embodiments, calculating the target risk level of the target vehicle using a collision safety index through a machine learning method specifically includes:

[0106] The target risk level of the target vehicle is calculated using the following formula 1.3 via a machine learning method.

[0107]

[0108] Among them, w IN and w M There are two types of parameters related to the risk of the vehicle, w IN +w M =1; Indicates the value of the collision conflict safety index when the leading vehicle decelerates at normal braking acceleration and the following vehicle decelerates at maximum braking acceleration in a following vehicle collision. It is the value of the collision conflict safety index when the front and rear vehicles in the following vehicle are decelerating at the maximum braking deceleration.

[0109] In this way, the total risk level within the Δ time period can be accurately assessed, and the risk level of each sub-scenario can be defined, that is, the collision conflict safety index within the above-mentioned 5-s time window, so that the level of all sub-scenarios in the following vehicle scenario library can be divided through the above method.

[0110] In some embodiments, the vehicle trajectory control method further includes: using a first boundary speed and a second boundary speed to calculate a reward function in a machine learning method; wherein the first boundary speed is calculated using the following formula 1.4:

[0111]

[0112] The second boundary velocity is calculated using the following formula 1.5:

[0113]

[0114] Among them, x vr =x r -1.5m, m is meter; a partialbraking represents the vehicle acceleration under partial braking conditions; a fullbraking Indicates vehicle acceleration under full braking, with partial braking providing a comfortable braking force and full braking providing maximum braking force.

[0115] Optionally, the first boundary speed can be understood as when the rear vehicle is traveling at the first boundary speed, the front vehicle suddenly uses a fullbraking After full braking, the following vehicle uses a partialbraking Partial braking is performed, and the front and rear vehicles can stop without collision and are 1.5m apart; the second boundary speed can be understood as when the rear vehicle is traveling at the second boundary speed, the front vehicle uses a partialbraking Perform partial braking, and the following vehicle uses a fullbraking After full braking, the two vehicles can stop without collision and are 1.5m (meters) apart.

[0116] In some embodiments, the use of the first boundary speed and the second boundary speed to calculate the reward function in the machine learning method specifically includes: using the following formula 1.6 to calculate the safety reward function included in the reward function based on the first boundary speed and the second boundary speed.

[0117]

[0118] The above safety reward function can promote the vehicle to travel at a speed close to the first boundary speed and away from the second boundary speed, so as to ensure that the vehicle switches between safe driving and efficient driving. Figure 5 , Figure 5 What is shown in FIG is a curve diagram of the safety reward function.

[0119] In some embodiments, the above reward function may also include a comfort reward function and a collision avoidance reward function. The comfort reward function is mainly aimed at the smoothness of the vehicle trajectory, thereby deriving the time derivative of the vehicle acceleration, ie Where a(t) is the vehicle's acceleration at time t; a(t-1) is the vehicle's acceleration at time t-1; and ΔT is the time interval. Reducing jerk can improve passenger comfort. The comfort reward value set by the comfort reward function can be calculated using the following formula 1.7.

[0120]

[0121] Furthermore, the anti-collision reward function can be calculated using the following formula 1.8.

[0122]

[0123] Based on the above, the reward function can be calculated using the following formula 1.9.

[0124] r=w S r SCR +w J r Jerk +r collision Formula (1.9);

[0125] Among them, w S and w J are the parameters of the reward function.

[0126] In some embodiments, when the above-mentioned machine learning method adopts the DDPG algorithm for calculation, two corresponding network models can be constructed according to the divided risk levels, wherein the first network model can be understood as a critic network model (Critic network), and the second network model can be understood as an actor network model (Actor network). Of course, it can also be other network models. This application does not make specific limitations on this. The following is an example of the first network model being the critic network model and the second network model being the actor network model.

[0127] Further, if Figure 6The figure shows a schematic diagram of the critic network model and actor network model for a high-risk state. Both the critic network model and the actor network model consist of an input layer, a hidden layer, and an output layer. For the actor network model, the input layer includes the distance between the following vehicles, the relative speed of the following vehicles, and the speed of the following vehicle; the output layer is the acceleration of the following vehicle. Compared to the policy network, the input layer of the critic network model includes the acceleration of the following vehicle, and the output layer is the value function.

[0128] Using the vehicle-following data included in each sub-scenario in the vehicle-following scenario library, multiple trajectory control models for different risk levels can be trained. Specifically, the DDPG algorithm can be used to update the parameters of the critic network and actor network to train multiple sets of network parameters for the critic network model and the actor network model. These sets of network parameters for the critic network model and the actor network model can represent multiple trajectory control models trained using sub-scenarios with different risk levels.

[0129] The embodiment of the present application also provides a vehicle trajectory control device 110. Figure 6 As shown, the vehicle trajectory control device 110 includes an acquisition module 101 and a determination module 102. The acquisition module 101 is configured to acquire the real-time driving state of a target vehicle. The determination module 102 is configured to determine the target risk level of the target vehicle based on the real-time driving state; determine a trajectory control model corresponding to the target risk level based on the target risk level, wherein the trajectory control models may be multiple, with different trajectory control models corresponding to different risk levels; and determine driving parameters output by the trajectory control model based on the real-time driving state, so as to control the driving state of the target vehicle according to the driving parameters.

[0130] The present application obtains the real-time driving status of the target vehicle, and determines the current target risk level of the target vehicle based on the actual driving status, and then determines the corresponding trajectory control model according to the target risk level after determining the target risk level. Therefore, the corresponding trajectory control model can be selected according to different risk levels to control the vehicle's driving status, thereby ensuring that the vehicle takes into account both safety and efficiency during operation. Especially for traffic environments with complex and changing risk levels, different trajectory control models can be used at different risk levels to output driving parameters for controlling the driving status of the target vehicle, thereby achieving the purpose of enabling the target vehicle to travel in the optimal driving state.

[0131] In some embodiments, the determining module 102 is further configured to:

[0132] determining, via the trajectory control model, a risk level of an adjacent lane adjacent to the target vehicle;

[0133] Comparing the risk level of the adjacent lane with the target risk level to obtain the comparison result;

[0134] The driving parameters of the target vehicle, including a lateral driving parameter, are determined based on the comparison result.

[0135] In some embodiments, the vehicle trajectory control device 110 further includes a processing module, which is configured to obtain the trajectory control model using the following method:

[0136] Acquire a database, the database comprising at least trajectory data of a plurality of vehicles in a time series;

[0137] Classifying the trajectory data in the database into risk levels to obtain different risk level groups;

[0138] The vehicle following data included in the different risk level groups are calculated based on a machine learning method to obtain trajectory control models corresponding to different risk levels.

[0139] In some embodiments, the processing module is further configured to:

[0140] Building a vehicle following scenario library comprising a plurality of sub-scenarios based on the database, each of the sub-scenario comprising at least vehicle following data of a vehicle following pair within a preset time;

[0141] Determine the risk level of each of the sub-scenarios in the vehicle following scenario library, and divide each of the sub-scenarios into different risk level groups based on the risk level.

[0142] In some embodiments, the determination module 102 is further configured to calculate the target risk level of the target vehicle using a collision conflict safety index through a machine learning method; wherein the collision conflict safety index is used to assess the collision risk during vehicle driving, and the collision conflict safety index is configured to be calculated using the following formula:

[0143]

[0144] in, is the collision safety index; x r is the relative position of the front and rear vehicles in the following direction; d p is the braking distance of the following vehicle to center the vehicle in front; d f is the braking distance of the following vehicle to the center of the following vehicle; v p (t) is the speed of the vehicle in front of the following vehicle at time t; v f (t) is the speed of the following vehicle at time t; t reaction is the reaction time of the driver of the following car; p is the acceleration of the following car centering the front car; a fIt is the acceleration of the following car relative to the car behind.

[0145] In some embodiments, the determination module 102 is further configured to calculate the target risk level of the target vehicle using the following formula via a machine learning method:

[0146]

[0147] Among them, w IN and w M There are two types of parameters related to the risk of the vehicle, w IN +w M =1; It indicates the value of the standard safety index when the front vehicle decelerates at normal braking acceleration and the rear vehicle decelerates at maximum braking acceleration; It is the value of the standard safety indicator when both the front and rear vehicles in a following vehicle are decelerating at the maximum braking deceleration.

[0148] In some embodiments, the processing module is further configured to:

[0149] The reward function in the machine learning method is calculated using a first boundary speed and a second boundary speed; wherein the first boundary speed is calculated using the following formula:

[0150]

[0151] The second boundary velocity is calculated using the following formula:

[0152]

[0153] Among them, x vr =x r -1.5m, m is meter; a partialbraking represents the vehicle acceleration under partial braking conditions; a fullbraking Indicates vehicle acceleration under full braking, with partial braking providing a comfortable braking force and full braking providing maximum braking force.

[0154] In some embodiments, the processing module is further configured to calculate the safety reward function included in the reward function based on the first boundary speed and the second boundary speed using the following formula:

[0155]

[0156] An embodiment of the present application further provides a storage medium storing a computer program, which implements the vehicle trajectory control method when executed by a processor.

[0157] Note that the various units in the various embodiments of the present application can be implemented as computer-executable instructions stored on a memory, which can implement the corresponding steps when executed by a processor; they can also be implemented as hardware with corresponding logical computing capabilities; or they can be implemented as a combination of software and hardware (firmware). In some embodiments, the processor can be implemented as any one of an FPGA, an ASIC, a DSP chip, a SOC (system on a chip), an MPU (for example, but not limited to Cortex), etc. The processor can be communicatively coupled to the memory and configured to execute the computer-executable instructions stored therein. The memory can include read-only memory (ROM), flash memory, random access memory (RAM), dynamic random access memory (DRAM) such as synchronous DRAM (SDRAM) or Rambus DRAM, static memory (for example, flash memory, static random access memory), etc., on which computer-executable instructions are stored in any format. The computer-executable instructions can be accessed by the processor, read from the ROM or any other suitable storage location, and loaded into the RAM for execution by the processor to implement the wireless communication method according to the various embodiments of the present application.

[0158] It should be noted that, among the various components of the system of the present application, the components are logically divided according to the functions to be implemented, but the present application is not limited to this, and the various components can be re-divided or combined as needed. For example, some components can be combined into a single component, or some components can be further decomposed into more sub-components.

[0159] The various component embodiments of the present application can be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that in practice, a microprocessor or digital signal processor (DSP) can be used to implement some or all of the functions of some or all of the components in the system according to the embodiments of the present application. The present application can also be implemented as an apparatus or device program (e.g., a computer program and a computer program product) for performing part or all of the methods described herein. Such a program implementing the present application can be stored on a computer-readable medium or can be in the form of one or more signals. Such a signal can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form. In addition, the present application can be implemented with the aid of hardware comprising several different elements and with the aid of a suitably programmed computer. In a unit claim that lists several means, several of these means can be embodied by the same hardware item. The use of the words first, second, and third, etc. does not indicate any order. These words can be interpreted as names.

[0160] Furthermore, although exemplary embodiments have been described herein, the scope includes any and all embodiments based on the present application with equivalent elements, modifications, omissions, combinations (e.g., solutions that intersect various embodiments), adaptations, or changes. The elements in the claims are to be interpreted broadly based on the language employed in the claims and are not limited to the examples described in this specification or during the prosecution of the application, which examples are to be interpreted as non-exclusive.

[0161] The above description is intended to be illustrative rather than restrictive. For example, the above examples (or one or more of their solutions) can be used in combination with each other. For example, a person of ordinary skill in the art may use other embodiments when reading the above description. In addition, in the above-mentioned specific embodiments, various features can be grouped together to simplify the application. This should not be interpreted as an intention that a disclosed feature that is not required to be protected is necessary for any claim. On the contrary, the subject matter of the present application may be less than all the features of a specific disclosed embodiment. Thus, the following claims are incorporated into the specific embodiments as examples or embodiments, wherein each claim is independently a separate embodiment, and it is considered that these embodiments can be combined with each other in various combinations or arrangements. The scope of this application should be determined with reference to the appended claims and the full scope of equivalents to which these claims are entitled.

[0162] The above embodiments are merely exemplary embodiments of the present application and are not intended to limit the scope of the present application. The scope of protection of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and scope of protection of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the scope of protection of the present application.

Claims

1. A vehicle trajectory control method, characterized in that: include: Obtain the real-time driving status of the target vehicle; Determining a target risk level of the target vehicle based on the real-time driving status; Determining a trajectory control model corresponding to the target risk level according to the target risk level, wherein there are multiple trajectory control models, and different trajectory control models correspond to different risk levels; Determine the driving parameters output by the trajectory control model based on the real-time driving state, so as to control the driving state of the target vehicle according to the driving parameters; wherein, Determining the target risk level of the target vehicle based on the real-time driving status specifically includes: The target risk level of the target vehicle is calculated using a collision conflict safety index through a machine learning method; wherein the collision conflict safety index is used to assess the collision risk during vehicle driving, and the collision conflict safety index is configured to be calculated using the following formula: in, is the collision safety index; x r is the relative position of the front and rear vehicles in the following direction; d p is the braking distance of the following vehicle to center the vehicle in front; d f is the braking distance of the following vehicle to the center of the following vehicle; v p (t) is the speed of the vehicle in front of the following vehicle at time t; v f (t) is the speed of the following vehicle at time t; t reaction is the reaction time of the driver of the following car; p is the acceleration of the following car centering the front car; a f is the acceleration of the following vehicle relative to the vehicle behind; The method of calculating the target risk level of the target vehicle using the collision conflict safety index through a machine learning method specifically includes: The target risk level of the target vehicle is calculated using the following formula via machine learning: Among them, w IN and w M There are two types of parameters related to the risk of the vehicle, w IN +w M =1; It indicates the value of the standard safety index when the front vehicle decelerates at normal braking acceleration and the rear vehicle decelerates at maximum braking acceleration; It is the standard safety index value when both the leading and following vehicles in a following vehicle alignment are decelerating at their maximum braking deceleration. The vehicle trajectory control method further includes: The reward function in the machine learning method is calculated using a first boundary speed and a second boundary speed; wherein the first boundary speed is calculated using the following formula: The second boundary velocity is calculated using the following formula: Among them, x vr =x r -1.5m, m is meter; a partialbraking represents the vehicle acceleration under partial braking conditions; a fullbraking Indicates vehicle acceleration under full braking, with partial braking providing a comfortable braking force and full braking providing maximum braking force.

2. The vehicle trajectory control method according to claim 1, characterized in that: Determining the driving parameters output by the trajectory control model based on the real-time driving state specifically includes: determining, via the trajectory control model, a risk level of an adjacent lane adjacent to the target vehicle; Comparing the risk level of the adjacent lane with the target risk level to obtain a comparison result; The driving parameters of the target vehicle, including a lateral driving parameter, are determined based on the comparison result.

3. The vehicle trajectory control method according to claim 1, characterized in that: The trajectory control model is obtained by the following method: Acquire a database, the database comprising at least trajectory data of a plurality of vehicles in a time series; Classifying the trajectory data in the database into risk levels to obtain different risk level groups; The vehicle following data included in the different risk level groups are calculated based on a machine learning method to obtain trajectory control models corresponding to different risk levels.

4. The vehicle trajectory control method according to claim 3, characterized in that: The risk classification of the trajectory data in the database to obtain different risk level groups specifically includes: Building a vehicle following scenario library comprising a plurality of sub-scenarios based on the database, each of the sub-scenario comprising at least vehicle following data of a vehicle following pair within a preset time; Determine the risk level of each of the sub-scenarios in the vehicle following scenario library, and divide each of the sub-scenarios into different risk level groups based on the risk level.

5. The vehicle trajectory control method according to claim 1, characterized in that: The method of calculating the reward function in the machine learning method using the first boundary speed and the second boundary speed specifically includes: The safety reward function included in the reward function is calculated based on the first boundary speed and the second boundary speed using the following formula:

6. A vehicle trajectory control device, characterized in that: include: an acquisition module configured to acquire the real-time driving status of a target vehicle; a determination module configured to determine a target risk level of the target vehicle based on the real-time driving status; as well as, Determining a trajectory control model corresponding to the target risk level according to the target risk level, wherein there are multiple trajectory control models, and different trajectory control models correspond to different risk levels; Determine the driving parameters output by the trajectory control model based on the real-time driving state, so as to control the driving state of the target vehicle according to the driving parameters; wherein, The determining module is further configured to: The target risk level of the target vehicle is calculated using a collision conflict safety index through a machine learning method; wherein the collision conflict safety index is used to assess the collision risk during vehicle driving, and the collision conflict safety index is configured to be calculated using the following formula: in, is the collision safety index; x r is the relative position of the front and rear vehicles in the following direction; d p is the braking distance of the following vehicle to center the vehicle in front; d f is the braking distance of the following vehicle to the center of the following vehicle; v p (t) is the speed of the vehicle in front of the following vehicle at time t; v f (t) is the speed of the following vehicle at time t; t reaction is the reaction time of the driver of the following car; P is the acceleration of the following car centering the front car; a f is the acceleration of the following vehicle relative to the vehicle behind; The determining module is further configured to: The target risk level of the target vehicle is calculated using the following formula via machine learning: Among them, w IN and w M There are two types of parameters related to the risk of the vehicle, w IN +w M =1; It indicates the value of the standard safety index when the front vehicle decelerates at normal braking acceleration and the rear vehicle decelerates at maximum braking acceleration; It is the standard safety index value when both the leading and following vehicles in a following vehicle alignment are decelerating at their maximum braking deceleration. The vehicle trajectory control device further includes a processing module, which is further configured to: The reward function in the machine learning method is calculated using a first boundary speed and a second boundary speed; wherein the first boundary speed is calculated using the following formula: The second boundary velocity is calculated using the following formula: Among them, x vr =x r -1.5m, m is meter; a partialbraking represents the vehicle acceleration under partial braking conditions; a fullbraking Indicates vehicle acceleration under full braking, with partial braking providing a comfortable braking force and full braking providing maximum braking force.

7. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the vehicle trajectory control method according to any one of claims 1 to 5 is implemented.

Images