Vehicle trajectory prediction method, trajectory prediction model training method and device

Abstract

The embodiment of the application provides a vehicle trajectory prediction method, a trajectory prediction model training method and device, relates to the technical field of intelligent driving, and comprises the following steps: obtaining a first trajectory representation sequence of a first target vehicle, inputting the first trajectory representation sequence into a vehicle trajectory prediction model, obtaining a second trajectory representation sequence output by the vehicle trajectory prediction model, optimizing each trajectory point data in the second trajectory representation sequence according to trajectory point data of the first target vehicle at t moment, obtaining a target trajectory of the first target vehicle at t+1 moment to t+n moment which is predicted, and performing intelligent driving control on a second target vehicle according to the target trajectory. The trajectory of the vehicle is predicted by using the trained vehicle trajectory prediction model, and the predicted trajectory is optimized. The accuracy of the trajectory prediction of the vehicle can be effectively improved, and the driving safety is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent driving, and in particular to a vehicle trajectory prediction method, a trajectory prediction model training method and device. BACKGROUND

[0002] Trajectory prediction refers to establishing a trajectory prediction model by using historical trajectory data of an object, taking a known part of a trajectory as input of the prediction model, and deriving a road where the trajectory is located at a next time by model inference, which belongs to a kind of position prediction.

[0003] Currently, trajectory prediction of a vehicle is usually performed by using a dynamics model, a Kalman filter algorithm model, etc. However, when a vehicle is driving on a road, the driving state is constrained by vehicle state, road traffic rules, etc. The establishment process of the dynamics model is relatively complex and difficult. The Kalman filter algorithm model based on kinematics has certain advantages in a short time and is relatively high in calculation efficiency, but the prediction accuracy is obviously reduced when prediction is performed in a long time domain. SUMMARY

[0004] The present application provides a vehicle trajectory prediction method, a trajectory prediction model training method and device, which can improve the accuracy of vehicle trajectory prediction.

[0005] In a first aspect, the present application provides a vehicle trajectory prediction method, comprising:

[0006] obtaining a first trajectory representation sequence of a first target vehicle, the first trajectory representation sequence comprising trajectory point data of the first target vehicle at t-m time to t time, the trajectory point data comprising position and speed of a trajectory point;

[0007] inputting the first trajectory representation sequence into a vehicle trajectory prediction model to obtain a second trajectory representation sequence output by the vehicle trajectory prediction model, the second trajectory representation sequence comprising predicted trajectory point data of the first target vehicle at t+1 time to t+n time; m and n are both integers greater than 1;

[0008] optimizing each trajectory point data in the second trajectory representation sequence according to the trajectory point data of the first target vehicle at t time to obtain a target trajectory of the first target vehicle at t+1 time to t+n time;

[0009] controlling intelligent driving of a second target vehicle according to the target trajectory.

[0010] Optionally, the optimization of each trajectory point data in the second trajectory characterization sequence according to the trajectory point data of the first target vehicle at the t moment, to obtain the predicted target trajectory of the first target vehicle at the t+1 moment to the t+n moment, comprises:

[0011] taking each trajectory point data in the second trajectory characterization sequence as observation data, and acquiring estimated trajectory point data of each trajectory point data in multiple motion modes according to the observation data and the trajectory point data of the first target vehicle at the t moment;

[0012] acquiring each trajectory point data after optimization according to the estimated trajectory point data of each trajectory point data in multiple motion modes;

[0013] obtaining the predicted target trajectory of the first target vehicle at the t+1 moment to the t+n moment according to each trajectory point data after optimization.

[0014] Optionally, the acquisition of the estimated trajectory point data of each trajectory point data in multiple motion modes according to the observation data and the trajectory point data of the first target vehicle at the t moment, comprises:

[0015] acquiring initial estimated trajectory point data of each trajectory point data in multiple motion modes according to a state equation corresponding to each motion mode and the trajectory point data of the first target vehicle at the t moment;

[0016] respectively acquiring a gain matrix of each initial estimated trajectory point data in each motion mode;

[0017] acquiring the estimated trajectory point data of each trajectory point data in multiple motion modes according to the observation data, the gain matrix of each initial trajectory point data in each motion mode, and the initial estimated trajectory point data of each trajectory point data in multiple motion modes.

[0018] Optionally, the acquisition of each trajectory point data after optimization according to the estimated trajectory point data of each trajectory point data in multiple motion modes, comprises:

[0019] weighting the estimated trajectory point data of each trajectory point data in multiple motion modes according to a weight corresponding to each motion mode, to obtain each trajectory point data after optimization.

[0020] In a second aspect, an embodiment of the present application provides a vehicle trajectory prediction model training method, comprising:

[0021] acquiring historical trajectory data of M sample vehicles;

[0022] generate a sample data set according to historical trajectory point data of M sample vehicles at t-m time to t+n time, the sample data set comprising: a first sample trajectory representation sequence generated by using historical trajectory data of each of the sample vehicles, and a label trajectory representation sequence, the first sample trajectory representation sequence comprising historical trajectory point data of the sample vehicles at t-m time to t time, and the label trajectory representation sequence comprising historical trajectory point data of the sample vehicles at t+1 time to t+n time; the historical trajectory point data comprising: position and speed of a trajectory point; the M, the m and the n are all integers greater than 1;

[0023] train a vehicle trajectory prediction model by using the sample data set to obtain a trained vehicle trajectory prediction model.

[0024] Optionally, the training of the vehicle trajectory prediction model by using the sample data set to obtain a trained vehicle trajectory prediction model comprises:

[0025] divide the sample data set into a training set and a test set, the training set comprising: first sample trajectory representation sequences of M1 sample vehicles, and a label trajectory representation sequence, and the test set comprising: first sample trajectory representation sequences of M2 sample vehicles, and a label trajectory representation sequence, M1+M2≤M;

[0026] in the process of training the vehicle trajectory prediction model:

[0027] if a first loss function value of the vehicle trajectory prediction model based on the training set is greater than a preset value, update parameters of the vehicle trajectory prediction model;

[0028] if the first loss function value is less than or equal to the preset value, and a second loss function value of the vehicle trajectory prediction model based on the test set is greater than a preset value, add a feature extraction layer in the vehicle trajectory prediction model;

[0029] if the first loss function value is less than or equal to the preset value, and the second loss function value is greater than the preset value or equal to, determine that the training of the vehicle trajectory prediction model is completed.

[0030] Optionally, the generation of the sample data set according to historical trajectory point data of M sample vehicles at t-m time to t+n time comprises:

[0031] preprocess historical trajectory point data of M sample vehicles at t-m time to t+n time;

[0032] generate the sample data set according to the historical trajectory point data of the sample vehicles at t-m time to t+n time after preprocessing.

[0033] In a third aspect, an embodiment of the present application provides a vehicle trajectory prediction device, comprising:

[0034] an acquisition module configured to acquire a first trajectory representation sequence of a first target vehicle, the first trajectory representation sequence comprising trajectory point data of the first target vehicle at a time t-m to a time t, the trajectory point data comprising a position and a speed of a trajectory point;

[0035] a processing module configured to input the first trajectory representation sequence into a vehicle trajectory prediction model to obtain a second trajectory representation sequence output by the vehicle trajectory prediction model, the second trajectory representation sequence comprising predicted trajectory point data of the first target vehicle at a time t+1 to a time t+n; the m and the n are both integers greater than 1;

[0036] an optimization module configured to optimize each trajectory point data in the second trajectory representation sequence according to the trajectory point data of the first target vehicle at the time t to obtain a target trajectory of the first target vehicle at the time t+1 to the time t+n;

[0037] a control module configured to control intelligent driving of a second target vehicle according to the target trajectory.

[0038] Optionally, the vehicle trajectory prediction device can execute the vehicle trajectory prediction method according to any one of the first aspect.

[0039] In a fourth aspect, an embodiment of the present application provides a vehicle trajectory prediction model training device, comprising:

[0040] an acquisition module configured to acquire historical trajectory data of M sample vehicles;

[0041] a processing module configured to generate a sample data set according to historical trajectory point data of the M sample vehicles at a time t-m to a time t+n, the sample data set comprising a first sample trajectory representation sequence generated by using the historical trajectory data of each of the sample vehicles and a label trajectory representation sequence, the first sample trajectory representation sequence comprising historical trajectory point data of the sample vehicle at the time t-m to the time t, and the label trajectory representation sequence comprising historical trajectory point data of the sample vehicle at the time t+1 to the time t+n; the historical trajectory point data comprising a position and a speed of a trajectory point; the M, the m and the n are all integers greater than 1;

[0042] a training module configured to train a vehicle trajectory prediction model by using the sample data set to obtain a trained vehicle trajectory prediction model.

[0043] Optionally, the vehicle trajectory prediction model training device can execute the vehicle trajectory prediction model training method described in any of the second aspects.

[0044] Fifthly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0045] The memory is used to store computer instructions; the processor is used to execute the computer instructions stored in the memory to implement the method of either the first aspect or the second aspect.

[0046] In a sixth aspect, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the method of any one of the first and / or second aspects.

[0047] In a seventh aspect, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the method of either the first aspect or the second aspect.

[0048] The vehicle trajectory prediction method, trajectory prediction model training method, and apparatus provided in this application obtain a first trajectory representation sequence of a first target vehicle, input the first trajectory representation sequence into a vehicle trajectory prediction model, and obtain a second trajectory representation sequence output by the vehicle trajectory prediction model. Based on the trajectory point data of the first target vehicle at time t, each trajectory point data in the second trajectory representation sequence is optimized to obtain the predicted target trajectory of the first target vehicle from time t+1 to time t+n. Based on the target trajectory, intelligent driving control is performed on the second target vehicle. The trained vehicle trajectory prediction model is used to predict the vehicle's trajectory, and the predicted trajectory is optimized. This can effectively improve the accuracy of vehicle trajectory prediction, thereby improving driving safety. Attached Figure Description

[0049] Figure 1 This application provides a scenario illustration for its embodiments;

[0050] Figure 2 A flowchart illustrating the vehicle trajectory prediction method provided in this application embodiment. Figure 1 ;

[0051] Figure 3 A schematic diagram of the vehicle trajectory prediction model provided in the embodiments of this application;

[0052] Figure 4 A flowchart illustrating the vehicle trajectory prediction method provided in this application embodiment. Figure 2 ;

[0053] Figure 5 This is a schematic diagram illustrating the optimization of prediction results provided in an embodiment of this application;

[0054] Figure 6 A flowchart of a vehicle trajectory prediction model training method provided for an embodiment of the present application Figure 1 ;

[0055] Figure 7 A flowchart of a vehicle trajectory prediction model training method provided for an embodiment of the present application Figure 2 ;

[0056] Figure 8 A vehicle trajectory prediction schematic diagram provided for an embodiment of the present application

[0057] Figure 9 A structural schematic diagram of a vehicle trajectory prediction device provided for an embodiment of the present application

[0058] Figure 10 A structural schematic diagram of a vehicle trajectory prediction model training device provided for an embodiment of the present application

[0059] Figure 11 A structural schematic diagram of an electronic device provided for an embodiment of the present application DETAILED DESCRIPTION

[0060] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0061] In the embodiments of the present application, the same items or similar items with basically the same functions and effects are distinguished by using “first”, “second” and the like, and the sequence is not limited. Those skilled in the art can understand that “first”, “second” and the like do not limit the quantity and execution sequence, and “first”, “second” and the like do not necessarily mean different.

[0062] It should be noted that in the embodiments of the present application, “exemplary” or “for example” and the like are used to represent examples, illustrations or descriptions. Any embodiment or design scheme described as “exemplary” or “for example” in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, “exemplary” or “for example” and the like are used to present the relevant concept in a specific manner.

[0063] The vehicle equipped with the advanced auxiliary driving system can use the sensors on the vehicle to sense the surrounding environment during the driving process of the vehicle, identify static and dynamic objects in the surrounding environment, and feed back potential dangers in the environment to the driver, thereby effectively increasing the comfort and safety during the driving process of the vehicle. Predicting the obstacles in the driving environment, obtaining accurate obstacle vehicle information in real time, and predicting the driving dynamics are important for the advanced auxiliary driving system and are necessary operations for realizing L3 and above levels of automatic driving.

[0064] Traditional trajectory prediction includes a dynamics model, a Kalman filter algorithm model, and a machine learning model. It is difficult to accurately establish a model for trajectory prediction of surrounding vehicles of the ego vehicle, and the prediction accuracy is generally low.

[0065] When the vehicle drives on the road, the driving state is constrained by the vehicle state, road traffic rules, etc. The Kalman filter equation is established based on the kinematic state equation to predict the trajectory, and the prediction accuracy has certain advantages in a short time and the calculation efficiency is high, but the accuracy is obviously reduced when predicting in a long time domain.

[0066] Therefore, the present application proposes a trajectory prediction method and a trajectory prediction model training method and device for a vehicle. The trajectory of the vehicle is predicted by using a deep learning trajectory prediction model, and the predicted trajectory is optimized by combining the kinematic state equation. The advantages of deep learning in a long time domain can effectively improve the accuracy of vehicle trajectory prediction.

[0067] The technical solutions of the present application and how the technical solutions of the present application solve the above technical problems will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present application will be described below with reference to the accompanying drawings.

[0068] Figure 1 The scene schematic diagram provided for the embodiments of the present application is shown in FIG. 1, which includes a vehicle 101, a vehicle 102, and a sensor 103 carried in the vehicle 101. Figure 1

[0069] The sensor 103 in the vehicle 101 can collect the historical trajectory data of the vehicle 102, for example, the position and speed of the vehicle 102 in the past 4 seconds. The sensor 103 can input the collected historical trajectory data of the vehicle 102 into the processing unit of the vehicle 101, and the processing unit can obtain the predicted trajectory data of the vehicle 102, for example, the position and speed in the future 5 seconds, based on the received data and the pre-trained vehicle trajectory prediction model.

[0070] ​The vehicle 101 can perform intelligent driving control on the vehicle 101 according to the predicted trajectory data obtained by the processing unit. For example, if the vehicle 101 can determine that the vehicle 102 will turn in 3 seconds according to the predicted trajectory data, the vehicle 101 can take braking measures to increase the vehicle distance between the vehicle 101 and the vehicle 102.

[0071] Optionally, in the embodiment of the application, the vehicle 102 can be any vehicle around the vehicle 101, and the types of the vehicle 101 and the vehicle 102 can be vehicles or other types of vehicles. The type of the vehicle is not limited in the embodiment of the application.

[0072] Optionally, in the embodiment of the application, the vehicle trajectory prediction model can be pre-stored in the storage unit of the vehicle 101.

[0073] Optionally, the sensor 103 can include an industrial camera, a camera support of an illumination device, and the like. The illumination device includes an illumination sensor and an auxiliary light source. The type of the sensor 103 is not limited in the embodiment of the application.

[0074] Figure 1 The application scenario of the embodiment of the application is simply described, and the vehicle 101 in the application scenario will be taken as an example to describe in detail the vehicle trajectory prediction method provided by the embodiment of the application. Figure 1

[0075] Figure 2 The flowchart of the vehicle trajectory prediction method provided by the embodiment of the application is shown in FIG. 2, which includes the following steps. Figure 2

[0076] S201, obtaining a first trajectory representation sequence of a first target vehicle, the first trajectory representation sequence including trajectory point data of the first target vehicle at t-m time to t time.

[0077] In the embodiment of the application, as shown in FIG. 1, the first target vehicle can be any vehicle on the road except the vehicle 101. The first trajectory representation sequence can be trajectory point data representing m time of the first target vehicle, and m is an integer greater than 1. The trajectory point can be the trajectory of the first target vehicle at any time, and the trajectory point data includes the position and speed of the trajectory point. Figure 1 The position of the trajectory point includes the position of the trajectory point in the x direction and the y direction, and the speed of the trajectory point includes the speed of the trajectory point in the x direction and the y direction. For example, for any time t, the trajectory point data at t time can be represented as C t ={x t ,y t ,v xt ,v yt}.

[0078] ​​​

[0079] In this embodiment of the application, the first trajectory representation sequence of the first target vehicle can be obtained through sensors mounted on the vehicle.

[0080] S202. Input the first trajectory representation sequence into the vehicle trajectory prediction model to obtain the second trajectory representation sequence output by the vehicle trajectory prediction model. The second trajectory representation sequence includes the predicted trajectory point data of the first target vehicle from time t+1 to time t+n.

[0081] In this embodiment of the application, the vehicle trajectory prediction model can be a deep learning-based network model.

[0082] For example, such as Figure 3 As shown, the vehicle trajectory prediction model in this embodiment consists of two main components: a Gate Recurrent Unit (GRU) and an Attention Mechanism (AM). The model comprises an input layer, a GRU layer, an AM layer, and an output layer. The GRU layer includes at least one GRU, and the AM layer includes at least one attention unit h. i The GRU layer, also known as the feature extraction layer, is used to extract features and transform data from the input layer. The AM layer can further analyze the data input from the GRU layer.

[0083] In this embodiment of the application, the vehicle trajectory prediction model can be a trained model that meets the usage requirements.

[0084] In this embodiment, the first trajectory representation sequence can be input into the input layer of the vehicle trajectory prediction model. After passing through the input layer, the first trajectory representation sequence is input into the GRU layer. The GRU layer performs preliminary prediction on the first trajectory representation sequence to obtain a preliminary prediction result. Then, the preliminary prediction result is passed into the AM layer for further prediction to obtain a more accurate prediction result. Finally, the result is output through the output layer to obtain the second trajectory representation sequence output by the vehicle trajectory prediction model.

[0085] Optionally, to make the real-world significance of the prediction results output by the vehicle trajectory prediction model more intuitive and readable, the prediction results of the vehicle trajectory prediction model can also be denormalized, that is, the second trajectory representation sequence representation data can be denormalized.

[0086] S203. Based on the trajectory point data of the first target vehicle at time t, optimize the trajectory point data of each trajectory point in the second trajectory representation sequence to obtain the predicted target trajectory of the first target vehicle from time t+1 to time t+n.

[0087] In the embodiments of the present application, since the vehicle trajectory prediction model is trained according to the actually sampled data of the vehicle, it is difficult to avoid the influence of noise in the data sampling process. Therefore, the second trajectory representation sequence output by the vehicle trajectory prediction model may have errors with the actual trajectory points of the first target vehicle, and the second trajectory representation sequence output by the vehicle trajectory prediction model needs to be further optimized to obtain a more accurate second trajectory representation sequence.

[0088] In the embodiments of the present application, the second trajectory representation sequence output by the vehicle trajectory prediction model can be optimized based on the kinematic state equation.

[0089] For example, the Kalman filtering equation can be established based on the kinematic state equation to optimize the second trajectory representation sequence output by the vehicle trajectory prediction model, and obtain the target trajectory of the first target vehicle at the t+1 time to the t+n time.

[0090] For example, the estimated value of the first target vehicle at the t+1 time to the t+n time can be obtained according to the trajectory point data of the first target vehicle at the t time in combination with the Kalman filtering equation, the trajectory point data of each time corresponding to the second trajectory representation sequence is taken as the observation, the optimized trajectory point data of each time is obtained, and then the predicted target trajectory of the first target vehicle at the t+1 time to the t+n time is obtained.

[0091] S204, according to the target trajectory, intelligently control the second target vehicle.

[0092] In the embodiments of the present application, when the target trajectory of the first target vehicle is obtained, the driving strategy of the second target vehicle can be adjusted according to the target trajectory.

[0093] For example, please continue to refer to Figure 1 If the vehicle 101 obtains the target trajectory of the vehicle 102, it can be confirmed that the vehicle 102 will turn after 3 seconds according to the target trajectory, and the vehicle 101 can take braking or turning to avoid the influence of the turning of the vehicle 102. Optionally, the second target vehicle can be a heavy vehicle.

[0094] The vehicle trajectory prediction method provided in this application involves acquiring a first trajectory representation sequence of a first target vehicle, inputting this sequence into a vehicle trajectory prediction model to obtain a second trajectory representation sequence output by the model. Based on the trajectory point data of the first target vehicle at time t, each trajectory point in the second trajectory representation sequence is optimized to obtain the predicted target trajectory of the first target vehicle from time t+1 to time t+n. Intelligent driving control is then performed on the second target vehicle based on this target trajectory. By using a trained vehicle trajectory prediction model to predict the vehicle's trajectory and optimizing the predicted trajectory, the accuracy of vehicle trajectory prediction can be effectively improved, thereby enhancing driving safety.

[0095] Figure 4 A flowchart illustrating the vehicle trajectory prediction method provided in this application embodiment. Figure 2 ,exist Figure 4 Based on the illustrated embodiment, a method for optimizing the data of each trajectory point in the second trajectory representation sequence will be further explained, such as... Figure 5 As shown, it includes the following steps:

[0096] S401. Take each trajectory point data in the second trajectory representation sequence as observation data, and based on the observation data and the trajectory point data of the first target vehicle at time t, obtain the estimated trajectory point data in multiple motion modes corresponding to each trajectory point data.

[0097] In this embodiment, the second trajectory representation sequence output by the vehicle trajectory prediction model can be optimized according to the Kalman estimation idea to obtain estimated trajectory point data under multiple motion modes corresponding to each trajectory point data. The specific method is as follows:

[0098] a1: Based on the state equations corresponding to each motion mode and the trajectory point data of the first target vehicle at time t, obtain the initial estimated trajectory point data for each trajectory point data in multiple motion modes.

[0099] In this embodiment of the application, the Kalman filter algorithm often assumes a certain motion state for calculation. If the assumed state is incorrect, it will bring new errors. Therefore, when optimizing the second trajectory representation sequence, this embodiment of the application divides the motion state of the first target vehicle into a stationary mode, a uniform speed mode, and a uniform acceleration mode.

[0100] Based on the state equation in the Kalman filter algorithm and the trajectory point data of the vehicle at time t, the initial estimated trajectory point data of different motion modes at time t+1 can be obtained.

[0101] For example, the state equation is as follows:

[0102]

[0103] wherein, is the initial estimated trajectory point data at t+1 time, F t+1|t is the state transition matrix from t time to t+1 time, X t is the trajectory point data at t time.

[0104] The state transition matrix of different motion modes is brought into the state equation, and the initial estimated trajectory point data of different motion modes at t+1 time can be obtained.

[0105] The state transition matrix of different motion modes is as follows:

[0106] Stationary mode:

[0107]

[0108] Uniform speed mode:

[0109]

[0110] Uniform acceleration mode:

[0111]

[0112] wherein, Δt is the difference between t+1 time and t time.

[0113] In the embodiment of the application, the value of Δt is changed, and the initial estimated trajectory point data of different motion modes at t+2 to t+n time can be obtained according to the state equation, that is, the initial estimated trajectory point data of each trajectory point data in multiple motion modes.

[0114] a2: respectively obtain the gain matrix of each initial estimated trajectory point data in each motion mode.

[0115] In the embodiment of the application, the gain matrix is the Kalman gain matrix, which can be determined by the following method:

[0116] S1: respectively obtain the variance matrix of the initial estimated trajectory point data at t+1 time in different motion modes.

[0117] The variance matrix of the initial estimated trajectory point data at t+1 time is determined according to the following formula:

[0118]

[0119] wherein, P t+1|t is the variance matrix from t time to t+1 time, is the transpose matrix of the state transition matrix, P t is the variance matrix at t time, which can be a unit matrix, Q tis a noise matrix at time t.

[0120] S2: Obtain a gain matrix of the initial estimated trajectory point data at time t+1 according to a variance matrix of the initial estimated trajectory point data at time t+1 under different motion modes.

[0121] The gain matrix of the initial estimated trajectory point data at time t+1 is determined according to the following formula:

[0122]

[0123] wherein H t+1 is a conversion matrix at time t+1, is a transpose matrix of the conversion matrix at time t+1, R t+1 is a noise matrix at time t+1.

[0124] S3: Obtain the gain matrix of each initial estimated trajectory point data.

[0125] The variance matrix of the initial estimated trajectory point data at time t+1 is updated according to the following formula.

[0126] P t+1 = P t+1|t (I-K t+1 H t+1 )

[0127] The process of S1 to S3 is repeated to obtain the gain matrix of the initial estimated trajectory point data corresponding to each time, that is, the gain matrix of each initial estimated trajectory point data under each motion mode can be obtained.

[0128] a3: Obtain the estimated trajectory point data corresponding to each motion mode of each trajectory point data according to the observation data, the gain matrix of each initial estimated trajectory point data under each motion mode, and the initial estimated trajectory point data of each trajectory point data corresponding to multiple motion modes.

[0129] In the embodiments of the present application, the estimated trajectory point data corresponding to multiple motion modes of each trajectory point data can be obtained according to each trajectory point data in the observation data.

[0130] For example, the estimated trajectory point data corresponding to multiple motion modes at time t+1 can be determined according to the following formula.

[0131]

[0132] wherein Z t+1 is the trajectory point data at time t+1 in the observation data.

[0133] The time in the above formula can correspond to the estimated trajectory point data in multiple motion modes at each time, that is, the estimated trajectory point data in multiple motion modes corresponding to each trajectory point data can be obtained.

[0134] S402, according to the estimated trajectory point data in multiple motion modes corresponding to each trajectory point data, obtaining the optimized each trajectory point data.

[0135] In the embodiment of the application, the estimated trajectory point data in multiple motion modes can be comprehensively processed to obtain the optimized each trajectory point data.

[0136] Specifically, according to the weight corresponding to each motion mode, the estimated trajectory point data in multiple motion modes corresponding to each trajectory point data is weighted and processed to obtain the optimized each trajectory point data.

[0137] For example, each trajectory point data can be determined by the following formula:

[0138]

[0139] wherein, The i-th trajectory point data after optimization, The i-th estimated trajectory point data, α i The weight corresponding to the i-th motion mode.

[0140] S403, according to the optimized each trajectory point data, obtaining the predicted target trajectory of the first target vehicle at t+1 time to t+n time.

[0141] In the embodiment of the application, the optimized each trajectory point data is obtained, and the optimized each trajectory point data can be connected in time sequence to obtain the predicted target trajectory of the first target vehicle at t+1 time to t+n time.

[0142] In summary, please refer to Figure 6 The embodiment of the application optimizes each trajectory point data output by the vehicle trajectory prediction model using Kalman filtering method, obtains the optimized data corresponding to each trajectory point data in three motion modes, and performs weighted calculation on the optimized data corresponding to each trajectory point data in three motion modes according to the weight corresponding to three motion modes, and finally obtains the predicted target trajectory of the first target vehicle at t+1 time to t+n time.

[0143] The embodiment of the application can reduce the error of the model output result caused by the noise of the sample data in the model training process by optimizing each trajectory point data output by the vehicle trajectory prediction model, improve the accuracy of the vehicle trajectory prediction, and further improve the safety of the vehicle driving.

[0144] The method for predicting the trajectory of a vehicle using a vehicle trajectory prediction model provided in the embodiments of the present application is described above, and the training method of the vehicle trajectory prediction model will be described below.

[0145] Figure 6 The flowchart of the training method of the vehicle trajectory prediction model provided in the embodiments of the present application, the execution subject of the embodiments of the present application can be an electronic device, or a model training platform, etc. The platform can be a platform located in the cloud, or a distributedly deployed platform (for example, part of which is deployed in a cloud environment, and part of which is deployed in an edge environment, etc.). As shown in the figure, the following steps are included: Figure 7

[0146] S601, obtain historical trajectory data of M sample vehicles.

[0147] In the embodiments of the present application, the historical trajectory data refers to data obtained according to the historical driving trajectory of the sample vehicle, including the position coordinates (x, y) and the corresponding speed (v x , v y ) of the sample vehicle at any time.

[0148] In the embodiments of the present application, the historical trajectory data of N sample vehicles can be obtained according to the driving trajectories of multiple roads and multiple vehicles.

[0149] In the embodiments of the present application, there can be trajectories with abnormal data in the obtained historical trajectory data of N sample vehicles. For example, a certain sample vehicle trajectory data point is particularly small; there is trajectory drift, that is, the moving distance is very long in a short time; there is a point with high signal delay, that is, the moving distance is very short in a long time; there are obvious outliers in the data, that is, there is a large distance between a certain point and other points at adjacent times, and there is obviously an anomaly.

[0150] In the embodiments of the present application, the trajectories with abnormal data in the historical trajectory data of the N sample vehicles can be filtered and deleted, and the interpolation method can be used to correct and complete the trajectories with missing data, to obtain the historical trajectory data of the M sample vehicles.

[0151] S602, generate a sample data set according to the historical trajectory point data of the M sample vehicles at t-m time to t+n time.

[0152] In the embodiments of the present application, when obtaining the historical trajectory data of the M sample vehicles, the historical trajectory in any end time length in the historical trajectory data can be used as the data for training the vehicle trajectory prediction model. That is, the historical trajectory point data of the M sample vehicles at t-m time to t+n time is used as the training data.

[0153] ​In the embodiments of the present application, model training generally requires sample data and corresponding label data. Therefore, the historical trajectory point data of the M sample vehicles at the t-m time to the t+n time can be preprocessed, and a sample data set is generated according to the preprocessed historical trajectory point data of the sample vehicles at the t-m time to the t+n time. The sample data set includes sample data and corresponding label data.

[0154] Specifically, the first sample trajectory representation sequence generated by the historical trajectory data of each sample vehicle and the label trajectory representation sequence can be used, the first sample trajectory representation sequence includes the historical trajectory point data of the sample vehicle at the t-m time to the t time, and the label trajectory representation sequence includes the historical trajectory point data of the sample vehicle at the t+1 time to the t+n time. The historical trajectory point data includes the position and speed of the trajectory point.

[0155] Optionally, to reduce errors in the model training process, the historical trajectory point data of the M sample vehicles at the t-m time to the t+n time can be normalized to convert all data to data in the range of [0, 1].

[0156] The normalization process can be performed according to the following formula:

[0157]

[0158] In the formula, x is a certain trajectory representation value in the original data, such as X or Y in the world coordinate, max and min represent the maximum and minimum values of the representation value, and x* represents the standardized representation value.

[0159] S603, using the sample data set, training the vehicle trajectory prediction model to obtain a trained vehicle trajectory prediction model.

[0160] In the embodiments of the present application, the first sample trajectory representation of each sample vehicle in the sample data can be input into the vehicle trajectory prediction model to obtain the prediction result output by the vehicle trajectory prediction model.

[0161] The value of the loss function of the prediction result and the corresponding label trajectory representation sequence is obtained, and if the value of the loss function meets the preset threshold, the model training is completed. If the value of the loss function does not meet the preset threshold, the parameters of the vehicle trajectory prediction model can be updated according to the value of the loss function, and the vehicle trajectory prediction model with updated parameters is used to repeat the above training process until a trained vehicle trajectory prediction model is obtained.

[0162] Optionally, the loss function value can be calculated using the root mean square error RMSE, that is, the square root of the ratio of the sum of the squares of the difference between the predicted value and the label value to the number.

[0163]

[0164] wherein, real t is a label value, pre t is a prediction value.

[0165] Optionally, the optimizer of the vehicle trajectory prediction model can be an Adam optimizer, the learning rate can be set to 0.001, the batch size can be set to 256, and the activation function can be set to a Leaky Relu function.

[0166] The training method of the vehicle trajectory prediction model provided in the embodiments of the present application can improve the accuracy of the vehicle trajectory prediction model in predicting vehicle trajectories by training the vehicle trajectory prediction model.

[0167] Figure 2 The training method of the vehicle trajectory prediction model provided in the embodiments of the present application can improve the accuracy of the vehicle trajectory prediction model in predicting vehicle trajectories by training the vehicle trajectory prediction model. Figure 6 , on the basis of the embodiments shown in Figure 7 , the process of model training is further described, as shown in Figure 8 , comprising the following steps:

[0168] S701, dividing the sample data set into a training set and a test set.

[0169] In the embodiments of the present application, to improve the accuracy of training the vehicle trajectory prediction model, the sample data set can be divided into a training set and a test set, the model is trained according to the training set, and the trained model is tested according to the test set.

[0170] Illustratively, the sample data set can be divided into a training set and a test set according to a ratio of 7:3, the training set includes: M1 first sample trajectory representation sequences of M1 sample vehicles, and label trajectory representation sequences, and the test set includes: M2 first sample trajectory representation sequences of M2 sample vehicles, and label trajectory representation sequences, M1+M2≤M.

[0171] S702, training the vehicle trajectory prediction model based on the training set.

[0172] In the embodiments of the present application, the first sample trajectory representation sequence of each sample vehicle in the training set can be input into the vehicle trajectory prediction model respectively to obtain the prediction result of the vehicle trajectory prediction model.

[0173] The value of the loss function of the prediction result and the corresponding label trajectory representation sequence is obtained, if the value of the loss function meets the preset threshold, the model training is completed. If the value of the loss function does not meet the preset threshold, the parameters of the vehicle trajectory prediction model can be updated according to the value of the loss function, and the vehicle trajectory prediction model with updated parameters is used to repeat the above training process until the loss function meets the preset threshold.

[0174] The vehicle trajectory prediction model meeting the needs of the training set is brought into the test set for testing.

[0175] S703, test the vehicle trajectory prediction model trained based on the test set on the training set, to obtain the trained vehicle trajectory prediction model.

[0176] In the embodiments of the present application, the first sample trajectory representation sequence of each sample vehicle in the test set can be input into the vehicle trajectory prediction model respectively, to obtain the prediction result of the vehicle trajectory prediction model.

[0177] The value of the loss function of the prediction result and the corresponding label trajectory representation sequence is obtained, and if the value of the loss function meets the preset threshold, the trained vehicle trajectory prediction model is obtained.

[0178] If the loss function does not meet the preset threshold, the number of layers of GRU in the vehicle trajectory prediction model can be increased, and the steps shown in S701 to S703 are re-executed until the value of the loss function of the prediction result output by the vehicle trajectory prediction model in the test set and the corresponding label trajectory representation sequence meets the preset threshold, and the trained vehicle trajectory prediction model is obtained.

[0179] In summary, the vehicle trajectory prediction method and the trajectory prediction model training method provided by the embodiments of the present application, as shown in Figure 9 The vehicle trajectory prediction model is trained by the historical trajectory data of the target vehicle, the prediction trajectory output by the vehicle trajectory prediction model is obtained, the Kalman filter is used to optimize the prediction trajectory, and the optimized prediction trajectory is output.

[0180] According to the prediction trajectory output by the vehicle trajectory prediction model, the accuracy of trajectory prediction in a long time domain can be improved. By optimizing the prediction trajectory through Kalman filtering, the influence of noise in the training process can be reduced, and the accuracy of vehicle trajectory prediction is further improved, thereby improving the driving safety.

[0181] On the basis of the above-mentioned embodiments, the embodiments of the present application further provide a vehicle trajectory prediction device.

[0182] Figure 10 A structural schematic diagram of a vehicle trajectory prediction device 90 provided by the embodiments of the present application, comprising:

[0183] The acquisition module 901 is configured to acquire a first trajectory representation sequence of a first target vehicle, wherein the first trajectory representation sequence comprises trajectory point data of the first target vehicle at t-m time to t time, and the trajectory point data comprises position and speed of the trajectory point.

[0184] The processing module 902 is configured to input the first trajectory representation sequence into a vehicle trajectory prediction model to obtain a second trajectory representation sequence output by the vehicle trajectory prediction model, the second trajectory representation sequence comprising predicted trajectory point data of the first target vehicle at times t+1 to t+n, where m and n are integers greater than 1.

[0185] The optimization module 903 is configured to optimize each trajectory point data in the second trajectory representation sequence according to the trajectory point data of the first target vehicle at time t to obtain a target trajectory of the first target vehicle at times t+1 to t+n.

[0186] The control module 904 is configured to perform intelligent driving control on the second target vehicle according to the target trajectory.

[0187] Optionally, the optimization module 903 is further configured to obtain estimated trajectory point data of each trajectory point data in the second trajectory representation sequence according to the observation data and the trajectory point data of the first target vehicle at time t, and obtain optimized trajectory point data according to the estimated trajectory point data of each trajectory point data in the multiple motion modes corresponding to the trajectory point data.

[0188] Optionally, the optimization module 903 is further configured to obtain initial estimated trajectory point data of each trajectory point data in the multiple motion modes corresponding to the trajectory point data according to a state equation corresponding to each motion mode and the trajectory point data of the first target vehicle at time t, obtain a gain matrix of each initial estimated trajectory point data in each motion mode, and obtain estimated trajectory point data of each trajectory point data in the multiple motion modes corresponding to the trajectory point data according to the observation data, the gain matrix of each initial trajectory point data in each motion mode, and the initial estimated trajectory point data of each trajectory point data in the multiple motion modes corresponding to the trajectory point data.

[0189] Optionally, the optimization module 903 is further configured to perform weighted processing on the estimated trajectory point data of each trajectory point data in the multiple motion modes corresponding to the trajectory point data according to a weight corresponding to each motion mode to obtain optimized trajectory point data.

[0190] The vehicle trajectory prediction device provided by the embodiments of the present application can execute the technical solutions of the vehicle trajectory prediction method provided by any of the above embodiments, and has similar implementation principles and technical effects, which will not be described here again.

[0191] On the basis of the above embodiments, the embodiments of the present application further provide a vehicle trajectory prediction model training device.

[0192] Figure 11 FIG. 1 is a schematic structural diagram of a trajectory prediction device 100 of a vehicle according to an embodiment of the present application, comprising:

[0193] The acquisition module 1001 is configured to acquire historical trajectory data of M sample vehicles.

[0194] The processing module 1002 is configured to generate a sample data set according to historical trajectory point data of the M sample vehicles at t-m time to t+n time.

[0195] The sample data set comprises a first sample trajectory representation sequence generated by using the historical trajectory data of each sample vehicle, and a label trajectory representation sequence. The first sample trajectory representation sequence comprises historical trajectory point data of the sample vehicle at t-m time to t time, and the label trajectory representation sequence comprises historical trajectory point data of the sample vehicle at t+1 time to t+n time. The historical trajectory point data comprises position and speed of a trajectory point. M, m and n are all integers greater than 1.

[0196] The training module 1003 is configured to train a vehicle trajectory prediction model by using the sample data set, to obtain a trained vehicle trajectory prediction model.

[0197] Optionally, the training module 1003 is further configured to divide the sample data set into a training set and a test set. The training set comprises a first sample trajectory representation sequence of M1 sample vehicles, and a label trajectory representation sequence. The test set comprises a first sample trajectory representation sequence of M2 sample vehicles, and a label trajectory representation sequence. M1+M2≤M.

[0198] Optionally, the training module 1003 is further configured to, in the process of training the vehicle trajectory prediction model: if a first loss function value of the vehicle trajectory prediction model based on the training set is greater than a preset value, update parameters of the vehicle trajectory prediction model. If the first loss function value is less than or equal to the preset value, and a second loss function value of the vehicle trajectory prediction model based on the test set is greater than the preset value, add a feature extraction layer in the vehicle trajectory prediction model. If the first loss function value is less than or equal to the preset value, and the second loss function value is greater than the preset value or equal to, determine that the training of the vehicle trajectory prediction model is completed.

[0199] Optionally, the processing module 1002 is further configured to pre-process historical trajectory point data of the M sample vehicles at t-m time to t+n time. The sample data set is generated according to the pre-processed historical trajectory point data of the sample vehicles at t-m time to t+n time.

[0200] The vehicle trajectory prediction model training apparatus provided by the embodiments of the present application can execute the technical solutions of the vehicle trajectory prediction model training method provided by any of the embodiments described above, and has similar implementation principles and technical effects, which will not be described here again.

[0201] Figure 11 An electronic device 110 structure schematic diagram is provided by the embodiments of the present application. As shown in the figure, the electronic device 110 can include at least one processor 1101, a memory 1102. ​

[0202] The memory 1102 is configured to store a program. Specifically, the program can include program code, and the program code includes computer operation instructions. The memory 1102 can include a high-speed RAM memory, and can also include a non-volatile memory such as at least one disk memory.

[0203] The processor 1101 is configured to execute the computer operation instructions stored in the memory 1102 to implement the technical solutions of the vehicle trajectory prediction model training method or the vehicle trajectory prediction method embodiment described in the foregoing method embodiments.

[0204] The processor 1101 can be a central processing unit (CPU) or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0205] Optionally, when the electronic device 110 executes the technical solutions of the vehicle trajectory prediction method embodiment, the electronic device 110 can further include a sensor 1103.

[0206] The sensor 1103 is configured to collect vehicle trajectory data in front of the road and transmit the collected vehicle trajectory data to the memory 1102 to implement the technical solutions of the vehicle trajectory prediction method embodiment described in the foregoing method embodiments.

[0207] ​Optionally, the electronic device 110 can further include a communication interface 1104, through which the electronic device 110 can communicate with external devices, such as user terminals (e.g., mobile phones, tablets). In a specific implementation, if the communication interface 1104, the memory 1102, and the processor 1101 are implemented independently, the communication interface 1104, the memory 1102, and the processor 1101 can be connected to each other through a bus and complete communication between each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc., but it does not mean that there is only one bus or one type of bus.

[0208] Optionally, in a specific implementation, if the communication interface 1104, the memory 1102, and the processor 1101 are integrated on a chip, the communication interface 1104, the memory 1102, and the processor 1101 can complete communication through an internal interface.

[0209] The embodiments of the present application also provide a computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the technical solutions of the training method of the vehicle trajectory prediction model or the trajectory prediction method of the vehicle according to the embodiments described above, and the implementation principles and technical effects are similar, and will not be repeated here.

[0210] The embodiments of the present application also provide a computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the technical solutions of the training method of the vehicle trajectory prediction model or the trajectory prediction method of the vehicle according to the embodiments described above, and the implementation principles and technical effects are similar, and will not be repeated here.

[0211] In a possible implementation, the computer readable medium can include a random access memory (RAM), a read-only memory (ROM), a compact disc read-only memory (CD-ROM) or other optical memory, a magnetic disk storage or other magnetic storage devices, or any other medium that is suitable for storing desired program code in the form of instructions or data structures and that can be accessed by a computer. Moreover, any connection is properly termed a computer readable medium. For example, if the software is transmitted from a website, server, or other remote source using a coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of medium. Disk and disc, as used herein, include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk, and Blu-ray® disc where disks usually reproduce data magnetically, while discs reproduce data optically with lasers. Combinations of the above should also be included within the scope of computer readable media.

[0212] The embodiments of the present application also provide a computer program product, including a computer program, which, when executed by a processor, implements the technical solutions of the above-mentioned vehicle trajectory prediction model training method or vehicle trajectory prediction method embodiments, and has similar implementation principles and technical effects, which will not be described here.

[0213] In the specific implementation of the terminal device or the server, it should be understood that the processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as hardware processor execution, or executed by a combination of hardware and software modules in the processor.

[0214] Those skilled in the art can understand that all or part of the steps of any of the method embodiments described above can be completed by hardware related to program instructions. The foregoing program can be stored in a computer readable storage medium, and when the program is executed, all or part of the steps of the foregoing method embodiments are executed.

[0215] If the technical solutions of the present application are realized in the form of software and sold or used as products, they can be stored in a computer readable storage medium. Based on this understanding, all or part of the technical solutions of the present application can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a computer program or a plurality of instructions. The computer software product enables a computer device (which can be a personal computer, a server, a network device or similar electronic equipment) to execute all or part of the steps of the method described in the embodiments of the present application.

[0216] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A vehicle trajectory prediction method, characterized by, The method comprises the following steps: obtaining a first trajectory representation sequence of a first target vehicle, the first trajectory representation sequence comprising trajectory point data of the first target vehicle at time t-m to time t, the trajectory point data comprising position and speed of a trajectory point; inputting the first trajectory representation sequence into a vehicle trajectory prediction model to obtain a second trajectory representation sequence output by the vehicle trajectory prediction model, the second trajectory representation sequence comprising predicted trajectory point data of the first target vehicle at time t+1 to time t+n; m and n are both integers greater than 1; obtaining initial estimated trajectory point data of each trajectory point data in each motion mode according to the state equation corresponding to each motion mode and the trajectory point data of the first target vehicle at time t; obtaining a gain matrix of each initial estimated trajectory point data in each motion mode respectively; obtaining estimated trajectory point data of each trajectory point data in each motion mode according to observation data, the gain matrix of each initial trajectory point data in each motion mode, and the initial estimated trajectory point data of each trajectory point data in each motion mode; the observation data is each trajectory point data in the second trajectory representation sequence; obtaining optimized each trajectory point data according to the estimated trajectory point data of each trajectory point data in each motion mode; obtaining a target trajectory of the first target vehicle at time t+1 to time t+n according to the optimized each trajectory point data; controlling intelligent driving of a second target vehicle according to the target trajectory.

2. The method of claim 1, wherein, The method of obtaining optimized each trajectory point data according to the estimated trajectory point data of each trajectory point data in each motion mode comprises: weighting the estimated trajectory point data of each trajectory point data in each motion mode according to the weight corresponding to each motion mode to obtain optimized each trajectory point data.

3. The method according to claim 1 or 2, characterized in that, The method further comprises: obtaining historical trajectory data of M sample vehicles; generating a sample data set according to historical trajectory point data of the M sample vehicles at time t-m to time t+n, the sample data set comprising a first sample trajectory representation sequence generated by using the historical trajectory data of each sample vehicle and a label trajectory representation sequence, the first sample trajectory representation sequence comprising historical trajectory point data of the sample vehicle at time t-m to time t, and the label trajectory representation sequence comprising historical trajectory point data of the sample vehicle at time t+1 to time t+n; the historical trajectory point data comprises position and speed of a trajectory point; M, m and n are all integers greater than 1; training a vehicle trajectory prediction model by using the sample data set to obtain a trained vehicle trajectory prediction model.

4. The method of claim 3, wherein, The method of training a vehicle trajectory prediction model by using the sample data set to obtain a trained vehicle trajectory prediction model comprises: The sample data set is divided into a training set and a test set, the training set comprising: a first sample trajectory representation sequence of M1 sample vehicles, and a label trajectory representation sequence, and the test set comprising: a first sample trajectory representation sequence of M2 sample vehicles, and a label trajectory representation sequence, M1+M2≤M; In the process of training the vehicle trajectory prediction model: If a first loss function value of the vehicle trajectory prediction model based on the training set is greater than a preset value, the parameters of the vehicle trajectory prediction model are updated; If the first loss function value is less than or equal to the preset value, and a second loss function value of the vehicle trajectory prediction model based on the test set is greater than the preset value, a feature extraction layer in the vehicle trajectory prediction model is added; If the first loss function value is less than or equal to the preset value, and the second loss function value is less than or equal to the preset value, it is determined that the training of the vehicle trajectory prediction model is completed.

5. A vehicle trajectory prediction device characterized by comprising: Comprising: An acquisition module is configured to acquire a first trajectory representation sequence of a first target vehicle, the first trajectory representation sequence comprising trajectory point data of the first target vehicle at a time t-m to a time t, the trajectory point data comprising a position and a speed of a trajectory point; A processing module is configured to input the first trajectory representation sequence into a vehicle trajectory prediction model to obtain a second trajectory representation sequence output by the vehicle trajectory prediction model, the second trajectory representation sequence comprising predicted trajectory point data of the first target vehicle at a time t+1 to a time t+n, m and n are both integers greater than 1; An optimization module is configured to acquire initial estimated trajectory point data of each trajectory point data in multiple motion modes according to a state equation corresponding to each motion mode and trajectory point data of the first target vehicle at the time t, acquire a gain matrix of each initial estimated trajectory point data in each motion mode respectively, acquire estimated trajectory point data of each trajectory point data in multiple motion modes according to observation data, the gain matrix of each initial trajectory point data in each motion mode, and the initial estimated trajectory point data of each trajectory point data in multiple motion modes, the observation data being each trajectory point data in the second trajectory representation sequence, and acquire each optimized trajectory point data according to the estimated trajectory point data of each trajectory point data in multiple motion modes, and obtain a target trajectory of the first target vehicle at the time t+1 to the time t+n according to each optimized trajectory point data; A control module is configured to perform intelligent driving control on a second target vehicle according to the target trajectory.

6. An electronic device, comprising: Comprising: A memory is configured to store a computer program; A processor is configured to execute the computer program to implement the method in any one of claims 1-4.

7. A computer readable storage medium characterized by A computer program is stored thereon, and the computer program is executed by a processor to implement the method in any one of claims 1-4.

Images