Vehicle track prediction method considering in-transit driving style

Abstract

The invention belongs to the technical field of automatic driving safety, and particularly relates to a vehicle track prediction method considering an in-transit driving style. According to the invention, from the scene perception layer, the mode layer, the operation layer and the vehicle state layer, the state of the target vehicle is comprehensively understood, and the provided trajectory prediction method considering the in-transit driving style integrates the advantages of rule data driving. And a knowledge-driven method is adopted to convert the current state of the target vehicle from an abstract data level into a driving behavior semantic space. According to the method, the understanding of human beings on the real world can be simulated, and the prediction effect of the trajectory prediction model is displayed in different driving behavior semantic spaces and different in-transit driving style types.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving safety, and in particular relates to a vehicle trajectory prediction method that takes into account on-the-road driving style. Background Art

[0002] Autonomous vehicles have attracted widespread attention as a potentially safer and more sustainable mode of transportation. Researchers aim to improve traffic safety and passenger comfort through autonomous driving technology, thereby improving the travel experience of travelers. Vehicle trajectory prediction is an important research topic for autonomous vehicles.

[0003] Before entering the era of fully autonomous driving, there will be a mixed traffic state of human-driven cars and autonomous cars for a long time in the future. Autonomous cars need to interact with surrounding human-driven vehicles and predict their running trajectories to ensure driving safety. Vehicle trajectory prediction already has a relatively complete research system. At this stage, the research on trajectory prediction focuses more on long-term trajectory characteristics. The trajectory prediction model aims to improve the accuracy of trajectory prediction, and rarely considers the impact of individual characteristics of human drivers and driving environment on model results. The complexity of the real traffic environment and the uncertainty of human driver driving style bring new opportunities and challenges to trajectory prediction. Human drivers in the surrounding environment have individual heterogeneity and inconsistency. Individual heterogeneity refers to the differences in characteristics between different drivers, and inconsistency refers to the differences between the same driver in different driving scenarios. At the same time, the existing trajectory prediction model has achieved certain success in the data-driven mode, but the data-driven method is a training for specific driving scenarios or specific data. Due to the influence of data, overfitting problems may occur. The effect is significant on the training data set, but the model lacks generalization and scalability on other data sets. This problem leads to huge challenges for autonomous vehicles in the face of diverse and unpredictable scenarios in real complex scenarios. Summary of the invention

[0004] In order to solve the above problems, the present invention provides a method that takes into account on-the-road driving, the heterogeneity and inconsistency of driving styles during driving operations, and integrates the attention mechanism to propose an improved bidirectional long short-term memory neural network. The network introduces the attention mechanism, which can assign different weights to historical trajectory sequences. The attention mechanism can make the model pay more attention to time series data that have a greater impact on future results, and assign higher weights to historical sequences that have a greater impact on results. The network improves the BiLSTM model, and improves the accuracy of the model in long-term trajectory prediction problems by integrating the attention mechanism, and significantly improves the prediction accuracy when the driver's driving mode and operating state change.

[0005] Compared with the traditional data-driven method, the knowledge-driven method enables the autonomous vehicle to fully understand the surrounding environment of the target vehicle, improves the interpretability of the trajectory prediction system, and makes it easier for humans to understand the decision-making behavior of the autonomous vehicle. The on-the-way driving style recognition method based on driving behavior semantic understanding proposed in the present invention starts from the scene perception layer, mode layer, operation layer and vehicle state layer, and comprehensively understands the state of the target vehicle. The proposed trajectory prediction method considering the on-the-way driving style integrates the advantages of rule data driving, and adopts a knowledge-driven method to convert the current state of the target vehicle from the abstract data level into the driving behavior semantic space. This method can simulate human understanding of the real world and display the prediction effect of the trajectory prediction model in different driving behavior semantic spaces and different on-the-way driving style types.

[0006] The technical solution of the present invention:

[0007] The vehicle trajectory prediction method considering the on-the-road driving style has the following specific steps:

[0008] Step (1) Data acquisition and data preprocessing

[0009] The vehicle trajectory data is obtained through the V2X device with a frequency of no less than 10Hz, including vehicle position, longitudinal speed, lateral speed, longitudinal acceleration, lateral acceleration, headway (THW), expected time to collision (TTC) and perception data (surrounding vehicle information, weather, road information). The original vehicle trajectory data is smoothed using the Savitzky-Golay filter.

[0010] Step (2) Construction of driving behavior semantic space and on-the-road driving style recognition

[0011] The driving behavior semantic space is a multi-dimensional concept that aims to comprehensively capture and analyze the driver's behavior patterns in different driving situations. The present invention divides the driving behavior semantic space into four levels from the macro level to the micro level: scene perception layer, mode layer, operation layer and vehicle status layer. Among them, the scene perception layer (macro level) mainly refers to the weather information, road environment information, surrounding vehicle information, etc. obtained by the vehicle during driving. The mode layer refers to the state information such as the vehicle chooses to follow closely, follow long distances or change lanes frequently in a specific scenario after determining the driving path. The operation layer is manifested as whether the driver has sudden acceleration, sudden deceleration and other behaviors in the current mode. The vehicle status layer (micro level) refers to micro-level information such as the current vehicle speed and acceleration.

[0012] The vehicle trajectory data obtained in step (1) is used to understand the driving behavior semantics. The abstract vehicle trajectory data is converted into understandable driving behavior semantic information, and a driving behavior semantic space is constructed to comprehensively describe and understand the driver's driving process. The converted driving behavior semantic data is input into the LDA topic model for topic identification. The number of driving style topics is determined through topic perplexity, topic consistency, and driving style topic visualization distribution. Finally, the characteristics of each driving style topic are determined through topic analysis to obtain the probability of the driver behaving as aggressive, moderate, conservative, or stable.

[0013] The LDA topic model is a widely used topic discovery model, including a three-layer Bayesian structure of documents, words, and topics. It was originally proposed in the field of natural language processing and can reveal the potential topics of documents. The main idea is that the generation of an article is subject to probability distribution, that is, each word selects a topic with a certain probability, and selects a certain word from this topic with a certain probability. In the LDA model, each document in the document collection is generated by a mixture of multiple topics, and each topic is composed of multiple words. The mathematical representation of this model usually involves several steps, including:

[0014]

[0015] z m,n ∣θ m ~Miltinomial(θ m )(2)

[0016] θ m ~Dirichlet(α)(3)

[0017]

[0018] Among them, θ m Represents the topic distribution in document m, sampled from the Dirichlet distribution Dirichlet(α). represents the word distribution in topic k, obtained by sampling from the Dirichlet distribution (β). m,n refers to the topic assignment of the nth word in document m, which is distributed by Multinomial(θ m ) is sampled. m,n refers to the nth word in document m, and z is assigned according to the topic of the word m,n From the multinomial distribution Formula (1) represents word w m,n The generation process of m,nSelect words from the corresponding word distribution. Formula (2) shows how to select topic z for each word m,n , which depends on the topic distribution of document m. Formula (3) shows how the topic distribution of each document is sampled, which depends on the Dirichlet parameter α. Formula (4) shows how the word distribution in each topic is sampled, which depends on the Dirichlet parameter β. The LDA topic model assumes that the topic generation process is the same for all documents. The parameters α and β of these distributions usually need to be obtained through training data.

[0019] Step (3) Design of trajectory prediction dataset integrating on-the-road driving style

[0020] Design a data input module that contains basic information about people, vehicles, and the environment to ensure that the influence of different factors can be fully considered during trajectory prediction.

[0021] The human factor (EDS) is the four driving styles identified by the LDA topic model in step (2): stable (DS0), aggressive (DS1), moderate (DS2), and conservative (DS3). The on-the-road driving style of each driver is the probability that the driver exhibits these four driving styles, and the sum of the probabilities of the driver exhibiting these four driving styles is 1.

[0022] EDS={DS0,DS1,DS2,DS3} (5)

[0023] Vehicle factors (TV) include lateral position x, longitudinal position y, vehicle speed information (including lateral speed v x and longitudinal velocity v y ) and acceleration information (including lateral acceleration a x and longitudinal acceleration a y ).

[0024] TV={x,y,v x ,v y ,a x ,a y} (6)

[0025] Environmental factors (EV) take into account that the front vehicle has the most significant impact on the predicted vehicle, and the headway distance (DHW), time headway (THW) and collision time (TTC) are selected as characteristic parameters.

[0026] EV={DHW,THW,TTC} (7)

[0027] Step (4) Model network structure design

[0028] The attention mechanism is widely used in the field of artificial intelligence. It imitates the attention process in human vision and thinking. Its main purpose is to enable the model to selectively focus on specific parts of the input data instead of processing all information equally.

[0029] The EDTA-BiLSTM model considering on-the-road driving style is constructed by integrating the attention mechanism. The complex dependencies of time series data are captured by the bidirectional long short-term memory neural network (BiLSTM) layer. First, the 13-dimensional feature vector {DS0, DS1, DS2, DS3, x, y, v x ,v y ,a x ,a y ,DHW,THW,TTC} as input, the characteristic factors affecting the target vehicle’s driving trajectory are passed to the network. The input feature vector is first expanded through a fully connected layer to expand the feature space, and then the input feature is passed to a fully connected layer with 256 hidden units for linear transformation to improve the feature representation capability and adapt to the high-dimensional processing requirements of the subsequent BiLSTM layer. The linearly transformed feature vector is sent to the BiLSTM layer. The BiLSTM layer consists of two stacked LSTM sublayers, each of which contains 256 hidden units. It integrates forward and backward information through bidirectional transmission, which optimizes the processing of time series data. The EDTA-BiLSTM network integrates the attention mechanism. By assigning different weights to different parts of the sequence, the network can pay more attention to the information that is effective for prediction during prediction. The output of the BiLSTM layer is passed to the attention layer, and the weighted sum is calculated through the attention layer. The weight parameter is determined by the tanh(x) activation function and a trainable weight vector. The weight is normalized using the Softmax function to generate a probabilistic attention distribution. The attention layer can highlight the importance of the time step, focus limited attention on important information, and pass the output features to two consecutive fully connected layers. Finally, the output layer outputs the prediction results of the target vehicle's position information and speed information, including the lateral position x, longitudinal position y, and lateral speed v. x and longitudinal velocity v y .

[0030] The EDTA-BiLSTM model optimizes the accuracy of trajectory prediction, especially in scenarios that consider historical and future contextual information. Based on the original network, the network structure is improved by integrating the attention mechanism to capture the complex dynamics of the vehicle's motion trajectory. The introduction of the attention mechanism can further improve the sensitivity to key sequence features, and the network can weigh the impact between past trajectory data points and future trajectory data points.

[0031] Step (5) Model training and prediction result evaluation

[0032] The model training set, test set and validation set are randomly divided according to the proportion. The root mean square error (RMSE) and final displacement error (FDE) are used for evaluation.

[0033]

[0034] in, are the values ​​of the horizontal and vertical coordinates of the predicted vehicle trajectory, are the values ​​of the actual horizontal and vertical coordinates of the vehicle trajectory, and n is the number of trajectory points.

[0035] The beneficial effects of the present invention are as follows: the method can realize long-term trajectory prediction for different human drivers, and at the same time, can explain the prediction results for drivers in different driving behavior semantic spaces, thereby improving the problem of poor interpretability of traditional neural network model predictions. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a workflow diagram of the vehicle trajectory prediction method considering the on-the-road driving style of the present invention.

[0037] Figure 2 It is the driving behavior semantic space structure diagram.

[0038] Figure 3 It is a schematic diagram of the vehicle trajectory prediction model considering on-the-road driving style.

[0039] Figure 4 , Figure 5 , Figure 6 This is a visualization of the vehicle trajectory prediction results for three different on-the-road driving styles. DETAILED DESCRIPTION

[0040] The specific implementation modes of the present invention are described in detail below with reference to examples, and the implementation effects of the invention are simulated.

[0041] The vehicle trajectory prediction method considering the on-the-road driving style of the present invention is as follows Figure 1 This example uses the highD vehicle trajectory open source dataset, and through data screening, finally selects 9427 vehicle trajectory data as the research object, and uses the Savitzky-Golay filter to smooth the original trajectory data.

[0042] The present invention establishes a driving behavior semantic division structure including scene perception layer, mode layer, operation layer and vehicle status layer from macro level to micro level, such as Figure 2As shown in the figure, the driving behavior semantics of the trajectory data are understood, the abstract trajectory data is converted into understandable driving behavior semantic information, and the driving behavior semantic space is constructed to comprehensively describe and understand the driver's driving process. The converted driving behavior semantic data is input into the LDA topic model for topic identification, and the number of driving style topics is determined through topic perplexity, topic consistency, and driving style topic visualization distribution; finally, the characteristics of each driving style topic are determined through topic analysis, and the probability of the driver being aggressive, moderate, conservative, and stable is obtained.

[0043] Design a data input module containing basic information about people, vehicles, and the environment to ensure that the influence of different factors can be fully considered in the process of trajectory prediction. The input feature is a 13-dimensional feature vector {DS0, DS1, DS2, DS3, x, y, v x ,v y ,a x ,a y ,DHW,THW,TTC}.

[0044] The EDTA-BiLSTM model of the present invention adopts a sequence-to-sequence (seq2seq) learning framework, such as Figure 3 As shown in Figure 2, it is possible to capture the temporal dependency of the sequence, which is crucial when processing time series data such as vehicle motion paths. The dimension of the input variables is 13, including the human factor (EDS), i.e., the on-the-way driving style of the target vehicle, and the vehicle factor (TV), i.e., {x, y, v x ,v y ,a x ,a y}, and environmental factors (EV), namely {DHW, THW, TTC}. The output of the prediction model is the trajectory information at the future moment, including {x, y, v x ,v y}, using the historical 5s vehicle trajectory data to predict the vehicle trajectory information for the next 5s. The input layer of the model receives a 13-dimensional feature vector, which provides the model with a comprehensive view of the current state of the vehicle. The input feature vector first passes through a fully connected layer, which expands the feature space to accommodate the high-dimensional processing requirements of the subsequent BiLSTM layer. After passing through the fully connected layer, the feature vector is sent to the BiLSTM layer, which consists of two stacked LSTM sublayers, each containing 256 hidden units. This bidirectional structure enables the network to consider both past and future information at the same time, optimizing the processing of time series data. Finally, the output data of the BiLSTM layer is passed to the output layer, outputting a 4-dimensional prediction vector. At the same time, the model also integrates a dropout mechanism to prevent overfitting, and improves the generalization ability of the model by randomly discarding some neuron connections during the training process.

[0045] The EDTA-BiLSTM model training of the present invention is carried out under the Pytorch deep learning framework built in Pycharm. The specific experimental environment is shown in the table. The operating system is Linux system, the memory is 512G, the GPU is NVIDIA GeForce RTX 3090, and the programming language used is Python.

[0046] The parameters that need to be passed in the training process of the EDTA-BiLSTM model of the present invention include n_feature, input_size, hidden_size, n_layers, and output_size. Among them, n_feature is the number of features contained in the input trajectory point. Input_size is the length of the vector input to the LSTM layer, hidden_size is the number of hidden nodes of the LSTM, n_layers is the number of hidden layers, and output_size is the dimension of the output predicted trajectory data. The batch (batch_size) is set to 128, the model optimizer is selected as Adam, the learning rate (learning rate) weight decay (weight decay) is 0.0001, and dropout is 0.2. Dropout is a regularization technique for randomly discarding (setting the weight to zero) a part of neurons during the training process of the neural network to prevent overfitting. This helps to improve the generalization ability of the model and make it perform better on new data. By randomly ignoring certain neurons, dropout forces the network not to rely on specific neurons, thereby reducing the risk of overfitting. The loss function adopts the mean square error loss function (MSELoss).

[0047] This embodiment designs a comparative experiment to verify the effect of the trajectory prediction model considering the on-the-road driving style.

[0048] (1) LSTM: Long short-term memory neural network.

[0049] (2) BiLSTM: Bidirectional long short-term memory neural network.

[0050] (3) ED-BiLSTM: A bidirectional long short-term memory neural network considering on-the-road driving style.

[0051] (4) EDTA-BiLSTM (the present invention): an improved bidirectional long short-term memory neural network integrating attention mechanism.

[0052] The root mean square error and final displacement error results of the model in the 1 to 5 s prediction time domain are shown in Tables 1 and 2.

[0053] Table 1 Comparison of root mean square error (RMSE) of prediction results of different models

[0054]

[0055] Table 2 Comparison of final displacement errors of prediction results of different models

[0056]

[0057] This example aims at the trajectory prediction results of the model for individual drivers with different on-the-road driving styles in different driving behavior semantic spaces. Figure 4 The driving semantic space of the driver in the example is as follows: the speed level is high speed, the acceleration level is safe acceleration, the driving mode is gradually following, the operating state is normal deceleration, the surrounding driving environment is obstacles ahead, and the on-the-way driving style is [0.1666, 0.0001, 0.7983, 0.0348]. The driver mainly exhibits a conservative driving style. The EDTA-BiLSTM prediction result of 5s FDE is 1.23m, and the ED-BiLSTM FDE of the next 5s is 1.48m. Figure 5 The speed level of the driver is moderate, the acceleration level is risk acceleration, the driving mode of the vehicle is restricted lane change, the operating state is normal acceleration, and the surrounding driving environment is obstacles ahead. The probability combination of the driver's on-the-way driving style is [0.1276, 0.0001, 0.0306, 0.8415]. At this time, the driver mainly exhibits an aggressive driving style. According to the visualization diagram, the improved model is significantly improved compared with the original model within 1 to 5 seconds. The 5s final displacement error predicted by the EDTA-BiLSTM model is 1.39m, and the 5s final displacement error of the ED-BiLSTM model is 1.91m. Figure 6 The semantic space of the driver's driving behavior is speed level high speed, acceleration level risk acceleration, driving mode restricted lane change, operation mode continuous acceleration, surrounding environment obstacles ahead, obstacles diagonally ahead, and the probability combination of his on-the-way driving style is [0.0001, 0.0253, 0.0002, 0.9744]. At this time, the driver is mainly aggressive. From the visualization of the prediction results, it can be seen that the original model has a poor prediction effect on the lane change position and lane change trajectory, while the improved model integrating the attention mechanism can well predict the vehicle's lane change trajectory and the starting point of the lane change.

[0058] The present invention integrates the attention mechanism to improve the trajectory prediction model, and by comparing the model prediction effect, evaluates the performance changes of the model before and after the introduction of the attention mechanism and the on-the-way driving style. The RMSE and FDE of the overall prediction results of the model of the present invention are improved. In the long-term trajectory prediction problem in the 5s prediction time domain, the RMSE and FDE can reach 1.28m and 1.17m respectively. The example verification results show that the improved BiLSTM model integrating the attention mechanism is superior to the original model in accuracy and prediction details. In addition, the example discusses the performance of the prediction results of the improved model for different driving behavior semantic spaces and drivers with different on-the-way driving styles. Combining the advantages of data-driven and rule-driven, the knowledge-driven approach is used to improve the interpretability and scalability of the model, which not only improves the accuracy of the vehicle trajectory prediction model, but also contributes an effective tool to the field of long-term series analysis.

Claims

1. A vehicle trajectory prediction method considering on-the-road driving style, characterized in that: The specific steps are as follows: Step (1) Data acquisition and data preprocessing Obtain vehicle trajectory data through V2X equipment at a frequency of no less than 10 Hz, including vehicle position, longitudinal velocity, lateral velocity, longitudinal acceleration, lateral acceleration, headway, expected collision time, and perception data; use Savitzky-Golay filter to smooth the raw vehicle trajectory data; Step (2) Construction of driving behavior semantic space and on-the-road driving style recognition From the macro level to the micro level, the driving behavior semantic space is divided into four levels: scene perception layer, mode layer, operation layer and vehicle status layer; among them, the scene perception layer is the macro level, including the weather information, road environment information and surrounding vehicle information obtained by the vehicle during driving; the mode layer refers to the state information of the vehicle choosing close-following, long-following or frequent lane changing in a specific scenario after determining the driving path; the operation layer is manifested in whether the driver has sudden acceleration, sudden deceleration and other behaviors in the current mode; the vehicle status layer is the micro level, which refers to the current vehicle speed and acceleration information; The vehicle trajectory data obtained in step (1) is used to understand the driving behavior semantics, and the abstract vehicle trajectory data is converted into understandable driving behavior semantic information, and a driving behavior semantic space is constructed. The converted driving behavior semantic data is input into the LDA topic model for topic identification, and the number of driving style topics is determined through topic perplexity, topic consistency, and driving style topic visualization distribution; finally, the characteristics of each driving style topic are determined through topic analysis, and the probability of the driver behaving as aggressive, moderate, conservative, or stable is obtained; Step (3) Design of trajectory prediction dataset integrating on-the-road driving style Design a data input module that contains basic information about people, vehicles, and the environment to ensure that the impact of different factors can be fully considered during trajectory prediction; The human factor EDS is the four driving styles identified by the LDA topic model in step (2): stable DS0, aggressive DS1, moderate DS2, and conservative DS3. The on-the-road driving style of each driver is the probability that the driver exhibits these four driving styles, and the sum of the probabilities of the driver exhibiting these four driving styles is 1. EDS={DS0,DS1,DS2,DS3} (5) The vehicle factor TV includes the lateral position x, longitudinal position y, vehicle speed information and acceleration information; wherein the vehicle speed information includes the lateral speed v x and longitudinal velocity v y ; Acceleration information includes lateral acceleration a x and longitudinal acceleration a y ; TV={x,y,v x ,v y ,to x ,to y } (6) Environmental factor EV takes into account the most significant impact of the preceding vehicle on the predicted vehicle, and selects headway DHW, headway THW and collision time TTC as characteristic parameters; EV={DHW,THW,TTC} (7) Step (4) Model network structure design The EDTA-BiLSTM trajectory prediction model considering on-the-road driving style is constructed by integrating the attention mechanism. The complex dependencies of time series data are captured by the bidirectional long short-term memory neural network BiLSTM layer. First, the 13-dimensional feature vector {DS0, DS1, DS2, DS3, x, y, v x ,v y ,a x ,a y ,DHW,THW,TTC} as input, and pass the characteristic factors affecting the target vehicle's driving trajectory to the network. The input feature vector is first expanded through a fully connected layer to expand the feature space, and then the input feature is passed to a fully connected layer with 256 hidden units for linear transformation; the linearly transformed feature vector is sent to the BiLSTM layer, which is composed of two stacked LSTM sublayers, each of which contains 256 hidden units and integrates forward and backward information through bidirectional transmission; the output of the BiLSTM layer is passed to the attention layer, and the weighted sum is calculated through the attention layer, where the weight parameter is determined by the tanh(x) activation function and a trainable weight vector. The Softmax function is used to normalize the weight to generate a probabilistic attention distribution, and the output feature is passed to two consecutive fully connected layers. Finally, the output layer outputs the predicted results of the target vehicle's position information and speed information, including the lateral position x, longitudinal position y, and lateral speed v x and longitudinal velocity v y ; Step (5) Model training and prediction result evaluation The model training set, test set and validation set are randomly divided according to the proportion; the root mean square error RMSE and final displacement error FDE are used for evaluation; in, are the values ​​of the horizontal and vertical coordinates of the predicted vehicle trajectory, are the values ​​of the actual horizontal and vertical coordinates of the vehicle trajectory, and n is the number of trajectory points.

2. The vehicle trajectory prediction method considering on-the-road driving style according to claim 1, characterized in that: In the LDA model of step (2), each document in the document collection is generated by mixing multiple topics, and each topic is composed of multiple words; the mathematical representation of the LDA model includes: With m,n ∣θ m ~Miltinomial(θ m )(2) i m ~Dirichlet(a)(3) Among them, θ m represents the topic distribution in document m, sampled from the Dirichlet distribution Dirichlet(α); represents the word distribution in topic k, obtained by sampling from the Dirichlet distribution (β); z m,n refers to the topic assignment of the nth word in document m, which is distributed by Multinomial(θ m ) is sampled; w m,n refers to the nth word in document m, and z is assigned according to the topic of the word m,n From the multinomial distribution The formula (1) represents word w m,n The generation process of m,n Select words from the corresponding word distribution; Formula (2) shows how to select topic z for each word m,n , which depends on the topic distribution of document m; Formula (3) shows how the topic distribution of each document is sampled, which depends on the Dirichlet parameter α; Formula (4) shows how the word distribution in each topic is sampled, which depends on the Dirichlet parameter β.

Images