Target Vehicle Research Articles

Vehicle Trajectory Prediction Considering Multi-feature Independent Encoding Based on Graph Neural Network

Background: Today, self-driving cars are already on the roads. However, driving safety remains a huge challenge. Trajectory prediction of traffic targets is one of the important tasks of an autonomous driving environment perception system, and its output trajectory can provide necessary information for decision control and path planning. Although there are many patents and articles related to trajectory prediction, the accuracy of trajectory prediction still needs to be improved. Objective: This paper aimed to propose a novel scheme that considers multi-feature independent encoding trajectory prediction (MFIE). Methods: MFIE is an independently coded trajectory prediction algorithm that consists of a spacetime interaction module and trajectory prediction module, and considers speed characteristics and road characteristics. In the spatiotemporal interaction module, an undirected and weightless static traffic graph is used to represent the interaction between vehicles, and multiple graph convolution blocks are used to perform data mining on the historical information of target vehicles, capture temporal features, and process spatial interaction features. In the trajectory prediction module, three long short-term memory (LSTM) encoders are used to encode the trajectory feature, motion feature, and road constraint feature independently. The three hidden features are spliced into a tensor, and the LSTM decoder is used to predict the future trajectory. Results: On datasets, such as Apollo and NGSIM, the proposed method has shown lower prediction error than traditional model-driven and data-driven methods, and predicted more target vehicles at the same time. It can provide a basis for vehicle path planning on highways and urban roads, and it is of great significance to the safety of autonomous driving. Conclusion: This paper has proposed a multi-feature independent encoders’ trajectory prediction data-driven algorithm, and the effectiveness of the algorithm is verified with a public dataset. The trajectory prediction algorithm considering multi-feature independent encoders provides some reference value for decision planning.

Lightweight air-to-air unmanned aerial vehicle target detection model

The rapid expansion of the drone industry has resulted in a substantial increase in the number of low-altitude drones, giving rise to concerns regarding collision avoidance and countermeasure strategies among these unmanned aerial vehicles. These challenges underscore the urgent need for air-to-air drone target detection. An effective target detection model must exhibit high accuracy, real-time capabilities, and a lightweight network architecture to achieve a balance between precision and speed when deployed on embedded devices. In response to these requirements, we initially curated a dataset comprising over 10,000 images of low-altitude operating drones. This dataset encompasses diverse and intricate backgrounds, significantly enhancing the model’s training capacity. Subsequently, a series of enhancements were applied to the YOLOv5 algorithm to realize lightweight object detection. A novel feature extraction network, CF2-MC, streamlined the feature extraction process, while an innovative module, MG, in the feature fusion section aimed to improve detection accuracy and reduce model complexity. Concurrently, the original CIoU loss function was replaced with the EIoU loss function to further augment the model’s accuracy. Experimental results demonstrate an enhancement in the accuracy of drone target detection, achieving mAP values of 95.4% on the UAVfly dataset and 82.2% on the Det-Fly dataset. Finally, real-world testing conducted on the Jetson TX2 revealed that the YOLOv5s-ngn model achieved an average inference speed of 14.5 milliseconds per image. The code utilized in this paper can be accessed via https://github.com/lucien22588/yolov5-ngn.git.

Structure design and advantage research for bidirectional synchronous force-increasing EMB actuator

Electro-Mechanical Brake (EMB) can provide an ideal brake actuator for the active braking system of intelligent vehicles and the decoupled regenerative braking system of new energy vehicles. In order to solve the problem of large volume, weight, and axial size of the existing EMB actuator, a novel bidirectional synchronous force-increasing EMB (BSF-EMB) actuator scheme is proposed. Based on the maximum required braking force of the target vehicle under emergency braking conditions and the limitation of installation space in the wheel hub. The rotary motor and the ball screw are selected, the force-increasing mechanism parameters calculation and structure design are completed. In order to minimize the difference between the force-increasing ratio of the two-stage force-increasing mechanism and the lever force-increasing mechanism. The parameters of the force-increasing mechanism are optimized based on the weight adaptive particle swarm optimization algorithm. The three-dimensional model of the EMB actuator is established. In addition, when the required braking force is the same, the mass and volume of the BSF-EMB actuator can be decreased by about 43.4% and 55.8%, compared with the existing planetary gear force-increasing EMB (PGF-EMB) actuator scheme. Therefore, the BSF-EMB actuator can effectively decrease the unsprung mass of the vehicle and resolve the issue of limited installation space in the wheel hub of compact vehicles. The EMB models for two EMB actuator structures are built based on MATLAB/Simulink, and the EMB braking response speed and tracking performance of braking force are compared. The simulation results show that the braking response speed of the BSF-EMB actuator can be increased by 11%, 10.6%, 12.7%, and 14% compared to the PGF-EMB actuator under emergence braking condition and other three general service braking conditions, and the braking force adjustment is more precise. Finally, the effectiveness for the control strategy is verified by hardware in loop (HIL) experiment under sin wave input condition for braking force. The related research can provide reference for the design and application of EMB.

An Empirical Study of the Policy Processes behind Norway’s BEV-Olution

Norway’s large battery electric vehicle (BEV) market and fleet are not the result of a comprehensive policy plan. Using the multiple streams (MS) framework and document analysis, it was identified that the most important Norwegian BEV policy decisions were made using inadequate policy processes that fall outside of traditional politics. This is contrary to the MS framework postulate that three independent streams of problems, policy solutions, and politics must align to pave the way for new policies. Politicians had limited information about the effects of policies they introduced in this “learning by doing process”. Impact assessments were rarely made. The decision rationale was often not documented. The future market expectation and thus the national budget consequences were low when important policy decisions were made, whereas the political gain was high. The processes were more aligned with traditional politics after 2014. The ambitious ZE vehicle targets for 2025 and the climate policy targets for 2030 locked in incentives, despite rising tax losses. In sum, these developments created the world’s largest per-capita BEV market. To avoid negative issues and keep the BEV policies’ potential to support the BEV transition, politicians should ensure that sufficient knowledge is available when making decisions about future policies. Such decisions should be taken transparently within traditional politics, be properly assessed as with EU policy processes, and regularly reviewed as with the California ZEV mandate process. The required knowledge should be developed in open-access research.

Mining the Micro-Trajectory of Two-Wheeled Non-Motorized Vehicles Based on the Improved YOLOx

Two-wheeled non-motorized vehicles (TNVs) have become the primary mode of transportation for short-distance travel among residents in many underdeveloped cities in China due to their convenience and low cost. However, this trend also brings corresponding risks of traffic accidents. Therefore, it is necessary to analyze the driving behavior characteristics of TNVs through their trajectory data in order to provide guidance for traffic safety. Nevertheless, the compact size, agile steering, and high maneuverability of these TNVs pose substantial challenges in acquiring high-precision trajectories. These characteristics complicate the tracking and analysis processes essential for understanding their movement patterns. To tackle this challenge, we propose an enhanced You Only Look Once Version X (YOLOx) model, which incorporates a median pooling-Convolutional Block Attention Mechanism (M-CBAM). This model is specifically designed for the detection of TNVs, and aims to improve accuracy and efficiency in trajectory tracking. Furthermore, based on this enhanced YOLOx model, we have developed a micro-trajectory data mining framework specifically for TNVs. Initially, the paper establishes an aerial dataset dedicated to the detection of TNVs, which then serves as a foundational resource for training the detection model. Subsequently, an augmentation of the Convolutional Block Attention Mechanism (CBAM) is introduced, integrating median pooling to amplify the model’s feature extraction capabilities. Subsequently, additional detection heads are integrated into the YOLOx model to elevate the detection rate of small-scale targets, particularly focusing on TNVs. Concurrently, the Deep Sort algorithm is utilized for the precise tracking of vehicle targets. The process culminates with the reconstruction of trajectories, which is achieved through a combination of video stabilization, coordinate mapping, and filtering denoising techniques. The experimental results derived from our self-constructed dataset reveal that the enhanced YOLOx model demonstrates superior detection performance in comparison to other analogous methods. The comprehensive framework accomplishes an average trajectory recall rate of 85% across three test videos. This significant achievement provides a reliable method for data acquisition, which is essential for investigating the micro-level operational mechanisms of TNVs. The results of this study can further contribute to the understanding and improvement of traffic safety on mixed-use roads.

Research on speed predictive control system of intelligent connected vehicle based on V2X

To enhance the efficiency of urban road traffic and reduce waiting time at traffic lights, this paper proposes a V2X based speed prediction control method for intelligent connected vehicles. Firstly, this paper uses PreScan to build an intelligent network environment model based on V2X communication technology. Then, the traffic signal and target vehicle information transmitted by V2X are analyzed and processed, leading to the design of an intelligent connected vehicle speed prediction control system. Finally, comparative experiments were conducted using HIL between the conventional vehicle speed decision control system and the V2X-based intelligent connected vehicle speed prediction control system under various operating conditions. The experimental results demonstrate that the proposed speed prediction control system can reduce waiting time for traffic lights and enhance road traffic efficiency. Therefore, this paper’s designed speed prediction control system is both effective and feasible, providing a reference for future research on intelligent connected vehicle speed decision control.

Information Extraction and Three-Dimensional Contour Reconstruction of Vehicle Target Based on Multiple Different Pitch-Angle Observation Circular Synthetic Aperture Radar Data

The circular synthetic aperture radar (CSAR) has the ability of all-round continuous observation and high-resolution imaging detection, and can obtain all-round scattering information and higher-resolution images of the observation scene, so as to realize the target information extraction and three-dimensional (3D) contour reconstruction of the observation targets. However, the existing methods are not accurate enough to extract the information of vehicle targets. Through the analysis of the vehicle target scattering model and CSAR image characteristics, this paper proposes a vehicle target information extraction and 3D contour reconstruction method based on multiple different pitch-angle observation CSAR data. The proposed method creatively utilizes the projection relationship of the vehicle in 2D CSAR imaging to reconstruct the 3D contour of the vehicle, without prior information. Firstly, the CSAR data obtained from multiple different pitch-angle observations are fully utilized, and the scattering points of odd-bounce reflection and even-bounce reflection echoes are extracted from the two-dimensional (2D) coherent CSAR images of the vehicle target. Secondly, the basic contour of the vehicle body is extracted from the scattering points of the even-bounce reflected echoes. Then, the geometric projection relationship of the “top–bottom shifting” effect of odd-bounce reflection is used to calculate the height and position information of the scattering points of odd-bounce reflection, so as to extract the multi-layer 3D contour of the vehicle target. Finally, the basic contour and the multi-layer 3D contour of the vehicle are fused to realize high-precision 3D contour reconstruction of the vehicle target. The correctness and effectiveness of the proposed method are verified by using the CVDomes simulation dataset of the American Air Force Research Laboratory (AFRL), and the experimental results show that the proposed method can achieve high-precision information extraction and realize distinct 3D contour reconstruction of the vehicle target.

Research on Vehicle Detection in Infrared Aerial Images in Complex Urban and Road Backgrounds

The detection of vehicle targets in infrared aerial remote sensing images captured by drones presents challenges due to a significant imbalance in vehicle distribution, complex backgrounds, the large scale of vehicles, and the dense and arbitrarily oriented distribution of targets. The RYOLOv5_D model is proposed based on the YOLOv5-obb rotation model. Firstly, we reconstruct a new vehicle remote sensing dataset, BalancedVehicle, to achieve data balance. Secondly, given the challenges of complex backgrounds in infrared remote sensing images, the AAHE method is proposed to highlight infrared remote sensing vehicle targets while reducing background interference during the detection process. Moreover, in order to address the issue of detecting challenges under complex backgrounds, the CPSAB attention mechanism is proposed, which could be used together with DCNv2. GSConv is also used to reduce the model parameters while ensuring accuracy. This combination could improve the model’s generalization ability and, consequently, enhance the detection accuracy for various vehicle categories. The RYOLOv5s_D model, trained on the self-built dataset BalancedVehicle, demonstrates a notable improvement in its mean average precision (mAP), increasing from 73.6% to 78.5%. Specifically, the average precision (AP) for large aspect ratio vehicles such as trucks and freight cars increases by 11.4% and 8%, respectively. The RYOLOv5m_D and RYOLOv5l_D models achieve accuracies of 82.6% and 84.3%. The Param of RYOLOv5_D is similar to that of the YOLOv5-obb, while possessing a decrease in computational complexity of 0.6, 4.5, and 12.8GFLOPS. In conclusion, the RYOLOv5_D model’s superior accuracy and real-time capabilities in infrared remote sensing vehicle scenarios are validated by comparing various advanced models based on rotation boxes on the BalancedVehicle dataset.

Modeling the effect of daughter migration on dosimetry estimates for unlabeled actinium-225.

Actinium-225 (225Ac) is an alpha emitting radionuclide which has demonstrated promising results in Targeted Alpha Therapy (TAT). A concern with 225Ac is that the decay energy can break the bond to the targeting vehicle, resulting in the release of free alpha-emitting daughter radionuclides in the body. The aim of this work is to develop a compartment model to describe the movement of unlabeled 225Ac in a human where the daughter isotopes of 225Ac have unique biokinetics. The ICRP Occupational Intake of Radionuclides reports were used to construct a compartment model for the 225Ac decay chain where the daughter isotopes of 225Ac are assigned their own unique transfer coefficients (TCs) between compartments. Computer simulations were performed for unlabeled 225Ac uniformly placed in the plasma and only the dose from alpha particles was considered. Absorbed doses to normal organs were determined for the liver, kidneys, bone, soft tissue, active marrow, and blood. Simulations were performed for the case when: (1) the daughters have unique biokinetics and (2) the daughters decay at the site of 225Ac. When the daughters have unique biokinetics, the organs that receive the highest absorbed dose are the liver (male: 1466.6 mGy/MBq, female: 1885.7 mGy/MBq), bone (male: 293.6 mGy/MBq, female: 403.6 mGy/MBq) and kidneys (male: 260.8 mGy/MBq, female: 294.0 mGy/MBq). These doses were compared to the case when the daughters of 225Ac decay at the site of 225Ac. There was a 13.5% increase in kidney dose, a 0.8% decrease in liver dose, and <0.1% decrease in bone dose calculations when the daughters have unique biokinetics compared to assuming the daughters decay at the site of 225Ac. The kidneys received a large dose estimate (260-295 mGy/MBq) as well as a considerable change in dose of +13.5% when the daughters have unique biokinetics compared to assuming the daughters decay at the site of 225Ac. Therefore, to accurately determine the kidney dose from unlabeled 225Ac in a human, the biokinetics of the daughter isotopes should be considered.

ADAS Simulation Result Dataset Processing Based on Improved BP Neural Network

The autonomous driving simulation field lacks evaluation and forecasting systems for simulation results. The data obtained from the simulation of target algorithms and vehicle models cannot be reasonably estimated. This problem affects subsequent vehicle improvement and parameter calibration. The authors relied on the simulation results of the AEB algorithm. We selected the BP Neural Network as the basis and improved it with a genetic algorithm optimized via a roulette algorithm. The regression evaluation indicators of the prediction results show that the GA-BP neural network has better prediction accuracy and generalization ability than the original BP neural network and other optimized BP neural networks. This GA-BP neural network also fills the Gap in Evaluation and Prediction Systems.

Speed Change Pattern Optimization for Improving the Electricity Consumption of an Electric Bus and Its Verification Using an Actual Vehicle

In this study, we focused on the eco-driving of electric vehicles (EVs). The target vehicle is an electric bus developed by our research team. Using the parameters of the bus and speed pattern optimization algorithm, we derived the EV’s eco-driving speed pattern. Compared to the eco-driving of internal combustion engine vehicles (ICVs), we found several different characteristics. We verified these characteristics with actual vehicle driving test data of the target bus, and the results confirmed its rationality. The EV’s eco-driving method can improve electricity consumption by about 10–20% under the same average speed.

Typical motion-based modelling and tracking for vehicle targets in linear road segment

In this paper, a typical motion-based modelling and tracking issue is investigated for vehicle targets in linear road segment by a variable structure multiple model (VSMM) method. Vehicle target motions in a road with multiple lanes are described by typical lane keeping and lane changing manoeuvres. To describe trajectories of typical manoeuvres, an Ornstein–Uhlenbeck process and a sine half-cycle are used to model lane keeping and lane changing motions in lateral direction, respectively. Note that starting point and length of lane changing motion are unknown in target tracking. Time-varying model sets are designed based on the current motion model with different motion parameters. A VSMM tracking frame is constructed to obtain target state estimates with time-varying model set. The effectiveness of the proposed typical motion-based tracking scheme is displayed by simulation results on road vehicle tracking.

Research on Over-the-Horizon Perception Distance Division of Optical Fiber Communication Based on Intelligent Roadways.

With the construction and application of more and more intelligent networking demonstration projects, a large number of advanced roadside digital infrastructures are deployed on both sides of the intelligent road. These devices sense the road situation in real time through algorithms and transmit it to edge computing units and cloud control platforms through high-speed optical fiber transmission networks. This article proposes a cloud edge terminal architecture system based on cloud edge cooperation, as well as a data exchange protocol for cloud control basic platforms. The over-the-horizon scene division and optical fiber network communication model are verified by deploying intelligent roadside devices on the intelligent highway. At the same time, this article uses the optical fiber network communication algorithm and ModelScope large model to model inference on real-time video data. The actual data results show that the StreamYOLO (Stream You Only Look Once) model can use the Streaming Perception method to detect and continuously track target vehicles in real-time videos. Finally, the method proposed in this article was experimentally validated in an actual smart highway digital infrastructure construction project. The experimental results demonstrate the high application value and promotion prospects of the fiber optic network in the division of over the horizon perception distance in intelligent roadways construction.

A Car-Following Model Considering Missing Data Based on TransGAN Networks

Car-following behavior is closely related to the longitudinal control of the vehicle, affecting the safety of the vehicle and traffic flow stability. In order to interact with the preceding vehicle, the target vehicle usually collects the driving data of the preceding vehicle. However, data acquisition devices often face malfunctions caused by various unpredictable disruptions, resulting in missing value problems. This may cause the target vehicle to make wrong control decisions. Given this situation, a new car-following(CF) model considering missing data based on Transformer-Generative Adversarial Networks (TransGAN) is proposed. Firstly, Transformer Network with multi-head attention is used to deeply extract the potential features from incomplete vehicle state data, which can filter important information from the input and focus on these, while capturing long distance dependencies. Secondly, a Generative Adversarial Network is constructed. The Generator generates the future multi-step control states of the target vehicle based on the features extracted by Transformer Network. The Discriminator with a fully connected network is applied to simultaneously ensure the generation accuracy. Finally, our proposed model was trained and tested on a publicly available NGSIM I-80 dataset. Compared with other existing advanced works, our model can fit the actual control states of the target vehicle with higher accuracy under different data missing rates of the preceding vehicle, which demonstrates that the proposed method effectively improves the robustness of vehicle longitudinal car-following control under missing data.

Classification and Prediction of Vehicle Lane‐Changing Crash Risk Levels Based on Video Trajectory Data

To investigate the potential lane‐changing collision risks that may arise between vehicles during lane changes and those in the original lane, a model for vehicle lane‐changing collision risk is constructed specifically for this scenario, and a research analysis is conducted. First, based on vehicle trajectory data, a sample set capturing the relationships between vehicles traveling in a straight line and those changing lanes laterally is extracted and built. Interpolation methods are then applied to fill in missing values, outliers are eliminated, and data noise is smoothed during preprocessing. After preprocessing, a total of 468 vehicle pairs and 265,392 data points are obtained. Second, a real‐time collision time model is established based on the preprocessed data, and collision risk probabilities are calculated accordingly. Then, the collision risks are classified into four levels based on whether the vehicle on the side actually changes lanes and the severity of the collision risks. Finally, a light gradient boosting machine (LightGBM) learning method is adopted to predict the risk levels and analyze the main factors that significantly impact the severity of collision risks. The results indicate that the longitudinal distance between the target vehicle and the preceding vehicle is the most critical influencing factor, followed by the speed of the target vehicle itself, and then the speed difference between the target vehicle and the preceding vehicle. The influence of other factors is relatively similar and does not have a significant impact.

An Intelligent Control Method for Industrial Equipment Based on Distributed Electrical and Electronic Architecture

Abstract The article establishes the objective function of distributed electronic and electrical architecture from three dimensions of economy, quantization and loadability, and constructs a mathematical model for multi-objective optimization based on the satisfaction of constraints. The model was applied and tested in the intelligent control system of automotive industrial equipment. In practical applications, the intelligent control system can automatically reduce the speed to 16km/h to maintain a safe distance and prevent collision when the target vehicle is monitored to enter the lane it is in and the distance between the two cars is less than 30 meters. Under 1500 pcu/h traffic flow conditions, the average delays of vehicles at the intersection through intelligent control were only 8.06 seconds and 8.02 seconds, compared to 20.58 seconds and 23.29 seconds for vehicles under conventional control. Vehicle intelligent control systems based on distributed electronic and electrical architectures have great potential for development and significant safety performance improvement in automotive industrial equipment manufacturing. The research in this paper provides a new perspective and adequate technical support for the intelligent control of future industrial equipment.

Research on Intelligent Vehicle Operation Risk Assessment and Early Warning Based on Predictive Risk Field

In order to enhance the driving safety of intelligent vehicles in complex road scenarios, a method for vehicle operation risk assessment and early warning based on the predictive risk field is proposed. The temporal feature vector composed of the spatiotemporal state characteristics of the ego vehicle and surrounding traffic participants is taken as input data for the Attention‐Bidirectional Long‐Short Term Memory (Attention‐BiLSTM) model, which is trained to establish the desired mapping relationship. By predicting the motion state of the target vehicle and utilizing an improved risk field model based on the target vehicle of heading angle, the predictive risk field is obtained. This allows for the assessment of the ego vehicle operational risks. The risk warning model is integrated to provide risk early warning, and the safety path for the ego vehicle is planned based on the interaction between the predictive risk field equipotential lines and the cubic spline curves. Experimental results demonstrate that the proposed vehicle operation risk assessment and early warning model is effective in providing early warnings and safe path references for the ego vehicle in complex urban road test scenarios.

Research on Vehicle Detection Algorithm Based on Embedded ARM

Abstract Based on the theory of machine learning in the field of artificial intelligence, this paper proposes to use the computer vision platform OpenCV to construct an embedded ARM vehicle detection model. Determine the ARM embedded software and hardware and adopt Haar features for the Adaboost algorithm to design the OpenCV vehicle classifier. Cross-compile the ARM chip using Linux to generate new firmware for OpenMV. Use the DFU tool for embedded ARM chips to upgrade and re-burn them into the embedded development board for machine vision OpenMV. By using the classifier file and OpenCV’s image processing algorithm, the work of vehicle detection and recognition is completed, and the vehicle target is labeled with a candidate box in the picture and video. The results demonstrate that the algorithm in this paper maintains the leakage detection rate and false detection rate below 5% in four different working conditions: strong light, normal light, weak light, and nighttime, thereby fully demonstrating the effectiveness of the research conducted in this paper.

Nanoscale engineering of solid-state materials for boosting hydrogen storage.

The development of novel materials capable of securely storing hydrogen at high volumetric and gravimetric densities is a requirement for the wide-scale usage of hydrogen as an energy carrier. In recent years, great efforts via nanoscale tuning and designing strategies on both physisorbents and chemisorbents have been devoted to improvements in their thermodynamic and kinetic aspects. Increasing the hydrogen storage capacity/density for physisorbents and chemisorbents and improving the dehydrogenation kinetics of hydrides are still considered a challenge. The extensive and fast development of advanced nanotechnologies has fueled a surge in research that presents huge potential in designing solid-state materials to meet the ultimate U.S. Department of Energy capacity targets for onboard light-duty vehicles, material-handling equipments, and portable power applications. Different from the existing literature, in this review, particular attention is paid to the recent advances in nanoscale engineering of solid-state materials for boosting hydrogen storage, especially the nanoscale tuning and designing strategies. We first present a short overview of hydrogen storage mechanisms of nanoscale engineering for boosted hydrogen storage performance on solid-state materials, for example, hydrogen spillover, nanopump effect, nanosize effect, nanocatalysis, and other non-classical hydrogen storage mechanisms. Then, the focus is on recent advancements in nanoscale engineering strategies aimed at enhancing the gravimetric hydrogen storage capacity of porous materials, reducing dehydrogenation temperature and improving reaction kinetics and reversibility of hydrogen desorption/absorption for metal hydrides. Effective nanoscale tuning strategies for enhancing the hydrogen storage performance of porous materials include optimizing surface area and pore volume, fine-tuning nanopore sizes, introducing nanostructure doping, and crafting nanoarchitecture and nanohybrid materials. For metal hydrides, successful strategies involve nanoconfinement, nanosizing, and the incorporation of nanocatalysts. This review further addresses the points to future research directions in the hope of ushering in the practical applications of hydrogen storage materials.

Vehicle Lane Change Multistep Trajectory Prediction Based on Data and CNN_BiLSTM Model

In order to accurately predict the lane‐changing trajectory of the vehicle and improve the driving safety of the vehicle, a lane‐changing trajectory prediction model based on the combination of convolutional neural network (CNN) and bidirectional long short‐term memory (BiLSTM) neural network is proposed by comprehensively considering the historical driving behavior, the spatial characteristics of surrounding vehicles and the bidirectional time sequence information of the vehicle trajectory. Firstly, the vehicle trajectory data are filtered and smoothed, and it is divided into three categories: left lane change, right lane change, and straight driving, and a lane change trajectory sample set is constructed. Secondly, CNN‐BiLSTM model is constructed to identify the sample set of lane‐changing trajectory. Considering the interaction between vehicles in the driving process, the information of predicted vehicle, and surrounding vehicles is taken as the input of the model. The extracted feature vector is input to the BiLSTM layer for prediction after the CNN layer feature extraction, and the horizontal and vertical coordinates of the target vehicle at the next time are output. Thirdly, the trajectory data of the US‐101 dataset in NGSIM is selected to verify the performance of the CNN‐BiLSTM model, and at the same time, it is compared with models such as CNN‐LSTM, long short‐term memory (LSTM), BiLSTM, and CNN‐GRU‐Att. Finally, the verification result shows that the overall fitting degree of the vehicle lane change trajectory prediction of the proposed model reaches 99.50%, and the mean square error and mean absolute error are 0.0003076 and 0.01417, which are improved compared with other models. In the meanwhile, the research on multistep trajectory prediction in different prediction time domains is carried out. It was found that the longer the prediction time domain is, the lower the prediction performance of the model decreases, but the prediction accuracy still reached more than 96%, and it was able to accurately predict the lane change trajectory.