Left Lane Change Research Articles

Autonomous driving decision-making control algorithm based on hierarchical reinforcement learning

In the decision-making model of autonomous driving based on learning, the reinforcement learning model has the problems of slow convergence and single application scenario, while the imitation learning model has the problem of poor generalization. In order to solve this problem, a two-layer reinforcement learning framework is proposed to replace the decision layer and control layer in the autonomous driving task. The decision layer divides the driving behavior into lane keeping, left lane change and right lane change. After the decision layer selects the corresponding behavior, it completes the behavior by changing the input of the control layer. Then, a new method for training the control layer, RL_COE (Reinforcement learning combined with online experts), is proposed by combining reinforcement learning and online experts. Finally, a highway simulation environment is built in Carla to verify the proposed algorithm and compare it with the reinforcement learning baseline algorithm. The results show that this method greatly improves the convergence speed and stability of the algorithm and can better complete the driving task.

Multi-head attention-based intelligent vehicle lane change decision and trajectory prediction model in highways

With the aim to improve the interaction between intelligent vehicles and human drivers, this article proposes the MCLG (multi-head attention + convolutional social pooling + long short-term memory + Gaussian mixture model) lane change decision and trajectory prediction model, which includes a lane-changing intention decision module. The model comprises a lane change decision module responsible for determining three lane change intentions: left lane change, right lane change, and car-following. Subsequently, a multi-head attention mechanism processes complex vehicle interaction information to enhance modeling accuracy and intelligence. In addition, uncertainty in trajectory prediction is considered by using multimodal trajectory prediction and Gaussian mixture model, and diversity and uncertainty are combined by combining trajectory prediction from several different modalities through probabilistic combinatorial sampling patterns. Test results indicate that the MCLG model, based on the multi-head attention module, outperforms existing methods in trajectory prediction. The decision module, which takes interactive information into account, exhibits higher predictability and accuracy. Furthermore, the MCLG model, considering the lane-changing decision module, significantly enhances trajectory prediction accuracy, providing robust decision-making support for autonomous driving systems.

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.

Driving Intention Recognition of Surrounding Vehicles Based on a Time-Sequenced Weights Hidden Markov Model for Autonomous Driving.

Accurate perception, especially situational awareness, is central to the evolution of autonomous driving. This necessitates understanding both the traffic conditions and driving intentions of surrounding vehicles. Given the unobservable nature of driving intentions, the hidden Markov model (HMM) has emerged as a popular tool for intention recognition, owing to its ability to relate observable and hidden variables. However, HMM does not account for the inconsistencies present in time series data, which are crucial for intention recognition. Specifically, HMM overlooks the fact that recent observations offer more reliable insights into a vehicle's driving intention. To address the aforementioned limitations, we introduce a time-sequenced weights hidden Markov model (TSWHMM). This model amplifies the significance of recent observations in recognition by integrating a discount factor during the observation sequence probability computation, making it more aligned with practical requirements. Regarding the model's input, in addition to easily accessible states of a target vehicle, such as lateral speed and heading angle, we also introduced lane hazard factors that reflect collision risks to capture the traffic environment information surrounding the vehicle. Experiments on the HighD dataset show that TSWHMM achieves recognition accuracies of 94.9% and 93.4% for left and right lane changes, surpassing both HMM and recurrent neural networks (RNN). Moreover, TSWHMM recognizes lane-changing intentions earlier than its counterparts. In tests involving more complex roundabout scenarios, TSWHMM achieves an accuracy of 87.3% and can recognize vehicles' intentions to exit the roundabout 2.09 s in advance.

Driver Maneuver Detection and Analysis Using Time Series Segmentation and Classification.

The current paper implements a methodology for automatically detecting vehicle maneuvers from vehicle telemetry data under naturalistic driving settings. Previous approaches have treated vehicle maneuver detection as a classification problem, although both time series segmentation and classification are required since input telemetry data are continuous. Our objective is to develop an end-to-end pipeline for the frame-by-frame annotation of naturalistic driving studies videos into various driving events including stop and lane-keeping events, lane changes, left-right turning movements, and horizontal curve maneuvers. To address the time series segmentation problem, the study developed an energy-maximization algorithm (EMA) capable of extracting driving events of varying durations and frequencies from continuous signal data. To reduce overfitting and false alarm rates, heuristic algorithms were used to classify events with highly variable patterns such as stops and lane-keeping. To classify segmented driving events, four machine-learning models were implemented, and their accuracy and transferability were assessed over multiple data sources. The duration of events extracted by EMA was comparable to actual events, with accuracies ranging from 59.30% (left lane change) to 85.60% (lane-keeping). Additionally, the overall accuracy of the 1D-convolutional neural network model was 98.99%, followed by the long-short-term-memory model at 97.75%, then the random forest model at 97.71%, and the support vector machine model at 97.65%. These model accuracies were consistent across different data sources. The study concludes that implementing a segmentation-classification pipeline significantly improves both the accuracy of driver maneuver detection and the transferability of shallow and deep ML models across diverse datasets.

Ensemble deep learning-based lane-changing behavior prediction of manually driven vehicles in mixed traffic environments

<abstract><p>Accurately predicting lane-changing behaviors (lane keeping, left lane change and right lane change) in real-time is essential for ensuring traffic safety, particularly in mixed-traffic environments with both autonomous and manual vehicles. This paper proposes a fused model that predicts vehicle lane-changing behaviors based on the road traffic environment and vehicle motion parameters. The model combines the ensemble learning XGBoost algorithm with the deep learning Bi-GRU neural network. The XGBoost algorithm first checks whether the present environment is safe for the lane change and then evaluates the likelihood that the target vehicle will make a lane change. Subsequently, the Bi-GRU neural network is used to accurately forecast the lane-changing behaviors of nearby vehicles using the feasibility of lane-changing and the vehicle's motion status as input features. The highD trajectory dataset was utilized for training and testing the model. The model achieved an accuracy of 98.82%, accurately predicting lane changes with an accuracy exceeding 87% within a 2-second timeframe. By comparing with other methods and conducting experimental validation, we have demonstrated the superiority of the proposed model, thus, the research achievement is of utmost significance for the practical application of autonomous driving technology.</p></abstract>

Directional lane change prediction using machine learning methods

This research employs a series of machine learning methods to predict the direction of lane change. The response is a binary variable indicating changing the lane to the left or to the right. The employed methods include Decision Tree, Discriminant Analysis, Naïve Bayes, Support Vector Machine, k-Nearest Neighbor and Ensemble. The results are compared to the conventional logistic regression method. Both performance criteria and computational times are reported for comparison purposes. A design of experiments is run to test 25 classification methods at ratios of 25%, 50%, and 75% right to left lane change data. Moreover, samples are validated by cross and holdback validation methods. RUS Boosted trees, an ensemble method, shows improvement over logistic regression. This research provides valuable insights on lane change behaviour, including trajectories and driving styles, which falls into the field of microscopic lane change study.

Attention-Based Lane Change and Crash Risk Prediction Model in Highways

Lane change and crash risk prediction are critical technologies for autonomous driving. An attention-based LSTM model is proposed in this paper for lane change behavior prediction in highways, considering both the surrounding vehicles’ information around the target vehicle at the current time and the historical trajectory data of the vehicle, and it shows higher accuracy and better interpretability than other models. There are two sections in this prediction model, including a pre-judgment model based on the C4.5 decision tree and bagging ensemble learning, and a multi-step lane change prediction model of LSTM with attention mechanism. In addition, the model is applied in the NGSIM datasets to test practicability and accuracy, and the precision of left lane change and right lane change is 98% and 94% respectively 1 second before the vehicle changes lane. Finally, to judge the safety of vehicles in the driving process, a crash risk prediction model based on Time-To-Collision is proposed, which further verifies the effectiveness of the method.

An improved risk estimation model of lane change using naturalistic vehicle trajectories

Lane change (LC) behavior has critical effects on traffic flows and safety due to its complex interactions with surrounding vehicles. To ensure safe lane changes and prevent potential crashes, it is important to recognize the potential crash risk of lane change in real time. This study proposes an improved risk estimation (IRE) model to evaluate the potential collision risk of lane change (LCR) vehicle groups. The safety margin is introduced to consider the deceleration capability of vehicles to measure the reaction time of drivers during the LC. Then the IRE model is established, incorporating the collision probability and collision severity measured based on the safety margin. The trajectory data, extracted from the highD dataset, are used and 1536 LC samples are investigated. We compare the LCR under different contextual factors, including vehicle types (cars and trucks), two lane change directions (left and right lane change, LLC and RLC), and traffic flows (low and high traffic). It was found that truck drivers keep higher LCR compared with car drivers due to limited brake capacity, and the left lane change results in higher LCR compared with the right lane change. Additionally, lane change is associated with higher crash risk in high traffic flow, as compared to low traffic flow. The understanding of the crash risk of lane change behavior under different contextual factors, can be useful for real-time crash prediction and devising traffic management strategies.

Design and Implementation of an Asynchronous Finite State Controller for Wheeled Mobile Robots

Wheeled mobile robots (WMRs) can navigate in uncontrolled environments with the assistance of electronic or physical devices. Several works have been conducted on the control and management of the path-tracking of a vehicle in different road scenarios. This paper aims to create an asynchronous finite state control law for a WMR. The control law is based on a proportional–integral–derivative controller, and the performance of the proposed model is evaluated in virtual and real environments in two different scenarios. In the first one, the WMR must perform a zig-zag maneuver between obstacles, while the second one involves a double left lane change. In the proposed scenarios, the WMR drives along a path until an obstacle is detected at less than 50 cm, causing the WMR to check whether the first lane is free to go and move on. These scenarios and the related required engineering approaches seem particularly suitable for system engineering in a student’s laboratory for the design and implementation of automated guidance system modeling.

Guest editorial: Decision making and control for connected and automated vehicles

Guest editorial: Decision making and control for connected and automated vehicles

Lane change maneuver detection considering real-time vehicle dynamic features via V2X communication

More than 90% of traffic accidents are caused by driver behavior, with lane change behavior being a major contributor. Recently, driving assistance systems are being introduced on vehicles to reduce traffic accidents, and a reliable vehicle lane change collision detection system is a key component of these systems. Besides, the foundation of the vehicle lane change detection system is the effective vehicle lane change detection model. In this paper, based on the support vector machine, we propose a model for detecting driver lane change maneuvers and take into account the real-time vehicle dynamic features transmitted via Vehicle to X (V2X) Communication. The accuracy is ideal for lane keep and lane change situations, and it is also robust for zigzag driving situations, according to tests conducted using the NGSIM real traffic dataset. The detection accuracy for left and right lane change maneuvers is 97.5% and 99.09%, respectively, while the false alarm rate is 8.56%. Additionally, the average advance detection time is 1.7 s, which is suitable for actual driving application scenarios.

Design, Development and Validation of an Intelligent Collision Risk Detection System to Improve Transportation Safety: The Case of the City of Popayán, Colombia

Several approaches from different perspectives have been used to solve problems with traffic accidents (TA), which mainly affect low- and middle-income countries. Conditions of certain cities, regarding road infrastructure, enforcement of traffic safety regulations, and motor vehicle numbers, influence the increase in TAs. Therefore, medium-sized cities in developing countries (context of interest), which commonly have worrying conditions, are a relevant scenario. One of the approaches to reduce TAs has been the use of data analysis through Machine Learning (ML); however, these techniques require a large amount of data, and medium-sized cities commonly do not have enough. Techniques such as Naturalistic Driving (ND) can be applied as a data collection method. This work proposes an intelligent collision risk detection system (ICDRS) using ND and ML to improve sustainability and safety of transportation in medium-sized cities. The ICRDS design considered the limitations of the context of interest and uses two data collection devices in the vehicle. The ICRDS validation included the design and execution of tests using ND. This validation identified if the collected data in a certain time interval contained high-risk collision events (sudden acceleration, sudden braking, aggressive left or right turn, aggressive left or right lane change). The system implementation results were satisfactory. The developed ML algorithm obtained an average value 0.98 in all the metrics. Two data sets of driving on routes were collected. In addition, the performed tests were able to identify city areas with high accident rates.

Event‐triggered optimisation of overtaking decision‐making strategy for autonomous driving on highway

Appropriate overtaking can improve the safety and efficiency of driving. Practically, overtaking consists of two asymmetric lane-changes: firstly to the left lane and then return back. Existing literatures treated the two lane-changes as the same behaviour, which may result in frequent and unnecessary lane-changes. For the more complex first lane-change, discretionary lane-change strategies have been developed to gain speed advantage with least deceleration. Most of them, however, described vehicles' dynamics with rough simplifications different from actual execution (e.g. constant speed and constant acceleration), which may fail them in practical applications. This paper presents an event-triggered overtaking decision-making strategy for autonomous vehicles to make the overtaking meet the driver's expectations. Firstly, the difference between the left-lane change and right-lane change is analysed. Left-lane change is mainly concerned with the efficiency, whereas driving safety is the main target of the right-lane change. Secondly, an event-triggering condition covering the speed difference between the ego vehicle and its preceding vehicle is set. Afterwards, based on quantified speed advantage, considering surrounding vehicles' changes in both speed and acceleration, an optimisation of overtaking decision-making strategy is proposed. During right-lane change in an overtaking, safety is ensured by predicting whether the ego vehicle will collide with surrounding vehicles. Finally, several scenarios are defined in the combination of SCANeR studio and Simulink, which are used to test and demonstrate the effectiveness and applicability of the overtaking decision-making strategy. The results show that the proposed overtaking decision-making strategy, by mathematical quantification of speed advantage, performs well in measuring the motivation of lane-change and avoiding frequent lane changes, which correspond to driver habits.

Driver Glance Behavior Modeling Based on Semi-Supervised Clustering and Piecewise Aggregate Representation

Glance behavior is significant because whether and how the driver is scanning and observing the driving scene is closely related to driving safety. This paper aims to improve the accuracy of glance behavior modeling and realize the spatiotemporal representation and visualization of glance behavior. Forty subjects were recruited to perform a freeway driving task using a driving simulator. The vehicle data were collected by the simulator. Drivers’ gaze points were collected by an eye tracker. The prior knowledge on gaze points obtained through a statistical analysis were provided for K-means (KM) to form a semi-supervised K-means (SSKM), which classifies gaze points into different fixation zones. The classification results were compared with the results of KM. Furthermore, a clustering center-based piecewise aggregate representation (CCPAR) was proposed to characterize glance behavior. Maneuvers identification was taken as a case to evaluate the proposed method. The k-nearest neighbour (KNN) based on the similarity of CCPAR identified driving maneuvers into lane-keeping, left lane change, and right lane change. The identification results were compared with the results of the Hidden Markov Model (HMM). The average classification accuracies of KM and SSKM were 55.28% and 94.75%, respectively. The accuracies of maneuvers identified by CCPAR-KNN and by HMM were 87.50% and 85.83%, respectively. The results indicate that SSKM and CCPAR are feasible for glance behavior modeling. SSKM eliminates the randomness of initial cluster center selection and improves the accuracy of gaze points classification. CCPAR is intuitive and convenient to describe and visualize the spatiotemporal characteristics of glance behavior.

An integrated lane change prediction model incorporating traffic context based on trajectory data

Predicting lane change maneuvers is critical for autonomous vehicles and traffic management as lane change may cause conflict in traffic flow. Most existing studies do not consider the effect of traffic context (i.e., traffic level and vehicle type) on lane change maneuvers. Therefore, these models cannot adapt to different traffic environments. This study aims to address this problem and establish an integrated lane change prediction model incorporating traffic context using machine learning algorithms. In addition, lane change decisions and lane change trajectories are both predicted to capture the whole process, which have been less studied. The framework of the proposed model contains two parts: the traffic context classification model, which is used to predict traffic level and vehicle type, and the integrated lane change prediction model, which is used to predict lane change decision with XGBoost and lane change trajectories with LSTM incorporating context information. Instead of considering lane change, we establish trajectory prediction models for left lane change and right lane change, further improving the prediction accuracy. The naturalistic trajectories of the highD dataset are used to train and validate the model. The results show that the proposed model improves the accuracy from 97.02% to 98.20% when predicting lane change decision that incorporate traffic context. In addition, the MSE decreases from 11.21 to 6.62 when predicting trajectories. The proposed models are also validated on NGSIM dataset, proving the adaptability of the model. The proposed model can be applied to different environments to reduce collision risks caused by lane change maneuvers and improve traffic management and driving safety.

Does LSTM outperform 4DDTW-KNN in lane change identification based on eye gaze data?

Lane change identification based on long short-term memory (LSTM) neural networks has received increasing attention. The most commonly known disadvantage of the approach is the “black box” nature. This study, to improve the interpretability, proposed a novel method for identifying lane changes. The philosophy was to measure the similarity of the glance behavior during the maneuver. The glance behavior in the same maneuver is more similar, while that in different maneuvers is different. First, a driving simulator was used to collect driving behavior data. The eye gaze time series were captured by the eye tracker. The median of the eye gaze sequence of 3 s before the initial moment of lane change was obtained. With the median as the center, the driver's visual field plane was divided into several grids with the side length of 600 pixels. A sliding space-time cuboid algorithm was proposed to extract the scanning path. Second, four-dimensional dynamic time warping (4DDTW) distance was used to measure the similarity of the scanning paths. Third, the K nearest neighbor (KNN) algorithm was used to classify the driving maneuver into the lane-keeping (LK), the right lane change (RLC), and the left lane change (LLC) based on the 4DDTW distance of the scan paths. LSTM was also used to classify the driving maneuver into LK, RLC, and LLC. The performances of 4DDTW-KNN and LSTM were compared. The accuracies of 4DDTW-KNN and LSTM were 86.50% and 86.33%, respectively. LSTM is not better than 4DDTW-KNN in lane change identification based on eye gaze data with a comprehensive consideration of the time efficiency, the accuracy and the interpretability.

Lane change strategy analysis and recognition for intelligent driving systems based on random forest

The development of intelligent connected technology provides a platform for multidimensional information interaction and makes the recognition of a specific lane change strategy become a reality. Intelligent driving systems cannot make human-like decisions by only recognizing left or right lane change behaviors; in fact, poor environmental cognitive competence is the main cause of accidents of intelligent driving systems during road tests. Therefore, the recognition of specific lane change strategies can provide support for safety performance improvement. This research aims to identify the different lane change strategies of a subject vehicle with a rear approaching vehicle in the target lane. For the purpose of acquiring different strategies, a total of 42 experienced drivers including 37 males and 5 females participated in the on-road vehicle experiments, and we divided lane change strategies into three categories: mandatory, yielding, and waiting for lane change. The lane change duration time, lane crossing time, and characteristic parameters including distance to lane line, steering wheel angle, relative distance, and relative speed under different lane change strategies were compared and analyzed. Then the random forest classifier was employed to construct a lane change strategy identification model. Results indicated that the maximum value of global recognition accuracy under three input parameters reached 88.78% when the time window was 1.2 s, which was higher than the maximum value of recognition accuracy under four input parameters, but the identification time was 0.1 s later than the model with four input parameters. In addition, the recognition performance of the random forest model, attention-bidirectional long short-term memory (Atten-BiLSTM) model and genetic algorithm-support vector machine (GA-SVM) model was compared. The results demonstrated that the global recognition accuracy and identification time of the proposed model were better than that of GA-SVM model under different input parameters, and the performance was comparable to the Atten-BiLSTM model. The findings provide the basis for the intelligent driving system to make more human-like decisions and improve safety performance.

Online Risk-Bounded Motion Planning for Autonomous Vehicles in Dynamic Environments

A crucial challenge to efficient and robust motion planning for autonomous vehicles is understanding the intentions of the surrounding agents. Ignoring the intentions of the other agents in dynamic environments can lead to risky or overconservative plans. In this work, we model the motion planning problem as a partially observable Markov decision process (POMDP) and propose an online system that combines an intent recognition algorithm and a POMDP solver to generate risk-bounded plans for the ego vehicle navigating with a number of dynamic agent vehicles. The intent recognition algorithm predicts the probabilistic hybrid motion states of each agent vehicle over a finite horizon using Bayesian filtering and a library of pre-learned maneuver motion models. We update the POMDP model with the intent recognition results in real time and solve it using a heuristic search algorithm which produces policies with upper-bound guarantees on the probability of near colliding with other dynamic agents. We demonstrate that our system is able to generate better motion plans in terms of efficiency and safety in a number of challenging environments including unprotected intersection left turns and lane changes as compared to the baseline methods.

Lane Change Prediction With an Echo State Network and Recurrent Neural Network in the Urban Area

The prediction of lane changes can reduce traffic accidents and improve traffic flow. In this paper two classifiers, Echo State Network and a recurrent neural network with Long Short Term Memory cells, were compared to predict lane changes using the input variables steering angle and indicator. The input variables were extracted from a data set which was generated from a naturalistic driving study in the urban area of Chemnitz, Germany. Both classifiers predicted left and right lane changes successfully. They achieved high true positive rates and low false positive rates. The Echo State Network predicted left and the recurrent neural network predicted right lane changes better.