Abstract:Trajectory prediction of the ego vehicle is essential for advanced driver assistance systems to function properly. By recognizing various driving styles and predicting trajectories reflecting them, the prediction performance is enhanced, and a personalized trajectory can be generated. Therefore, we propose to combine driving style recognition and trajectory prediction tasks using only in-vehicle CAN-bus sensor data for possible application to normal vehicles. The DeepConvLstm network was utilized for driving s… Show more
“…In response, hybrid models combining traditional machine learning algorithms with deep learning have emerged. Zhang [22].…”
Section: Literature Reviewmentioning
confidence: 99%
“…In Equations ( 20)- (22), n denotes the total number of samples; x i denotes the i sample j's true value; and y i signifies the predicted value of the model for the same sample i. The smaller the value of MSE, RMSE, and MAE, the more reasonable the corresponding hidden layer's node count.…”
To enhance the lane-changing safety of autonomous vehicles, it is crucial to accurately identify the driving styles of human drivers in scenarios involving the coexistence of autonomous and human-driven vehicles, aiming to avoid encountering vehicles exhibiting hazardous driving patterns. In this study, based on the real traffic flow data from the Next Generation Simulation (NGSIM) dataset in the United States, 301 lane-changing vehicles that meet the criteria are selected. Six evaluation parameters are chosen, and principal component analysis (PCA) is employed for dimensionality reduction in the data. The K-means algorithm is then utilized to cluster the driving styles, classifying them into three categories. Finally, ant colony optimization (ACO) of a backpropagation (BP) neural network model was constructed, utilizing the dimensionality reduction results as inputs and the clustering results as outputs for the purpose of driving style recognition. Simulation experiments are conducted using MATLAB Version 9.10 (R2021a) for comparative analysis. The results indicate that the constructed ACO-BP model achieved an overall recognition accuracy of 96.7%, significantly higher than the recognition accuracies of the BP, artificial neural network (ANN), and gradient boosting machine (GBM) models. The ACO-BP model also exhibited the fastest recognition speed among the four models. Moreover, the ACO-BP model shows varied improvements in recognition accuracy for each of the three driving styles, with an increase of 13.7%, 4.4%, and 4.3%, respectively, compared to the BP model. The simulation results validate the high accuracy, real-time capability, and classification effectiveness of this model in driving style recognition, providing new insights for this field.
“…In response, hybrid models combining traditional machine learning algorithms with deep learning have emerged. Zhang [22].…”
Section: Literature Reviewmentioning
confidence: 99%
“…In Equations ( 20)- (22), n denotes the total number of samples; x i denotes the i sample j's true value; and y i signifies the predicted value of the model for the same sample i. The smaller the value of MSE, RMSE, and MAE, the more reasonable the corresponding hidden layer's node count.…”
To enhance the lane-changing safety of autonomous vehicles, it is crucial to accurately identify the driving styles of human drivers in scenarios involving the coexistence of autonomous and human-driven vehicles, aiming to avoid encountering vehicles exhibiting hazardous driving patterns. In this study, based on the real traffic flow data from the Next Generation Simulation (NGSIM) dataset in the United States, 301 lane-changing vehicles that meet the criteria are selected. Six evaluation parameters are chosen, and principal component analysis (PCA) is employed for dimensionality reduction in the data. The K-means algorithm is then utilized to cluster the driving styles, classifying them into three categories. Finally, ant colony optimization (ACO) of a backpropagation (BP) neural network model was constructed, utilizing the dimensionality reduction results as inputs and the clustering results as outputs for the purpose of driving style recognition. Simulation experiments are conducted using MATLAB Version 9.10 (R2021a) for comparative analysis. The results indicate that the constructed ACO-BP model achieved an overall recognition accuracy of 96.7%, significantly higher than the recognition accuracies of the BP, artificial neural network (ANN), and gradient boosting machine (GBM) models. The ACO-BP model also exhibited the fastest recognition speed among the four models. Moreover, the ACO-BP model shows varied improvements in recognition accuracy for each of the three driving styles, with an increase of 13.7%, 4.4%, and 4.3%, respectively, compared to the BP model. The simulation results validate the high accuracy, real-time capability, and classification effectiveness of this model in driving style recognition, providing new insights for this field.
“…Unsupervised learning [2][3][4][5][6] and semi-supervised learning [7][8][9] methods require a smaller amount of data but face challenges in obtaining reliable sample features within limited data. In situations where data are sufficiently abundant, researchers opt for supervised learning for driver style recognition [10][11][12][13][14][15][16]. This approach achieves high accuracy but demands high requirements for both the quantity and quality of training data.…”
With the increasing demand for road traffic safety assessment, global concerns about road safety have been rising. This is particularly evident with the widespread adoption of V2X (Vehicle-to-Everything) technology, where people are more intensively focused on how to leverage advanced technological means to effectively address challenges in traffic safety. Through the research of driving style recognition technology, accurate assessment of driving behavior and the provision of personalized safety prompts and warnings have become crucial for preventing traffic accidents. This paper proposes a risk field construction technique based on environmental data collected by in-vehicle sensors. This paper introduces a driving style recognition algorithm utilizing risk field visualization and mask learning technologies. The research results indicate that, compared to traditional classical models, the improved algorithm performs excellently in terms of accuracy, stability, and robustness, enhancing the accuracy of driving style recognition and enabling a more effective evaluation of road safety.
“…However, the occupant's driving experience may become worse with this change [5]- [7]. Furthermore, since individual driving styles and operating habits are characterized by a wide range of diversity and dynamic changes [8], [9], it is difficult for a universally designed TTC to meet the driving needs of all individuals [10], [11]. Therefore, the development of humanlike TTC method is of great research significance and value.…”
By improving the ability of Trajectory Tracking Control (TTC) algorithms to mimic the manipulation behaviors of real drivers, which is of great significance in improving the personalized driving experience of autonomous vehicles. In this paper, we propose a TTC method that combines the Adaptive Control of Thought-Rational (ACT-R) cognitive theory framework with the Preview Tracking (PT) theory. Firstly, by analyzing and describing the ACT-R cognitive framework and the PT theory, a TTC framework that combines the two theories is proposed. Secondly, a virtual driving simulator was built to collect driving data from 30 drivers and construct a driving memory database. Thirdly, a central production system for driver trajectory tracking is designed, which consists of: three control modes, production rules, a timing generator, and filtering methods for driving memory segments. Fourthly, the TTC method based on PT theory is designed to adapt to different control modes. Finally, statistical and comparative analyses of the human-like trajectory tracking results of the proposed method were carried out through co-simulation experiments, and it was verified that the human-like performance of the proposed new method had a high degree of similarity with the manipulation behaviors and the vehicle motion state of the real driver. The real-vehicle experiments are carried out, which verifies the consistency of the proposed method with the results of the simulation experiments.INDEX TERMS Trajectory tracking control; human-like driving; autonomous vehicles; ACT-R cognitive framework; preview tracking theory.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.