Search Results

Roundabouts are one of the most complex traffic environments in urban roads, and a key challenge for intelligent driving decision-making. Deep reinforcement learning, as an emerging solution for intelligent driving decisions, has the advantage of avoiding complex algorithm design and sustainable iteration. For the decision difficulty in roundabout scenarios, this paper proposes an artificial potential field based Soft Actor-Critic (APF-SAC) algorithm. Firstly, based on the Carla simulator and Gym framework, a reinforcement learning simulation system for roundabout driving is built. Secondly, to reduce reinforcement learning exploration difficulty, global path planning and path smoothing algorithms are designed to generate and optimize the path to guide the agent.

Highway safety remains a significant concern, especially in mixed traffic scenarios involving heavy-duty vehicles (HDV) and smaller passenger cars. The vulnerability of HDVs following closely behind smaller cars is evident in incidents involving the lead vehicle, potentially leading to catastrophic rear-end collisions. This paper explores how automatic speed enforcement systems, using speed cameras, can mitigate risks for HDVs in such critical situations. While historical crash data consistently demonstrates the reduction of accidents near speed cameras, this paper goes beyond the conventional notion of crash occurrence reduction. Instead, it investigates the profound impact of driver behavior changes within desired travel speed distribution, especially around speed cameras, and their contribution to the safety of trailing vehicles, with a specific focus on heavy-duty trucks in accident-prone scenarios.

Autonomous driving technology is more and more important nowadays, it has been changing the living style of our society. As for autonomous driving planning and control, vehicle dynamics has strong nonlinearity and uncertainty, so vehicle dynamics and control is one of the most challenging parts. At present, many kinds of specific vehicle dynamics models have been proposed, this review attempts to give an overview of the state of the art of vehicle dynamics models for autonomous driving. Firstly, this review starts from the simple geometric model, vehicle kinematics model, dynamic bicycle model, double-track vehicle model and multi degree of freedom (DOF) dynamics model, and discusses the specific use of these classical models for autonomous driving state estimation, trajectory prediction, motion planning, motion control and so on.

Driver steering feature clustering aims to understand driver behavior and the decision-making process through the analysis of driver steering data. It seeks to comprehend various steering characteristics exhibited by drivers, providing valuable insights into road safety, driver assistance systems, and traffic management. The primary objective of this study is to thoroughly explore the practical applications of various clustering algorithms in processing driver steering data and to compare their performance and applicability. In this paper, principal component analysis was employed to reduce the dimension of the selected steering feature parameters. Subsequently, K-means, fuzzy C-means, the density-based spatial clustering algorithm, and other algorithms were used for clustering analysis, and finally, the Calinski-Harabasz index was employed to evaluate the clustering results. Furthermore, the driver steering features were categorized into lateral and longitudinal categories.

Currently, the rapid expansion of the global road transport industry and the imperative to reduce carbon emissions are propelling the advancement of electrified highways (EH). In order to conduct a comprehensive economic analysis of EH, it is crucial to develop a detailed /8.and comprehensive economic model that takes into account various transportation modes and factors that influence the economy. However, the existing economic models for EH lack comprehensiveness in terms of considering different transportation modes and economic factors. This study aims to fill this gap by designing an economic model for an EH-based Online DC-driven system (ODS) for long distance heavy-duty transport vehicle incorporating multi-factor sensitivities. Firstly, the performance parameters of the key components of the system are calculated using vehicle dynamics equations which involves selecting and matching the relevant components and determining the fundamental cost of vehicle transformation.

Autonomous driving in real-world urban traffic must cope with dynamic environments. This presents a challenging decision-making problem, e.g. deciding when to perform an overtaking maneuver or how to safely merge into traffic. The traditional autonomous driving algorithm framework decouples prediction and decision-making, which means that the decision-making and planning tasks will be carried out after the prediction task is over. The disadvantage of this approach is that it does not consider the possible impact of ego vehicle decisions on the future states of other agents. In this article, a decision-making and planning method which considers longitudinal interaction is represented. The method’s architecture is mainly composed of the following parts: trajectory sampling, forward simulation, trajectory scoring and trajectory selection. For trajectory sampling, a lattice planner is used to sample three-dimensionally in both the time horizon and the space horizon.

Automated driving has become a very promising research direction with many successful deployments and the potential to reduce car accidents caused by human error. Automated driving requires automated path planning and tracking with the ability to avoid collisions as its fundamental requirement. Thus, plenty of research has been performed to achieve safe and time efficient path planning and to develop reliable collision avoidance algorithms. This paper uses a data-driven approach to solve the abovementioned fundamental requirement. Consequently, the aim of this paper is to develop Deep Reinforcement Learning (DRL) training pipelines which train end-to-end automated driving agents by utilizing raw sensor data. The raw sensor data is obtained from the Carla autonomous vehicle simulation environment here. The proposed automated driving agent learns how to follow a pre-defined path with reasonable speed automatically.

With the rapid development of intelligent driving technology, there has been a growing interest in the driving comfort of automated vehicles. As vehicles become more automated, the role of the driver shifts from actively engaging in driving tasks to that of a passenger. Consequently, the study of the passenger experience in automated driving vehicles has emerged as a significant research area. In order to examine the impact of automatic driving on passengers' riding experience in vehicle platooning scenarios, this study conducted real vehicle experiments involving six participants. The study assessed the subjective perception scores, eye movement, and electrocardiogram (ECG) signals of passengers seated in the front passenger seat under various vehicle speeds, distances, and driving modes. The results of the statistical analysis indicate that vehicle speed has the most substantial influence on passenger perception.

Under complex and extreme operating conditions, the road adhesion coefficient emerges as a critical state parameter for tire force analysis and vehicle dynamics control. In contrast to model-based estimation methods, intelligent tire technology enables the real-time feedback of tire-road interaction information to the vehicle control system. This paper proposes an approach that integrates intelligent tire systems with machine learning to acquire precise road adhesion coefficients for vehicles. Firstly, taking into account the driving conditions, sensor selection is conducted to develop an intelligent tire hardware acquisition system based on MEMS (Micro-Electro-Mechanical Systems) three-axis acceleration sensors, utilizing a simplified hardware structure and wireless transmission mode. Secondly, through the collection of real vehicle experiment data on different road surfaces, a dataset is gathered for machine learning training.

Accurate sensing of road conditions is one of the necessary technologies for safe driving of intelligent vehicles. Compared with the structured road, the unstructured road has complex road conditions, and the response characteristics of vehicles under different road conditions are also different. Therefore, accurately identifying the road categories in front of the vehicle in advance can effectively help the intelligent vehicle timely adjust relevant control strategies for different road conditions and improve the driving comfort and safety of the vehicle. However, traditional road identification methods based on vehicle kinematics or dynamics are difficult to accurately identify the road conditions ahead of the vehicle in advance. Therefore, this paper proposes an unstructured road region detection and road classification algorithm based on machine vision to obtain the road conditions ahead.

With the rapid development of science and technology, the automobile industry is developing rapidly, and intelligent networking and autonomous driving have become new research hotspots. The safety and efficiency of vehicle driving has always been an important research topic in the transportation field. Due to reducing the participation of drivers, autonomous vehicles can reduce traffic accidents caused by human factors. While the development of intelligent networking can achieve information sharing between vehicles, and improve driving efficiency to a certain extent. Based on the game theory and the minimum safe distance condition, this paper establishes a lane changing decision model of intelligent network-connected autonomous vehicles, puts forward a game payoff function and analyzes the game strategy.

Aiming at the problems of long lane change intention recognition, complicated lane change model, and huge amount of processing data in the current research, this paper uses principal component analysis to improve the driver’s lane change intention recognition model using traditional pattern recognition. Firstly collect 7 parameters including driver operation and vehicle running characteristics. After data standardization and PCA (principal component analysis), the top three principal components that can reflect the information content of the original data are nearly 90%. Then, a lane-change intent recognition model based on GMM-HMM was established, three lane change intents cannot be directly observed as the hidden state of the model; and three principal component quantities obtained through linear changes are used as observational measurements.

Autonomous driving technology, as the product of the fifth stage of the information technology revolution, is of great significance for improving urban traffic and environmentally friendly sustainable development. Autonomous driving can be divided into three main modules. The input of the decision module is the perception information from the perception module and the output of the control strategy to the control module. The deep reinforcement learning method proposes an end-to-end decision-making system design scheme. This paper adopts the Deep Deterministic Policy Gradient Algorithm (DDPG) that incorporates the Priority Experience Playback (PER) method. The framework of the algorithm is based on the actor-critic network structure model. The model takes the continuously acquired perception information as input and the continuous control of the vehicle as output.

With the increasingly complex traffic environment, the vehicle AEB system needs to go through a large number of testing processes, in order to drive more safely on the road. For speeding up the development process of AEB and solve the problems of long cycle, high cost and low efficiency in AEB testing, in this paper, a millimeter wave radar turntable is built, and a high-precision detection algorithm of turntable encoder is designed, at the same time, a test method of vehicle AEB based on the detection data of radar turntable encoder is designed. The verification results show that methods described in this paper can be used to develop the vehicle AEB test algorithm efficiently.

To quickly and efficiently match the planetary gearset characteristic parameter of power-spilt hybrid vehicles so that their oil-saving potential can be maximized, this study proposes a parameter matching method that comprehensively considers energy management strategy and driving cycle based on an analysis of vehicle instantaneous efficiency. The method is used to match the planetary characteristic parameter of a power-split hybrid light truck. The relevant conclusions are compared with the influence of various planetary characteristic parameters on fuel consumption obtained through simulation under typical operating conditions. The simulation results show that the influence laws of the various planetary characteristic parameters on vehicle average efficiency are similar to those on fuel consumption. The proposed parameter-matching method based on vehicle efficiency analysis can effectively match the planetary characteristic parameter for power-split hybrid powertrains.

With the development of cellular communication technology and for the sake of reducing drag resistance, the multi-lane platoon technology will be more prosperous in the future. In this article, the cooperative vehicle platoon method on the public road is represented. The method’s architecture is mainly composed of the following parts: decision-making, path planning and control command generation. The decision-making uses the finite state machine to make decision and judgment on the cooperative lane change of vehicles, and starts to execute the lane change step when the lane change requirements are met. In terms of path planning, with the goal of ensuring comfort, the continuity of the vehicle state and no collision between vehicles, a fifth-order polynomial is used to fit every vehicle trajectory. In terms of control command generation module, a model predictive control algorithm is used to solve the multi-vehicle centralized optimization control problem.

With the advent of increasing autonomous vehicles on public roads, the safety of vulnerable road users such as pedestrians, cyclists, etc. has never been more important. These especially include Blind or Visually Impaired (BVI) pedestrians who face difficulty in making confident decisions in road crossings without the help of accessible pedestrian signals (APS). This paper addresses some of the safety measures that can be taken to improve and assess the safety of BVI pedestrians in a controlled environment like a BVI school campus where autonomous vehicles are operated. The majority of research on autonomous vehicle safety does not consider the edge cases of encounters with BVI pedestrians. Based on this motivation, requirements and characteristics of Non-BVI and BVI pedestrians have been stated along with the motion models used to predict their future movements. Existing tools based on Bayesian multi-model filters were used for pedestrian tracking and motion predictions.

Roundabout is a unique approach of managing traffic at intersections because it relies on driver’s instincts of safety. Roundabouts are considered safer than other ways of intersection traffic management due to low speed limits, smoother merging, and reduced fatal accidents. Despite their benefits and increasing usage, there is lack of clear understanding of the roundabouts, particularly due to scarcity of data and simulation models and the complexity of the structure. Real-time and offline traffic data recorded at a roundabout provides a basis for 1) identification of the safety issues, 2) understanding unexpected and risky driver behavior, 3) proposing potential mobility solutions, and 4) developing simulation models. The processed data may be used in controlling metered roundabouts, communicating with connected and automated vehicles (CAVs) etc. In this paper an approach to obtain useful traffic information from video feed data at a roundabout is presented.

This study is focused on exploring the possibilities of using camera and route planner images for autonomous driving in an end-to-mid learning fashion. The overall idea is to clone the humans’ driving behavior, in particular, their use of vision for ‘driving’ and map for ‘navigating’. The notion is that we humans use our vision to ‘drive’ and sometimes, we also use a map such as Google/Apple maps to find direction in order to ‘navigate’. We replicated this notion by using end-to-mid imitation learning. In particular, we imitated human driving behavior by using camera and route planner images for predicting the desired waypoints and by using a dedicated control to follow those predicted waypoints. Besides, this work also places emphasis on using minimal and cheaper sensors such as camera and basic map for autonomous driving rather than expensive sensors such Lidar or HD Maps as we humans do not use such sophisticated sensors for driving.

It’s hard to achieve complete self-driving using hand-crafting generalized decision-making rules, while the end-to-end self-driving system is low in complexity, does not require hand-crafting rules, and can deal with complex situations. Modular-based self-driving systems require multi-task fusion and high-precision maps, resulting in high system complexity and increased costs. In end-to-end self-driving, we usually only use camera to obtain scene status information, so image processing is very important. Numerous deep learning applications benefit from multi-task learning, as the multi-task learning can accelerate model training and improve accuracy with combine all tasks into one model, which reduces the amount of calculation and allows these systems to run in real-time. Therefore, the approach of obtaining rich scene state information based on multi-task learning is very attractive. In this paper, we propose an approach to multi-task learning for semantics, geometry and motion.