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.

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.

Decision-making of lane-change for autonomous vehicles faces challenges due to the behavioral differences among human drivers in dynamic traffic environments. To enhance the performances of autonomous vehicles, this paper proposes a game theoretic decision-making method that considers the diverse Social Value Orientations (SVO) of drivers. To begin with, trajectory features are extracted from the NGSIM dataset, followed by the application of Inverse Reinforcement Learning (IRL) to determine the reward preferences exhibited by drivers with distinct Social Value Orientation (SVO) during their decision-making process. Subsequently, a reward function is formulated, considering the factors of safety, efficiency, and comfort. To tackle the challenges associated with interaction, a Stackelberg game model is employed.

In recent years, research on car-like robots has received more attention due to the rapid development of artificial intelligence from diverse disciplines. As essential parts, path planning and lateral path tracking control are the basis for car-like robots to complete automation tasks. Based on the two-degree-of-freedom vehicle dynamic model, this study profoundly analyzes the car-like robots’ path planning and lateral path tracking control. Three objectives: path length, path smoothness, and path safety, are defined and used to construct a multi-objective path planning model. By introducing an adaptive factor, redefining the selection of reference points, and using the cubic spline interpolation for path determination, an improved NGSA-III is proposed, which is mostly adapted in solving the multi-objective path planning problem.

Preventing skidding is essential for studying the safety of driving in curves. However, the adhesion of the vehicle during the driving process on the wet and slippery road will be significantly reduced, resulting in lateral slippage due to the low adhesion coefficient of the road surface, the speed exceeding the turning critical, and the turning radius being too small when passing through the corner. Therefore, to reduce the incidence of traffic accidents of passenger cars driving in curves on rainy and snowy days and achieve the purpose of planning safe driving speed, this paper proposes a curve active safety system based on a deep learning algorithm and vehicle dynamics model. First,we a convolutional neural network (CNN) model is constructed to extract and judge the characteristics of snow and ice adhesion on roads.

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.

Intelligent sweeper vehicle is gradually applied to human life, in which the accuracy of garbage identification and classification can improve cleaning efficiency and save labor cost. Although Deep Learning has made significant progress in computer vision and the application of semantic network segmentation can improve waste identification rate and classification accuracy. Due to the loss of some spatial information during the convolution process, coupled with the lack of specific datasets for garbage identification, the training of the network and the improvement of recognition and classification accuracy are affected. Based on the Unet algorithm, in this paper we adjust the number of input and output channels in the convolutional layer to improve the speed during the feature extraction part. In addition, manually generated datasets are used to greatly improve the robustness of the model.

At present, most of the game decision lane-changing models only consider the state data of the vehicle at the current moment. However, the driving style has a significant impact on the vehicle trajectories, which should be taken into account in the lane-changing process. Moreover, most of the game models are static and do not take into account the sequence of the vehicle lane-changing. This paper proposed a Stackelberg-game-based vehicle lane-changing model considering driving style. Firstly, the NGSIM public dataset is selected for this research and the data screen flow is processed. The K-means algorithm is applied to exchange data clustering. Based on the analysis of vehicle lane changing features under different driving style, the characteristics of the corresponding data under different style are extracted. The quantic-polynomial programming algorithm is used to generate a vehicle lane changing trajectory under different driving styles.

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.

Overloading of trucks can easily cause damage to roads, bridges and other transportation facilities, and accelerate the fatigue loss of the vehicles themselves, and accidents are prone to occur under overload conditions. In recent years, various countries have formulated a series of management methods and governance measures for truck overloading. However, the detection method for overload behavior is not efficient and accurate enough. At present, the method of dynamic load identification is not perfect. No matter whether it is the dynamic weight measurement method of reconstructing the road surface or the non-contact dynamic weight measurement method, little attention is paid to the difference of different vehicles. Especially for different vehicles, there should be different load limits, and the current devices are not smart enough.

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.

The sweeper vehicle plays a very key role in maintaining the urban environment. If the sweeper vehicle can accurately and efficiently identify and classify the ground garbage in the working process, it can greatly improve the working efficiency of the sweeper vehicle and reduce the consumption of manpower. Although the deep learning algorithm based on DUC and PSPNet has high accuracy, the recognition speed is low. ENet is a lightweight network, which greatly improves efficiency, but significantly sacrifices accuracy. This paper presents an improved real-time detection lightweight network based on PSPNet, which takes into account the operation speed and accuracy. The network takes PSPNet as the backbone network, and increases the stride in the convolution process, to reduce the size of the feature map and reduce the amount of calculation.

With the rapid development of new energy vehicles, the echelon utilization of retired power battery has become an important factor to promote the healthy development of this industry, while the Remaining Useful Life (RUL), as the key reference factor for the echelon utilization of retired power battery, has attracted the attention and research of many scholars in recent years. At present, most prediction methods are based on off-line data, which cannot process real-time data in time, so it is difficult to realize online prediction of RUL. In order to realize the real-time online monitoring and high-precision calculation of lithium-ion battery RUL, this paper proposes a lithium-ion battery RUL prediction method based on data-driven and multi-model fusion. The one-dimensional Convolutional Neural Network (1D_CNN) is used for fast online feature extraction of one-dimensional battery capacity time series data to mine potential hidden information.

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.

Lane detection is an important component in automatic pilot system and advanced driving assistance system (ADAS). The stability and precision of lane detection will directly determine precision of control and lane plan of vehicles. Traditional mechanical vision lane detection approaches in complicated environment have the deficiencies of low precision and feature semantic description disabilities. But the lane detection depending on deep learning, e.g. SCNN network, LaneNet network, ENet-SAD network have imbalance problems of splitting precision and storage usage. This paper proposes an approach of high-efficiency deep learning Segnet-LSTM semantic segmentation network. This network structure is composed with encoding network and corresponding decoding networks. First, convolution and maximum pooling. The proposal extracts texture details of five images and stores searching position of maximum pooling. Meanwhile, it will implement interpolate processing to the lost points.

Driving style and skill classification are of great significance in human-oriented advanced driver-assistance system (ADAS) development. In this paper, we propose Fully-Connected Deep Neural Networks (FC-DNN) to classify drivers’ styles and skills with naturalistic driving data. Followed by the data collection and pre-processing, FC-DNN with a series of deep learning optimization algorithms are applied. In the experimental part, the proposed model is validated and compared with other commonly used supervised learning methods including the k-nearest neighbors (KNN), support vector machine (SVM), decision tree (DT), random forest (RF), and multilayer perceptron (MLP). The results show that the proposed model has a higher Macro F1 score than other methods. In addition, we discussed the effect of different time window sizes on experimental results. The results show that the driving information of 1s can improve the final evaluation score of the model.

To reduce the interference and conflict of human-machine cooperative control, lighten the operation workload of drivers, and improve the friendliness and acceptability of intelligent vehicles, a personalized human-machine cooperative lane-change trajectory tracking control method was proposed. First, a lane-changing driving data acquisition test was carried out to collect different driving behaviors of different drivers and form the data pool for the machine learning method. Two typical driving behaviors from an aggressive driver and a moderate driver are selected to be studied. Then, a control structure combined by feedforward and feedback control based on Long Short Term Memory (LSTM) and model-based optimum control was introduced. LSTM is a machine learning method that has the ability of memory. It is used to capture the lane-changing behaviors of each driver to achieve personalization. For each driver, a specific personalized controller is trained using his driving data.

Driver distraction detection is crucial to driving safety when autonomous vehicles are co-piloted. Recognizing drivers’ behaviors that are highly related with distraction from real-time video stream is widely acknowledged as an effective approach mainly due to its non-intrusiveness. In recently years, deep learning neural networks have been adopted to by-pass the procedure of designing features artificially, which used to be the major downside of computer-vision based approaches. However, the detection accuracy and generalization ability is still not satisfying since most deep learning models extracts only spatial information contained in images. This research develops a driver distraction model based on a two-stream, spatial and temporal, convolutional neural network (CNN).

Artificial intelligence systems are highly accepted as a technology to offer an alternative way to tackle complex and non-linear problems. They can learn from data, and they are able to handle noisy and incomplete data. Once trained, they can perform prediction and generalization at high speed. The aim of the present study is to propose a novel approach utilizing the adaptive neuro-fuzzy inference system (ANFIS) and the fuzzy clustering method for automotive ride performance estimation. This study investigated the relationship between the automotive ride performance and relative parameters including speed, spring stiffness, damper coefficients, ratios of sprung and unsprung mass. A Takagi-Sugeno fuzzy inference system associated with artificial neuro network was employed. The C-mean fuzzy clustering method was used for grouping the data and identifying membership functions.

In this research, the interior noise of a passenger car was measured, and the sound quality metrics including sound pressure level, loudness, sharpness, and roughness were calculated. An artificial neural network was designed to successfully apply on automotive interior noise as well as numerous different fields of technology which aim to overcome difficulties of experimentations and save cost, time and workforce. Sound pressure level, loudness, sharpness, and roughness were estimated by using the artificial neural network designed by using the experiment values. The predicted values and experiment results are compared. The comparison results show that the realized artificial intelligence model is an appropriate model to estimate the sound quality of the automotive interior noise. The reliability value is calculated as 0.9995 by using statistical analysis.