Search

Infrared and visible image fusion is an effective solution for image quality enhancement. However, conventional fusion models require the decomposition of source images into image blocks, which disrupts the original structure of the images, leading to the loss of detail in the fused images and making the fusion results highly sensitive to matching errors. This paper employs Convolutional Sparse Representation (CSR) to perform global feature transformation on the source images, overcoming the drawbacks of traditional fusion models that rely on image decomposition. Inspired by neural networks, a multi-layer CSR model is proposed, which involves five layers in a forward-feeding manner: two CSR layers acquiring sparse coefficient maps, one fusion layer combining sparse maps, and two reconstruction layers for image recovery. The dataset used in this paper comprises infrared and visible images selected from public dataset, as well as registered images collected by an actual Unmanned Aerial Vehicle (UAV). The source images contain ground targets, marine targets, and natural landscapes. To validate the effectiveness of the proposed image fusion model in this paper, comparative analysis is conducted with state-of-the-art (SOTA) algorithms. Experimental results demonstrate that the proposed fusion model outperforms other state-of-the-art methods by at least 10% in SF, EN, MI and QAB/F fusion metrics in most image fusion cases, thereby affirming its favorable performance.

The automotive headlight stands out as a critical vehicle component, particularly emphasized during nighttime driving. The high beam, designed for optimal driver visibility on long-distance roads, traditionally relies on manual control by the driver. However, this manual control poses challenges, particularly when the high beam light temporarily blinds oncoming drivers. The resultant dazzle for drivers of opposing vehicles is a significant concern. In response to these issues, there is a growing demand for adaptive and intelligent headlights that can autonomously adjust beam intensity. The intelligent headlight system takes on the responsibility of modifying the beam intensities without requiring explicit input from the drivers. This study aims to systematically review various approaches to controlling intelligent headlight beam intensity. The paper identifies four prominent approaches to intelligent headlight beam intensity control, recognized as widely used techniques. Furthermore, the study uncovers intriguing connections between some of these intensity control approaches. A survey on utilization rates indicates that sensor-based and machine learning (ML)-based intensity control approaches are the most commonly employed methods by automotive headlight designers. The paper concludes by providing insights into the future prospects of intelligent headlight technology, offering guidance for future researchers in this field.

Abstract Controlling an autonomous vehicle's unprotected left turn at an intersection is a challenging task. Traditional rule‐based autonomous driving decision and control algorithms struggle to construct accurate and trustworthy mathematical models for such circumstances, owing to their considerable uncertainty and unpredictability. To overcome this problem, a rule‐constrained reinforcement learning (RCRL) control method is proposed in this work for autonomous driving. To train a reinforcement learning controller with rule constraints, outcomes of the path planning module are used as a goal condition in the reinforcement learning framework. Since they include vehicle dynamics, the proposed approach is safer and more reliable compared to end‐to‐end learning, thereby ensuring that the generated trajectories are locally optimal while adjusting to unpredictable situations. In the experiments, a highly randomized two‐way four‐lane intersection is established based on the CARLA simulator to verify the effectiveness of the proposed RCRL control method. Accordingly, the results show that the proposed method can provide real‐time safe planning and ensure high passing efficiency for autonomous vehicles in the unprotected left turn task.

Abstract In this paper, the authors investigate the issue of constructing and incorporating an accurate vehicle dynamic model for model predictive control (MPC) with an application to autonomous vehicle path following. To achieve the desired performance, MPC employs a precise dynamic model. However, the uncertainty of complex systems and their operating environments presents a challenge to the development of an adequately accurate vehicle dynamic model. This paper proposes a Deep Gaussian Process Regression (DGPR) method to improve model precision. Meanwhile, the learning model is incorporated into a novel MPC framework to enhance closed‐loop performance. High‐fidelity simulations using CarSim‐MATLAB demonstrate the validity of the proposed approach in terms of enhancing the path following performance and lateral stability under the condition of large curvature at medium to high speeds on roads with different friction coefficients when compared to the nominal MPC approach.

Abstract Cooperative adaptive cruise control (CACC) realizes efficient, intelligent control of vehicle acceleration, deceleration, and steering, through inter‐vehicle communication and cooperative control. However, the close combination of the platoon makes it difficult for other vehicles to cut‐in, which can lead to severe traffic jams on certain sections of the road. The control effect of the CACC depends on the platoon penetration rate, which is the percentage of connected and autonomous vehicles (CAVs) in the total number of platoon members. There is no quantitative control method for different penetration rates, and it is difficult to quantify the impact of CACC vehicles on traffic. Therefore, this paper proposes an innovative CACC control method based on deep reinforcement learning (DRL). First, the altruism control and the quantitative control of the car‐following strategy are realized by the virtual car‐following distance method to reduce the exclusivity of the CACC platoon or improve the road utilization efficiency. Second, a more appropriate platoon reward function and collision avoidance method are proposed. Finally, the Car Learning to Act (CARLA) simulator is used. The obtained results confirm that the CACC control of CAVs based on DRL can absorb speed oscillation and improve fuel economy.

Abstract The vehicle detection algorithm based on visual perception has been applied in all types of automatic driving scenes. However, there are still flaws in the current detection algorithm model, especially for small objects. The detection effect of vehicle objects with small pixels in the image is often missed and wrongly detected. This research proposes an improved feature fusion module based on double‐channel residual pyramid (DRP) structure for autonomous detection algorithm which named binary residual feature pyramid network (BiResFPN) to solve the above problems. Firstly, a DRP structure, which can effectively supplement the shallow information of the network, is proposed. The residual structure is added to the output feature map for further supplement. Then, an average sampling method of positive and negative samples based on intersection‐over‐union (IOU) value is proposed on the basis of this structure, aimed at the unbalanced sampling of positive and negative samples in the training stage of faster regions with CNN features (RCNN). It leads to the reduction of the interference of a large number of simple negative samples, which makes the learned model better. The experimental results based on the KITTI and BDD100K dataset datasets show that the capability of the feature fusion module based on DRP structure is strong for small object detection. Compared with Faster‐RCNN (FPN), the detection algorithm of small object detection accuracy APsmall was increased by 2.6%, APmedium and APlarge was increased by 1.1% and 0.3%.

Connected automated vehicles (CAV) can increase traffic efficiency, which is considered a critical factor in saving energy and reducing emissions in traffic congestion. In this paper, systematic traffic simulations are conducted for three car-following modes, including intelligent driver model (IDM), adaptive cruise control (ACC), and cooperative ACC (CACC), in congestions caused by rear-end collisions. From the perspectives of lane density, vehicle trajectory and vehicle speed, the fuel consumption of vehicles under the three car-following modes are compared and analysed, respectively. Based on the vehicle driving and accident environment parameters, an XGBoost algorithm-based fuel consumption prediction framework is proposed for traffic congestions caused by rear-end collisions. The results show that compared with IDM and ACC modes, the vehicles in CACC car-following mode have the ideal performance in terms of total fuel consumption; besides, the traffic flow in CACC mode is more stable, and the speed fluctuation is relatively tiny in different accident impact regions, which meets the driving desires of drivers.

Abstract Connected and Autonomous Vehicles (CAVs) can receive various information from surrounding vehicles through Vehicle‐to‐Everything (V2X) communication technologies and adjust their car‐following behaviour accordingly. Although several studies have evaluated the impact of CAVs on traffic flow stability in a small segment of networks, most approaches are focused on their specific applications considering the trajectory information, and there is a lack of studies analyzing the impact of CAVs on a large‐scale network. This paper proposes a novel viscous continuum traffic model considering the anticipation of space headway, the throttle angle, and brake torque information during cooperative car‐following. The methods employed to develop the new car‐following model and its counterpart continuum traffic model have been described. The linear and non‐linear stability analyses of the newly developed model have been conducted to obtain the critical stability factors in small perturbations. Numerical simulations have been carried out to investigate the effect of the anticipation, the throttle angle, and brake torque information on traffic stability, fuel consumption, and exhaust emissions. The numerical results reveal that the anticipation of space headway and the transmission of the throttle angle and brake torque information during cooperative car‐following manoeuvres can improve the traffic flow stability and reduce fuel consumption and emissions.

Although significant progress has been made in developing autonomous vehicle technology, further research is still needed to reduce the risk of collision by planning a suitable path. Drawing vehicle paths through pedestrian collision avoidance action is worth further study. In this paper, vehicle path planning considering pedestrian collision avoidance action is proposed to improve vehicle driving safety further. First, the movement features reflecting different types of pedestrian collision avoidance intention are summarized through the actual human-vehicle collision accident video. Then, the data representing the pedestrian action features are obtained through the deep neural network and input into the trained Hidden Markov Model to predict pedestrian intention. Finally, the vehicle collision avoidance path is planned through the cubic polynomial combined with the pedestrian intention. The algorithm is verified through the pedestrian collision avoidance action in the simulation scene. The results show that this algorithm can predict the intention of different types of pedestrians to avoid collisions at frames 9, 8, and 10, respectively, which is 0.6 seconds, 0.4 seconds, and 0.6 seconds faster than the Long Short Term Memory algorithm. Finally, a suitable path is planned, and the collision was successfully avoided.

This paper aims to establish a driving style recognition method that is highly accurate, fast and generalizable, considering the lack of data types in driving style classification task and the low recognition accuracy of widely used unsupervised clustering algorithms and single convolutional neural network methods. First, we propose a method to collect the information on driver’s operation time sequence in view of the imperfect driving data, and then extract the driver’s style features through convolutional neural network. Then, for the collected temporal data, the Long Short Term Memory networks (LSTM) module is added to encode and transform the driving features, to achieve the driving style classification. The results show that the accuracy of driving style recognition reaches over 93%, while the speed is improved significantly.

Abstract Trajectory prediction of surrounding vehicles is a crucial capability of intelligent driving vehicles. In a scene, a vehicle and its surrounding vehicles constitute an integral system, and the vehicle's future motion trajectory is affected by the actions of surrounding vehicles. The influencing mode and degree are hidden in the relevant historical information of the vehicle and its neighbour vehicle. The existing trajectory prediction methods either do not consider the confidence of the predicted trajectory, or the accuracy requirement is ignored when considering the confidence of the predicted trajectory. In order to address this problem, a mixed Conditional AutoEncoder Generative Adversarial Network (CAE‐GAN) model based on the multi‐loss function is proposed. The proposed model uses the encoder–decoder structure with a convolutional social pool to extract general features. The generative adversarial networks (GANs) are used to extract the confidence features of the generated trajectories, which enables the proposed model to generate trajectories that are close to the real trajectories. In addition, a classifier structure based on an LSTM network is added to output the probability that the predicted trajectory belongs to a particular lateral maneuver so that the generated trajectory lateral maneuver of the model is consistent with the real trajectory lateral maneuver. The proposed model is evaluated using the publicly available NGSIM US‐101 and I‐80 datasets, and results show that the accuracy of the proposed model is higher than that of the existing methods. The proposed model achieves an average accuracy improvement of 16.34% in comparison to the most advanced existing models.

Abstract The combination of reinforcement learning and platooning control has been widely studied, which has been considered to be altruistic on the basis of safety. However, the platooning control method based on reinforcement learning is not mature. This paper proposes a platoon sharing deep deterministic policy gradient algorithm (PSDDPG) for a multi‐vehicle network which overcomes the problem of low efficiency of continuous action space exploration. On the basis of the noise of deep deterministic policy gradient (DDPG) algorithm, additional platoon noise is added to enhance the diversity of training samples during exploration and avoid using a single‐vehicle data, which could weaken model robustness. The method of using the time sequence information as an input and replay buffer backup is proposed to prevent the problems of insufficient exploration and low sampling efficiency from deteriorating the training effect. The robustness of the proposed algorithm is verified by tests using the Carla simulator. Also, the platoon merging, overtaking, cruising following and obstacle avoidance control of a vehicle platoon is realized under typical traffic flow conditions. The experimental results show that the PSDDPG algorithm can provide an efficient control strategy and thus has the potential to reduce energy consumption and improve road efficiency.

Abstract Autonomous vehicles need to have the ability to predict the future behaviour of surrounding vehicles, which helps with proper trajectory planning and tracking. Many behaviour prediction methods have limited application because they have a very limited prediction horizon. This paper proposes a deep learning‐based hybrid model for behaviour prediction over long‐time periods, including manoeuvre recognition and a behaviour prediction module. In manoeuvre recognition module, the CNN extracts the social characteristics of the target vehicle and LSTM outputs the manoeuvre probability vector to form a contextual feature vector with the social features. In the behaviour prediction module, LSTM and Attention are based on the contextual feature vector to capture multi‐time step information in the behaviour time window to complete the prediction of the target vehicle behaviour. Real‐car collection and open‐source vehicle trajectory datasets were used for training and testing. The results show that the proposed algorithm could predict vehicle behaviour with an accuracy of 89.73% and an average prediction time of 2.032 s, which has a high engineering application value.

Particularly crucial throughout the mode transition procedure is the transmission properties of hydro–mechanical composite transmission devices. This paper describes a multipurpose power transmission device that integrates hydrostatic, hydro-mechanical, and mechanical transmission and mainly discusses the transmission characteristic optimization problem from the perspective of speed regulation characteristics, shift strategy, and efficiency characteristics. The kinematic and dynamic analysis of the transmission system, the assembly scheme, and relevant parameters of the power transmission device are analyzed, and the speed regulation characteristic curve is obtained. The shift strategy of power transmission devices involving clutches and brakes during the whole speed regulation process and the best switch time of each component are found. The efficiency expression of the static pressure system is obtained from the efficiency model of the pump-control-motor system, and the efficiency of the multi-purpose power transmission device is obtained using the efficiency definition method. The fitting curves of hydrostatic system efficiency are determined using experimental data, and the efficiency of the hydro–mechanical composite power transmission system is obtained using the conversion mechanism method. The results show that the shift quality of power transmission devices can be improved greatly by controlling the switch sequence of clutches and brakes reasonably.

This paper proposes a novel vehicle state estimation (VSE) method that combines a physics-informed neural network (PINN) and an unscented Kalman filter on manifolds (UKF-M). This VSE aimed to achieve inertial measurement unit (IMU) calibration and provide comprehensive information on the vehicle’s dynamic state. The proposed method leverages a PINN to eliminate IMU drift by constraining the loss function with ordinary differential equations (ODEs). Then, the UKF-M is used to estimate the 3D attitude, velocity, and position of the vehicle more accurately using a six-degrees-of-freedom vehicle model. Experimental results demonstrate that the proposed PINN method can learn from multiple sensors and reduce the impact of sensor biases by constraining the ODEs without affecting the sensor characteristics. Compared to the UKF-M algorithm alone, our VSE can better estimate vehicle states. The proposed method has the potential to automatically reduce the impact of sensor drift during vehicle operation, making it more suitable for real-world applications.

Abstract Radar and LiDAR are two environmental sensors commonly used in autonomous vehicles, Lidars are accurate in determining objects’ positions but significantly less accurate as Radars on measuring their velocities. However, Radars relative to Lidars are more accurate on measuring objects velocities but less accurate on determining their positions as they have a lower spatial resolution. In order to compensate for the low detection accuracy, incomplete target attributes and poor environmental adaptability of single sensors such as Radar and LiDAR, in this paper, an effective method for high-precision detection and tracking of surrounding targets of autonomous vehicles. By employing the Unscented Kalman Filter, Radar and LiDAR information is effectively fused to achieve high-precision detection of the position and speed information of targets around the autonomous vehicle. Finally, the real vehicle test under various driving environment scenarios is carried out. The experimental results show that the proposed sensor fusion method can effectively detect and track the vehicle peripheral targets with high accuracy. Compared with a single sensor, it has obvious advantages and can improve the intelligence level of autonomous cars.

Manure treatment with black soldier fly larvae (BSFL) and BSFL frass application in crop land is a sustainable strategy; however, whether residual antibiotic resistance genes (ARGs) and their transmission risk are related to the manure BSFL treatment process is still unknown. In this paper, the effect of BSFL addition density on residual tetracycline resistance genes (TRGs) and transmission from frass to pakchoi was determined. The results showed that BSFL frass can provide sufficient nutrients for growth, improve the economic value of pakchoi, and reduce the risk of transmission of TRGs in chicken manure regardless of BSFL density. The potential hosts of the TRGs we detected were found in BSFL frass (Oblitimonas and Tissierella), rhizosphere soil (Mortierella and Fermentimonas), and pakchoi endophytes (Roseomonas). The present study concluded that BSFL frass produced by adding 100 BSFL per 100 g of chicken manure has the advantages of high value and low risk. These findings will provide important strategic guidance for animal manure disposal and theoretical support for preventing the transmission of TRGs in BSFL applications.

Abstract Due to the bus characteristics of large quality, high center of gravity and narrow wheelbase, the research of its yaw stability control (YSC) system has become the focus in the field of vehicle system dynamics. However, the tire nonlinear mechanical properties and the effectiveness of the YSC control system are not considered carefully in the current research. In this paper, a novel adaptive nonsingular fast terminal sliding mode (ANFTSM) control scheme for YSC is proposed to improve the bus curve driving stability and safety on slippery roads. Firstly, the STI (Systems Technologies Inc.) tire model, which can effectively reflect the nonlinear coupling relationship between the tire longitudinal force and lateral force, is established based on experimental data and firstly adopted in the bus YSC system design. On this basis, a more accurate bus lateral dynamics model is built and a novel YSC strategy based on ANFTSM, which has the merits of fast transient response, finite time convergence and high robustness against uncertainties and external disturbances, is designed. Thirdly, to solve the optimal allocation problem of the tire forces, whose objective is to achieve the desired direct yaw moment through the effective distribution of the brake force of each tire, the robust least-squares allocation method is adopted. To verify the feasibility, effectiveness and practicality of the proposed bus YSC approach, the TruckSim-Simulink co-simulation results are finally provided. The co-simulation results show that the lateral stability of bus under special driving conditions has been significantly improved. This research proposes a more effective design method for bus YSC system based on a more accurate tire model.

This paper introduces a mechanic-electronic-hydraulic powertrain system (MEH-PS) composed of an electro-mechanical hybrid system and hydro-mechanical composite transmission according to current mainstream drive and transmission technologies. Firstly, the structural design concept of the system is introduced, and the power and transmission components are selected according to the actual working requirements of the tractor. Then, the principles of various drive modes and transmission modes of the powertrain system are explained, and the speed regulation characteristic curves of the hydro-mechanical transmission (HMT) are given; the speed characteristics, torque characteristics, power split characteristics, and efficiency characteristics of the powertrain system are analyzed. Finally, a tractor simulation and test model was developed to verify its performance under certain operating conditions. Simulation results show that: the tractor acceleration performance is improved, the speed range is wider, the power components and hydraulic components can also meet the requirements, the HMT in a wide range of speed to maintain the average efficiency of above 86%. The bench test results show that: the step-less speed regulation characteristics and efficiency characteristics of the powertrain system are basically consistent with the simulation results.

Following man-made rules in the traditional control method of autonomous driving causes limitations for intelligent vehicles under various traffic conditions that need to be overcome by incorporating machine learning-based method. The latter is inherently suitable for simple tasks of autonomous driving according to its limited characteristic under complex multi-lane traffic conditions. In this paper, a decision control method is proposed based on multi-task reinforcement learning to address the shortcomings of autonomous driving control under complex traffic conditions. Herein, the autonomous driving task is divided into several subtasks utilizing the proposed method to reduce the training time and improve traffic efficiency under complex multi-lane traffic condition. To ensure the efficiency and robustness of agent convergence to the optimal action space, an adaptive noise exploration method is designed for the subtasks with convex characteristics. Five-lane driving tasks scenarios embedded in Carla simulator have been conducted to verify the proposed method. The results of the simulation draw the conclusion that the proposed method increases the driving efficiency of intelligent vehicles under complex traffic conditions.

This paper provides a multi-agent coordinated control system to improve the real-time performance of intelligent vehicle active collision avoidance. At first, the functions and characteristics of longitudinal and lateral collision avoidance agents are analyzed, which are the main components of the multi-agent. Then, a coordinated solution mechanism of an intelligent vehicle collision avoidance system is established based on hierarchical control and blackboard model methods to provide a reasonable way to avoid collision in complex situations. The multi-agent coordinated control system can handle the conflict between the decisions of different agents according to the rules. Comparing with existing control strategies, the proposed system can realize multi decisions and planning at the same time; thus, it will reduce the operation time lag during active collision avoidance. Additionally, fuzzy sliding mode control theory is introduced to guarantee accurate path tracking in lateral collision avoidance. Finally, co-simulation of Carsim and Simulink are taken, and the results show that the real-time behavior of intelligent vehicle collision avoidance can be improved by 25% through the system proposed.

In this study, a novel traffic flow modeling framework is proposed considering the impact of driving system and vehicle mechanical behavior as two different units on the traffic flow. To precisely model the behavior of Connected and Autonomous (CA) vehicles, three submodels are proposed as a novel microscopic traffic flow framework, named Integrated-Hybrid (IH) model. Focusing on the realization of the car following behavior of CA vehicles, the driving system (vehicle control system) and the vehicle mechanical system are modeled separately and linked by throttle and brake actuators model. The IH model constitutes the key part of the Full Velocity Difference (FVD) model considering the mechanical capability of vehicles and dynamic collision avoidance strategies to ensure the safety of following distance between two consecutive vehicles. Linear stability conditions are derived for each model and developing methodology for each submodel is discussed. Our simulations revealed that the IH model successfully generates velocity and acceleration profiles during car following maneuvers and throttle angle/brake information in connected vehicles environment can effectively improve traffic flow stability. The vehicles’ departure and arrival process while passing through a signal-lane with a traffic light considering the anticipation driving behavior and throttle angle/brake information of direct leading vehicle was explored. Our numerical results demonstrated that the IH model can capture the velocity fluctuations, delay times, and kinematic waves efficiently in traffic flow.

This paper presents a novel algorithm to deal with dynamic multiobjective optimization problems, in which the objective functions change over time. The algorithm adopts the decomposition framework to decompose the multiobjective optimization problems into a number of scalar optimization subproblems. For each subproblem, its respective solutions obtained in several former consecutive environments can form a moving trajectory over time. A shift vector guided prediction model is proposed, which samples three intermediately previous solutions of each subproblem to construct two shift vectors. The shift vectors use the weighted summation to generate a new shift vector as the forthcoming motion of the target solution. Then a new location in the later environment is estimated based on the current location and the newly generated shift vector. When detecting an environmental change, the multiobjective evolutionary algorithm based on decomposition will update the population using the predicted solutions by the proposed model. Empirical results demonstrate that our proposed algorithm is effective in tracking dynamic optimal solutions and shows great superiority comparing with state-of-the-art methods.

In order to improve the mode switching stability of a power-split hybrid electric vehicle (HEV), a torque coordinated control strategy based on disturbance observation and compensation is proposed. Aiming at the problem of engine torque disturbances caused by engine modeling error, torque control error, working environment interference and other factors, and vehicle load torque interference caused by changes in vehicle driving conditions, an improved linear extended state observer (ILESO) is designed at first. According to the deviation control principle, the state variable regulation mechanism using the same error term in the traditional linear extended state observer (TLESO) is revised, and improved by adding the corresponding deviation between the state variable and its observed value separately. Then the Lyapunov stability of the error system for the ILESO is proved gradually. On this basis, a torque redistribution algorithm of two motors based on disturbances compensation is put forward. After that, simulation verification and road adaptability analysis are carried out subsequently. The results show that the coordinated control strategy based on ILESO, compared with the TLESO, has higher observation accuracy, and makes the HEV have better vehicle speed tracking stability and mode switching smoothness when the vehicle is subject to the same external disturbance, as well as excellent adaptability under a wide range of road conditions.

In order to solve the problem of high fuel consumption and poor emission performance in high horsepower tractors, a parallel hybrid tractor system was designed using a dual power source of an engine and motor matched with a hydro-mechanical continuously variable transmission (HMCVT). An equivalent fuel consumption minimization strategy (ECMS) was used for power distribution of this hybrid system. To address the problem of poor adaptability of the equivalence factor to different working cycles in the conventional ECMS, a fuzzy adaptive equivalent fuel consumption minimization strategy (FA-ECMS) was proposed. A fuzzy PI controller based on battery SOC (State of Charge) feedback was designed to adjust the equivalence factor in real time, so as to achieve adaptive control of the equivalence factor. The physical model of the system was built by SimulationX, and the model of the control strategy was built using Matlab/Simulink. Two typical cycles of tractor plowing and road transportation were simulated. Under ECMS, the fuel consumption of the hybrid agricultural tractor was 14.3 L and 1.19 L in one plowing cycle and one transport cycle, respectively, with final battery SOC values of 60.75% and 60.32%, respectively. Under FA-ECMS, the hybrid farm tractor consumed 13.34 L and 1.13 L in one plowing cycle and one transport cycle, respectively, with final battery SOC values of 60.27% and 60.17%, respectively. The results showed that, with the introduction of a fuzzy PI controller to dynamically adjust the equivalence factor, the overall fuel consumption was reduced by 6.71% and 5.04%, respectively, and the battery power maintenance performance was improved. The designed control strategy could achieve a more reasonable power distribution between the engine and motor while maintaining the balance of the battery SOC.

The existing intelligent vehicle trajectory-planning methods have limitations in terms of efficiency and safety. To overcome these limitations, this paper proposes an automatic driving trajectory-planning method based on a variable Gaussian safety field. Firstly, the time series bird’s-eye view is used as the input state quantity of the network, which improves the effectiveness of the trajectory planning policy network in extracting the features of the surrounding traffic environment. Then, the policy gradient algorithm is used to generate the planned trajectory of the autonomous vehicle, which improves the planning efficiency. The variable Gaussian safety field is used as the reward function of the trajectory planning part and the evaluation index of the control part, which improves the safety of the reinforcement learning vehicle tracking algorithm. The proposed algorithm is verified using the simulator. The obtained results show that the proposed algorithm has excellent trajectory planning ability in the highway scene and can achieve high safety and high precision tracking control.

To improve the curve driving stability and safety under critical maneuvers for four-wheel-independent drive autonomous electric vehicles, a three-stage direct yaw moment control (DYC) strategy design procedure is proposed in this work. The first stage conducts the modeling of the tire nonlinear mechanical properties, i.e. the coupling relationship between the tire longitudinal force and the tire lateral force, which is crucial for the DYC strategy design, in the STI (Systems Technologies Inc.) form based on experimental data. On this basis, a 7-DOF vehicle dynamics model is established and the direct yaw moment calculation problem of the four-wheel-independent drive autonomous electric vehicle is solved through the nonsingular fast terminal sliding mode (NFTSM) control method, thus the optimal direct yaw moment can be obtained. To achieve this direct yaw moment, an optimal allocation problem of the tire forces is further solved by using the trust-region interior-point method, which can effectively guarantee the solving efficiency of complex optimization problem like the tire driving and braking forces allocation of four wheels in this work. Finally, the effectiveness of the DYC strategy proposed for the autonomous electric vehicles is verified through the CarSim-Simulink co-simulation results.

Accurate and reliable vehicle state estimation results are very significant to the active safety, energy optimization, and the intelligent control of vehicles. In this paper, to improve the accuracy and adaptability of vehicle running state estimation, the vehicle running states fused estimation strategy is presented for in-wheel motor drive electric vehicle using the Kalman filters and tire force compensation method. The concept of electric drive wheel model (EDWM) is developed and deduced, and then, considering that the EDWM is a nonlinear model with an unknown input, the design concept of high-order sliding mode observer is used to construct the state space equation of longitudinal force. To improve the accuracy and the reliability of vehicle state estimation, an overall estimation strategy with information fusion and tire force compensation is designed, in which a weighted square-root cubature Kalman filter with an adaptive covariance matrix of measurement noise is developed for observer design. Finally, the simulations in CarSim-Simulink co-simulation model and experiments are carried out, and the effectiveness of the designed estimation strategy is validated.

As one of the active safety technologies, stability control of vehicles has recently received great attention. In order to improve the handling stability of distributed drive electric vehicles under various extreme conditions, a direct yaw moment control (DYC) method based on a novel fuzzy sliding mode control (FSMC) is proposed. First, a linear 2DOF reference vehicle model as ideal value reference, a 7DOF vehicle model used for sideslip angle estimation, an electric-driving wheel model used to provide tire motion parameters based on CarSim platform are established. Then, FSMC is designed as the core decision-making layer of the control method to calculate the required additional yaw moment on the premise of estimating the sideslip angle. Four hub motors are allocated by the distribution method based on axle load proportion. Finally, under two typical working conditions, the four hub motors are allocated. Compared with traditional sliding mode control (SMC), the results show that FSMC can not only maintain vehicle stability more effectively under different working conditions, but also greatly reduce the occurrence of buffeting phenomenon, which has practical significance for engineering applications.

The visual guidance of AGV (automated guided vehicle) has gradually become one of the most important perception methods. Aiming at the problem that it is difficult to extract lane line accurately when AGV is running in complex working environment (such as uneven illumination, overexposure, lane line is not obvious, etc.), a scheme of lane line recognition under complex environment is proposed. Firstly, the variable scale image correction is carried out for the uneven illumination area in ROI (region of interest), and the threshold of Canny algorithm is adjusted adaptively according to the luminance of ROI region by Fuzzy-Canny algorithm; Secondly, the edge points matching the lane width feature are extracted by the way of aerial view. Finally, the curve fitting method based on RANSAC (Random Sample Consensus) is used to fit a curve with the lowest error rate and then get the lane center curve. The experimental results show that the processing algorithm used in this paper is feasible and effective, has strong robustness and fast computing performance, and can meet the requirements of intelligent AGV in various complex environments.

Tractors are usually applied in field operations, road transport, and other operations. Modern agriculture has higher design requirements for tractor powertrains due to the complicated working environments and various operations. To meet the driving requirements of the tractor under multiple operations, a mechanic-electronic-hydraulic powertrain system (MEH-PS) for tractors has been designed according to the characteristics of the hydro-mechanical composite transmission and electromechanical hybrid system. The principle of multiple driven and transmission modes of MEH-PS are introduced, the speed regulation characteristic curve of hydro-mechanical transmission (HMT) is given, and the related power element model, tractor model, and efficiency model are established. The HMT optimal economy transmission ratio control strategy and hybrid rule-based optimization energy management strategy were developed. Take three typical tractor operations for analysis: ploughing, harvesting, and transport. The results show that the engine operating points are mainly distributed in the higher load area, the tractor maintains high system efficiency, and the relative error between simulated and tested fuel consumption is within 5%, which further proves the reliability of the model. The solution also showed lower fuel consumption in all three operations compared to DLG’s announced PowerShift tractors and CVT tractors. Thus, the powertrain system can meet the tractor’s drive requirements under complex operating conditions and maintain high efficiency and is therefore suitable for tractors that need to operate frequently in the field and on the road.

The efficient harvesting of mechanical energy from ambient vibrations is an ongoing project. Recent research has shown that nonlinear energy harvesters can generally overcome many significant disadvantages of linear harvesters arising from their narrow bandwidth. This paper proposes an energy harvester within an automotive tire that boasts the advantages of nonlinear systems to increase the harvesting bandwidth by combining stochastic resonance with high-energy orbit oscillations. A major challenge in this automotive application is the wide variation in tire speeds over which harvesting can take place. Stochastic resonance has some benefits here, particularly at low speeds at which high-energy orbit oscillations may not otherwise occur. Thus, as the speed of the vehicle and, therefore, a high-energy orbit oscillation of the harvester can be stimulated, the operating frequency of the harvester increases due to the presence of stochastic resonance and can be maintained as the speed of the vehicle increases or decreases. The results of numerical simulations and laboratory experiments show that the effective bandwidth of energy harvesting increased from 31 rad/s to 129.4 rad/s and the maximum power generated reached 0.21 mW, with a mean value of 35.6 μW, entirely through a combination of sustainable stochastic resonance and high-energy orbit oscillations within the harvester. Thus, by combining these two phenomena, the efficient bandwidth of rotation can be further extended to enhance the overall capability of tire-based energy harvesting.

The study on the spatial-temporal characteristics of highway traffic flow is helpful to deeply understand the inherent evolution of highway traffic system and provide a theoretical basis for prediction and control of highway traffic flow. This paper makes an empirical analysis on the spatial-temporal characteristics of highway traffic flow using manifold similarity index and manifold learning technology. The time series of highway traffic flow is converted into the distance series containing manifold features to calculate the manifold distance between multi-section traffic flow data points, which are highly similar to spatial-temporal distribution of traffic flow speed parameters, and then, the levels calibration of traffic state is carried out according to the manifold distance, so as to reveal the distribution rule of spatial-temporal characteristics of highway traffic flow. Its prediction error is obviously lower than the traditional distance measurement method, which has higher accuracy. The research of this paper can provide new ideas and methods to reveal the highway traffic flow evolution and traffic state prediction.

Recovering missing values from incomplete traffic sensor data is an important task for intelligent transportation system because most algorithms require data with complete entries as input. Self-representation-based matrix completion attempts to optimally represent each sample by linearly combining other samples when conducting missing values recovery. Typically, it implements sparse or dense combination through imposing either l1-norm or l2-norm regularization over the representation coefficients, which is not always optimal in practice. To permit more flexibility, we propose in this paper a novel approach termed as lp-norm regularized sparse self-representation (SSR-lp) by incorporating nonconvex lp-norm with 0 1-norm and in turn facilitates the accurate recovery of missing data. We further develop an efficient iterative algorithm for solving SSR-lp. The performance of this method is evaluated on a real-world road network traffic flow data set. The experimental results verify the advantage of our method over other competing algorithms in recovering missing values.

Electromagnetic slip coupling (EMSC), like other contactless transmission devices, has lots of advantages over contact transmissions such as clutch function, continuous speed regulation, and no mechanical wear. Nevertheless, the EMSC has inherent slip energy especially under the large-slip condition, which leads to low efficiency. A novel energy-regenerative EMSC was developed including wound-type EMSC and the energy-regenerative apparatus for slip energy recovery. To improve recovery efficiency of the EMSC, the field-oriented control strategy was proposed. In this case, the mathematical model of EMSC under synchronously rotating coordinate system was constructed and verified by experiment. Since direct-axis component and quadrature-axis component of current in three-phase windings of the EMSC are coupled, the feed forward decoupling control approach was employed. Space vector pulse width modulation technique was utilized to eliminate harmonic components of current in three-phase windings. Simulations and experiments were carried out to validate the proposed control strategy. The results showed that the voltage and current in three-phase windings were almost in-phase, the current only contained few of harmonic components, and slip energy of the EMSC decreased by 82.5%. In conclusion, the field-oriented control strategy is conducive to recycle the slip energy of the EMSC efficiently.

The energy conversion efficiencies among different sources of power split hybrid electric vehicle rely on the energy management strategy. In this paper, the energy management of a power split hybrid electric bus (HEB) is described as the predictive control problem based on the linear control-oriented model of the HEB. In order to ensure that the engine can output the desired power, the fuzzy PI controller, which can realize the optimal engine speed tracking, is further designed. Two real-world driving cycles are analyzed and formulated to evaluate the vehicle fuel economy under transient practical conditions. On this basis, the driver torque demand and vehicle velocity in the prediction horizon are derived with the stochastic one-step Markov chain. Finally, the hardware-in-the-loop (HIL) simulation platform is built to explore the validity of the controller. Compared with the adaptive equivalent fuel consumption minimization strategy, HIL test results demonstrate the real-time capability and benefits of the proposed approach in optimizing the energy management of HEB.

Short-term traffic state forecasting is critical for real-time traffic control, but due to its complexity and its nonlinear nature, it is difficult to obtain a high degree of precision. The “k-nearest neighbors”model has been widely used to solve nonlinear regression and time series forecasting. This paper presents a traffic state forecasting method using adaptive neighborhood selection based on expansion strategy to search manifold neighbors to get higher precision with manifold neighbors. We propose a method of linear structure to handle the traffic data in Euclidean space to find a manifold neighbor that is more suitable for predicting traffic states. The results of extensive comparison experiments indicate that the proposed model can produce more accurate forecasting results than other classic algorithms.

The tire longitudinal properties exhibit highly nonlinear dynamical behaviors influenced by the tire slip coefficient, the road longitudinal adhesion coefficient, and the tire vertical load. This paper presents a system identification approach to approximate such a nonlinear dynamic for autonomous driving controller design. The tire longitudinal properties tests are conducted using a flat-plate test bench. On the basis of the experimental data, the piecewise affine (PWA) identification of the tire longitudinal properties involves the classification of the cluster data and the parameter estimation of the affine submodels. Using the least-square algorithm, the parameter vectors of the affine submodels are estimated, and the problem of region partition is solved via heuristic approach. The simulation results of the identified PWA model match the experimental data accurately, which demonstrates the effectiveness of the proposed identification approach for the tire longitudinal properties.

In order to improve the extension mileage and ensure the yaw stability of the distributed drive electric vehicle simultaneously, a multi-objective coordination control strategy is presented in this paper, in which the optimal vehicle-state estimation method and the orientated tire force distribution method are designed. The longitudinal force estimation is designed using the unknown input observer Kalman filter, a novel yaw rate and a vehicle sideslip angle estimation scheme is proposed by an observer error compensation method. A multi-objective hierarchical vehicle motion control strategy is developed to monitor the yaw stability and improve the energy efficiency of vehicle in real time. In the upper layer controller design, a sliding mode controller is proposed to track the desired yaw rate. In the lower layer controller design, the orientated tire force distribution scheme is presented to improve the motor efficiency and reduce the energy loss caused by vehicle steering resistance while ensuring the yaw stability of vehicle. Finally, the effectiveness of the proposed estimation method and the global coordination control strategy is validated by simulation in CarSim-Simulink joint platform and road test.

Existing machine-learning-based vehicle detection algorithms for intelligent vehicles have an obvious disadvantage in that the detection effect decreases dramatically when the distribution of training samples and the scene target samples do not match. To address this issue, a scene-adaptive vehicle detection algorithm based on a composite deep structure is proposed in this paper. Inspired by the Bagging (Bootstrap aggregating) mechanism, multiple relatively independent source samples are first used to build multiple classifiers and then voting is used to generate target training samples with confidence scores. The automatic feature extraction ability of deep convolutional neural network is then used to perform source-target scene feature similarity calculations with a deep auto-encoder in order to design a composite deep-structure-based scene-adaptive classifier and its training method. Experiments on the KITTI data set and a data set captured by our group demonstrate that the proposed method performs better than existing machine-learning-based vehicle detection methods. In addition, compared with existing scene-adaptive object detection methods, our method improves the detection rate by an average of approximately 3%.

Night time pedestrian detection is more and more important in advanced driver assistant systems (ADAS). Traditional pedestrian detection algorithms in far infrared (FIR) images lack accuracy and have long processing times. Focusing on this issue, in this paper, a visual saliency-based pedestrian detection algorithm is proposed. First, areas that contain suspected pedestrians are detected using a fusion saliency-based method. Then, the sub-image of the suspected pedestrian is used as an input to a histogram of local intensity difference feature and cross kernel-based support vector machine classifier to make a final determination. Experiments performed using a real FIR road image data set demonstrated that the proposed fusion saliency-based region of interest (ROI) detection method has the largest pedestrian inclusion rate and the smallest ROI proportion compared with three other methods. Besides, compared with existing state-of-the-art pedestrian detection algorithms, the proposed method demonstrates a much higher pedestrian detection rate with a comparably short processing time.

The independent driving ability of the four-wheel independent drive electric vehicles can be used to improve the energy efficiency of the vehicle. The composition and structure of a prototype were introduced in this article and the rapid prototype was applied to design the upper-level control system. Experimental data obtained from the bench tests and the chassis dynamometer tests were employed to generate the torque response performance and power efficiency map of the in-wheel motor. The control logic of the in-wheel motor was obtained based on the least squares method. Besides, a front and rear axle torque distribution method was proposed to reduce the energy consumption of vehicle, and the simulating test of the prototype was conducted based on the in-wheel motor efficiency map. Simulation analysis of vehicle energy consumption in vehicle driving cycle was carried out in ways of rear axle drive, average torque allocation, and optimized torque energy-saving allocation; the results show that the optimized torque energy-saving allocation strategy can decrease the energy consumption up to approximately 0.117 and 0.426 kJ per unit mileage comparing with the rear axle drive and average torque allocation.

Short-term traffic flow forecasting is a difficult yet important problem in intelligent transportation systems. Complex spatiotemporal interactions between the target road segment and other road segments can provide important information for the accurate forecasting. Meanwhile, spatiotemporal variable selection and traffic flow prediction should be solved in a unified framework such that they can benefit from each other. In this article, we propose a novel sparse hybrid genetic algorithm by introducing sparsity constraint and real encoding scheme into genetic algorithm in order to optimize short-term traffic flow prediction model based on least squares support vector regression. This method can integrate spatiotemporal variable selection, parameter selection as well as traffic flow prediction in a unified framework, indicating that the “goodness,” that is, contribution, of selected spatiotemporal variables and optimized parameters directly depends on the final traffic flow prediction accuracy. The real-world traffic flow data are collected from 24 observation sites located around the intersection of Interstate 205 and Interstate 84 in Portland, OR, USA. The experimental results show that the proposed sparse hybrid genetic algorithm-least square support vector regression prediction model can produce better performance but with much fewer spatiotemporal variables in comparison with other related models.

Accurate and reliable estimation information of sideslip angle is very important for intelligent motion control and active safety control of an autonomous vehicle. To solve the problem of sideslip angle estimation of an autonomous vehicle, a sideslip angle fusion estimation method based on robust cubature Kalman filter and wheel-speed coupling relationship is proposed in this paper. The vehicle dynamics model, tire model, and wheel speed coupling model are established and discretized, and a robust cubature Kalman filter is designed for vehicle running state estimation according to the discrete vehicle model. An adaptive measurement-update solution of the robust cubature Kalman filter is presented to improve the robustness of estimation, and then, the wheel-speed coupling relationship is introduced to the measurement update equation of the robust cubature Kalman filter and an adaptive sideslip angle fusion estimation method is designed. The simulations in the CarSim-Simulink co-simulation platform and the actual vehicle road test are carried out, and the effectiveness of the proposed estimation method is validated by corresponding comparative analysis results.

Vehicle detection is a key component of environmental sensing systems for Intelligent Vehicles (IVs). The traditional shallow model and offline learning-based vehicle detection method are not able to satisfy the real-world challenges of environmental complexity and scene dynamics. Focusing on these problems, this work proposes a vehicle detection algorithm based on a multiple feature subspace distribution deep model with online transfer learning. Based on the multiple feature subspace distribution hypothesis, a deep model is established in which multiple Restricted Boltzmann Machines (RBMs) construct the lower layers and a Deep Belief Network (DBN) composes the superstructure. For this deep model, an unsupervised feature extraction method is applied, which is based on sparse constraints. Then, a transfer learning method with online sample generation is proposed based on the deep model. Finally, the entire classifier is retrained online with supervised learning. The experiment is actuated using the KITTI road image datasets. The performance of the proposed method is compared with many state-of-the-art methods and it is demonstrated that the proposed deep transfer learning-based algorithm outperformed existing state-of-the-art methods.

Visual-based vehicle detection has been studied extensively, however there are great challenges in certain settings. To solve this problem, this paper proposes a probabilistic framework combining a scene model with a pattern recognition method for vehicle detection by a stationary camera. A semisupervised viewpoint inference method is proposed in which five viewpoints are defined. For a specific monitoring scene, the vehicle motion pattern corresponding to road structures is obtained by using trajectory clustering through an offline procedure. Then, the possible vehicle location and the probability distribution around the viewpoint in a fixed location are calculated. For each viewpoint, the vehicle model described by a deformable part model (DPM) and a conditional random field (CRF) is learned. Scores of root and parts and their spatial configuration generated by the DPM are used to learn the CRF model. The occlusion states of vehicles are defined based on the visibility of their parts and considered as latent variables in the CRF. In the online procedure, the output of the CRF, which is considered as an adjusted vehicle detection result compared with the DPM, is combined with the probability of the apparent viewpoint in a location to give the final vehicle detection result. Quantitative experiments under a variety of traffic conditions have been contrasted to test our method. The experimental results illustrate that our method performs well and is able to deal with various vehicle viewpoints and shapes effectively. In particular, our approach performs well in complex traffic conditions with vehicle occlusion.

The emergence and development of deep learning theory in machine learning field provide new method for visual-based pedestrian recognition technology. To achieve better performance in this application, an improved weakly supervised hierarchical deep learning pedestrian recognition algorithm with two-dimensional deep belief networks is proposed. The improvements are made by taking into consideration the weaknesses of structure and training methods of existing classifiers. First, traditional one-dimensional deep belief network is expanded to two-dimensional that allows image matrix to be loaded directly to preserve more information of a sample space. Then, a determination regularization term with small weight is added to the traditional unsupervised training objective function. By this modification, original unsupervised training is transformed to weakly supervised training. Subsequently, that gives the extracted features discrimination ability. Multiple sets of comparative experiments show that the performance of the proposed algorithm is better than other deep learning algorithms in recognition rate and outperforms most of the existing state-of-the-art methods in non-occlusion pedestrian data set while performs fair in weakly and heavily occlusion data set.