Search Results

For intelligent vehicles, a robust perception system relies on training datasets with a large variety of scenes. The architecture of federated learning allows for efficient collaborative model iteration while ensuring privacy and security by leveraging data from multiple parties. However, the local data from different participants is often not independent and identically distributed, significantly affecting the training effectiveness of autonomous driving perception models in the context of federated learning. Unlike the well-studied issues of label distribution discrepancies in previous work, we focus on the challenges posed by scene heterogeneity in the context of federated learning for intelligent vehicles and the inadequacy of a single scene for training multi-task perception models. In this paper, we propose a federated learning-based perception model training system.

To improve the braking energy recovery rate of pure electric garbage removal vehicles and ensure the braking effect of garbage removal vehicles, a strategy using particle swarm algorithm to optimize the regenerative braking fuzzy control of garbage removal vehicles is proposed. A multi-section front and rear wheel braking force distribution curve is designed considering the braking effect and braking energy recovery. A hierarchical regenerative braking fuzzy control strategy is established based on the braking force and braking intensity required by the vehicle. The first layer is based on the braking force required by the vehicle, based on the front and rear axle braking force distribution plan, and uses fuzzy controllers.

Autonomous vehicles require the collaborative operation of multiple modules during their journey, and enhancing tracking performance is a key focus in the field of planning and control. To address this challenge, we propose a cooperative control strategy, which is designed based on the integration of model predictive control (MPC) and a dual proportional–integral–derivative approach, referred to as collaborative control of MPC and double PID (CMDP for short in this article).The CMDP controller accomplishes the execution of actions based on information from perception and planning modules. For lateral control, the MPC algorithm is employed, transforming the MPC’s optimal problem into a standard quadratic programming problem. Simultaneously, a fuzzy control is designed to achieve adaptive changes in the constraint values for steering angles.

The main objective of platoon control is coordinated motion of autonomous vehicle platooning with small intervehicle spacing while maintaining the same speed and acceleration as the leading vehicle, which can save energy consumption and improve traffic throughput. The conventional platoon control methods are confronted with the problem of manual parameter tuning. In order to addres this isue, a novel bifold platoon control approach leveraging a deep reinforcement learning-based model is proposed, which enables the platoon adapt to the complex traffic environment, and guarantees the safety of platoon. The upper layer controller based on the TD3 tuned PID algorithm outputs the desired acceleration. This integration mitigates the inconvenience of frequent manual parameter tuning asociated with the conventional PID algorithm. The lower layer controller tracks the desired acceleration based on the inverse vehicle dynamics model and feedback control.

The phenomenon of liquid transfer in the liquid tank of the semi-trailer vehicle for transporting dangerous cargo (SVTDC) during braking is analyzed and the relevant mathematical model is established. The braking dynamic model of the SVTDC considering the liquid sloshing in the tank is established, and the model is verified based on the co-simulation method. Based on the typical conditions, the braking deceleration and axle load calculation functions of the model are simulated and analyzed, and the application prospect of the model in the development of driving automation control strategy is discussed.

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.

This paper proposed a model predictive control(MPC) based car-following control strategy for electric vehicles considering comfort, in order to improve the comfort of the car-following control system of electric vehicles. The MPC algorithm is improved in the following three aspects to improve the comfort: Firstly, a five-state longitudinal car-following model is adopted, so that the MPC algorithm can optimize the acceleration and acceleration change rate of the ego vehicle. Secondly, for the weight coefficients of the output vector and the input vector of the objective function, the fixed weight coefficients are changed into variable weight coefficients by the way of Nash equilibrium game, so that the control system can improve the weight of the parameters used to control the comfort under suitable driving conditions.

For cooperative adaptive cruise control (CACC) system, a robust following control algorithm based on fuzzy PID principle is adopted in this paper. Firstly, a nonlinear vehicle dynamics model considering the lag of driving force and acceleration constraints was established. Then, with the vehicle’s control hierarchic, the upper controller takes the relative speed between vehicles and the spacing error as inputs to output the following vehicle's target acceleration, while the lower controller takes the target acceleration as inputs and the throttle opening and brake master cylinder pressure as outputs. For the setting of target spacing, this paper additionally considers the relative speed between vehicles and the acceleration of the front vehicle. Through testing, compared with the traditional variable safety distance model, the average distance reduces by 5.43% when leading vehicle is accelerating, while increases by 2.74% in deceleration.

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.

Accurate surrounding vehicle motion prediction is critical for enabling safe, high quality autonomous driving decision-making and motion planning. Aiming at the problem that the current deep learning-based trajectory prediction methods are not accurate and effective for extracting the interaction between vehicles and the road environment information, we design a target vehicle intention-aware dual attention network (IDAN), which establishes a multi-task learning framework combining intention network and trajectory prediction network, imposing dual constraints. The intention network generates an intention encoding representing the driver’s intention information. It inputs it into the attention module of the trajectory prediction network to assist the trajectory prediction network to achieve better prediction accuracy.

With the rapid development of the logistics and transportation industry, heavy-duty trucks play an increasingly important role in social life. However, due to the characteristics of large cargo loads, high center of mass and relatively narrow wheelbase, the driving stability of heavy trucks are poor, and it is easy to cause rollover accidents under high-speed driving conditions, large angle steering and emergency obstacle avoidance. To improve the roll stability of heavy trucks, it is necessary to design an active anti-rollover control system, through the analysis of the yaw rate and the load transfer rate of the vehicle, driving states can be estimated during the driving process. Under the intervention of the control system, the lateral transfer rate of heavy trucks can be reduced to correct the driving posture of the vehicle body and reduce the possibility of rollover accidents.

With the increasingly serious global environmental and energy problems, as well as the increasing number of vehicles, pure electric vehicles with its advantages of environmental protection, low noise and renewable energy, become an effective way to alleviate environmental pollution and energy crisis. Due to the current pure electric vehicle power battery technology is not perfect, the range of pure electric vehicle has a great limit. Through the braking energy recovery, the energy can be reused, the energy utilization rate can be improved, and the battery life of pure electric vehicles can be improved. In this paper, a pure electric vehicle is taken as the analysis object, and the whole vehicle analysis model is built. Through the comparative analysis, based on the driver's braking intention and vehicle running state, the braking energy recovery control strategy of double fuzzy control is proposed.

Road traffic accidents resulting from alcohol-impaired driving are increasing globally despite several measures, currently in place, to curb the trend. For this reason, recent research aims at integrating alcohol early-detection systems and driving simulator experiments to identify intoxicated drivers. However, driving simulator experiments on drunk driving have focused mostly on male participants than female drivers whose characteristics have scarcely been explored. Hence in this paper, vehicle dynamic control inputs on steering, braking, and acceleration performance of 75 licensed female drivers with an upshot of alcohol at four different blood alcohol concentration (BAC) levels (0%, 0.03%, 0.05%, and 0.08%) were investigated. The participants completed simulated driving in a fixed-based simulator experiment coupled with real-time ecological scenarios to extract discrete responses.

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.

Based on the fuzzy inference system, it constructs a discretionary lane-changing decision model for different types of preceding vehicles and compares and analyzes the parameter differences of their input membership functions. According to the driver questionnaire survey, the model uses three parameters that drivers can easily percept as the model input—preceding vehicle distance in the current lane, preceding vehicle distance in the target lane, and following-vehicle distance in the target lane—uses Next-Generation Simulation (NGSIM) vehicle trajectory data to optimize the input membership functions of models based on genetic algorithm according to different vehicle lane-changing trajectory data to analyze the impact of the preceding vehicle type before lane change to the intelligent lane-changing decision.

Discretionary lane changing is commonly seen in highway driving. Intelligent vehicles are expected to change lanes discretionarily for better driving experience and higher traffic efficiency. This study proposed to optimize the decision-making and trajectory-planning process so that intelligent vehicles made lane changes not only with driving safety taken into account, but also with the goal to improve driving comfort as well as to meet the driver’ s expectation. The mechanism of how various factors contribute to the driver’s intention to change lanes was studied by carrying out a series of driving simulation experiments, and a Lane-Changing Intention Generation (LCIG) model based on Bi-directional Long Short-Term Memory (Bi-LSTM) was proposed.

With the increase of car ownership and the complex and crowded parking environment, it is difficult for drivers to complete the parking operation quickly and accurately, which may cause traffic accidents such as vehicle collisions and road jams because of poor parking skills. The emergence of an automatic parking system can help drivers park safely and reduce the occurrence of safety accidents. In this paper, the neural network identifier on the control method of an adaptive integral derivative of a neural network is proposed for an automatic parallel parking system with front-wheel steering is studied by using MATLAB/Simulink environment, and the simulation is carried out. Firstly, according to vehicle parameters and obstacle avoidance constraints, the minimum parking space, and parking starting position are calculated. Meanwhile, the path planning of parallel parking spaces is carried out by quintic polynomial.

Deep Deterministic Policy Gradient (DDPG) is one of the Deep Reinforcement Learning algorithms. Because of the well perform in continuous motion control, DDPG algorithm is applied in the field of self-driving. Regarding the problems of the instability of DDPG algorithm during training and low training efficiency and slow convergence rate. An improved DDPG algorithm based on segmented experience replay is presented. On the basis of the DDPG algorithm, the segmented experience replay select the training experience by the importance according to the training progress to improve the training efficiency and stability of the training model. The algorithm was tested in an open source 3D car racing simulator called TORCS. The simulation results demonstrate the training stability is significantly improved compared with the DDPG algorithm and the DQN algorithm, and the average return is about 46% higher than the DDPG algorithm and about 55% higher than the DQN algorithm.

In order to ensure driving safety, heavy vehicles are often equipped with hydraulic retarder, which provides sustained, stable braking torque and converts the vehicle kinetic energy into heat taken away by the cooling system when traveling on a long downhill. The conventional hydraulic retarder braking torque is modulated by adjusting the liquid filling rate, which leads to slow response and difficult control. In this paper, a new kind of magnetorheological (MR) fluid hydraulic retarder is designed by replacing the traditional transmission oil with MR fluid and arranging the excitation coils outside the working chamber. The braking torque can be controlled by the fluid viscosity of MR fluid with the variation of magnetic field. Compared with the traditional hydraulic retarder, the system has the advantages of fast response, easy control and high adjustment sensitivity.

In this paper a new pressure control method of a modified accumulator-type Electro-hydraulic Braking System (EHB) is proposed. The system is composed of a hydraulic motor pump, an accumulator, an integrated master cylinder, a pedal feel simulator, valves and pipelines. Two pressurizing modes are switched between by-motor and by-accumulator to adapt different pressure boost demands. A differentiator filtering raw sensor signal and calculating pedal speed is designed. By using the pedal feel simulator, the relationship between wheel pressures and brake force is decoupled. The relationships among pedal displacement, pedal force and wheel pressure are calibrated by experiments. A model-based PI controller with predictor is designed to lower the influences caused by delay. Moreover, a self-tuning regulator is introduced to deal with the parameter’s time-varying caused by temperature, brake pads wearing and delay variation.