This paper proposes a multi-vehicle cooperative control framework to address cyber-attack interference and environmental uncertainty in complex traffic scenarios. First, to tackle the cooperative adaptive cruise control (CACC) problem under network attacks, an adaptive stochastic hybrid model predictive control (SHMPC) approach is developed by integrating deep deterministic policy gradient (DDPG) with mixed logical dynamics (MLD), enabling effective mode switching and online tuning of control parameters, thereby enhancing system robustness and adaptability. Second, for safe lane changing in dynamic traffic, an stochastic model predictive control (SMPC) controller incorporating a probabilistic control barrier function (PCBF) and iterative linear quadratic regulator (iLQR)-based trajectory planning is constructed to explicitly model environmental uncertainties and enforce safety through probabilistic constraints. Finally, comprehensive simulations validate the effectiveness of the proposed models: under stochastic cyber-attacks, the tracking distance error is reduced by about 11%; for lane change tasks, the success rate is increased by 24.6%, with significantly fewer collisions. The results demonstrate the proposed framework’s superiority in robustness, safety, and response efficiency.

This paper presents a systematic investigation into path planning and trajectory tracking for autonomous vehicles. By integrating an improved A* algorithm, a fuzzy dynamic window approach, and a Fuzzy PID control strategy, the proposed method enables effective driving of an autonomous vehicle. Firstly, in the global path planning phase, to address the issues of low computational efficiency and suboptimal path quality in traditional A* algorithms for large-scale map searches, an improved A* algorithm incorporating an enhanced heuristic function, redundant node removal strategy, and path smoothing approach is introduced, significantly increasing search efficiency and optimizing path quality. Secondly, in the local path planning phase, the dynamic adjustment of vehicle speed and steering is achieved by combining fuzzy logic control with the dynamic window approach. This allows for smooth obstacle avoidance in dynamic environments. Furthermore, a path smoothing algorithm is integrated to refine the generated trajectory, ensuring its continuity and smoothness. Finally, a Fuzzy PID control algorithm is integrated into the trajectory tracking controller. By introducing fuzzy logic, the PID parameters are adaptively adjusted to ensure precise vehicle following of the planned path, improving path tracking stability and response speed. The proposed method is validated and evaluated in a variety of complex road scenarios using a real vehicle based on ROS. The simulation and real-world experimental results clearly illustrate that the proposed method achieves substantially better performance than conventional approaches with regard to path planning efficiency, obstacle avoidance success rate, and path smoothness.

Safety has become a critical issue in the development of automated driving systems (ADSs). Drivers' perception of risk determines their acceptance, trust, and use of ADS. However, driver's perceived risk is subjective and difficult to evaluate using traditional risk assessment methods. To address this issue, a driver's subjective perceived risk (DSPR) model is proposed, which regards driver's perceived risk as a dynamically triggered mechanism that exhibits anisotropy and attenuation. Subsequently, 20 participants are recruited for a driver-in-the-loop experiment to report their real-time subjective risk ratings (SRRs) when experiencing various real-world automated driving scenarios. A convolutional neural network and bidirectional long short-term memory with temporal pattern attention (CNN-Bi-LSTM-TPA) network is embedded into a semisupervised learning strategy to predict driver's SRRs, aiming to reduce data noise caused by the subjective randomness of drivers. Results illustrate that our proposed DSPR model combined with the semisupervised learning strategy achieves the highest prediction accuracy of 87.91% in predicting driver's SRRs, outperforming other three state-of-the-art risk models. Compared to the model trained on the original data, the semisupervised method improves accuracy by 20.12%. The proposed CNN-Bi-LSTM-TPA network presents the highest among four different network structures. Finally, a genetic algorithm is applied to optimize parameters, which improves the accuracy to 89.95%. This study offers an effective method for assessing driver's perceived risk, providing support for the safety enhancement of ADS and driver's trust improvement.

Robust and reliable human–machine shared driving (HMSD) is essential for balancing safety and comfort. Within a connected urban arterial system, rare high-risk long-tail disturbances can trigger conflicts, lane departures, and oscillatory flow, degrading system safety and efficiency of the HMSD. To mitigate these effects, an interactive learning framework is built by coupling a scalable environment model with a dual-discriminator generative adversarial network to synthesise diverse, high-fidelity extreme scenarios, thereby enlarging the training domain of HMSD. A bidirectional loop between the environment model and the decision maker enables continuous refinement of the control policy, risk suppression, and improvement in response efficiency. The trained controller leverages cooperative vehicle–infrastructure sensing to derive shared risk states and to adaptively allocate authority between the human driver and automation in real time. The robustness of the proposed method is validated by comparing it with other shared control frameworks on a hierarchical validation platform, including a driver-in-the-loop (DiL) system. The results demonstrate that this method offers a superior balance between driving safety, stability, and pleasure, while demonstrating practical and robust control performance under long-tail events.

In this paper, we focus on trajectory planning for lane change maneuvers and elaborately consider the impacts of dynamic uncertainties from surrounding vehicles. First, a risk-aware corridor is developed to guarantee that the collision probability is below an acceptable risk level. In contrast to the existing studies, a more succinct and intuitive corridor is constructed based on an explicit safety check theorem. Since it relies solely on the distances between autonomous vehicles and surrounding ones, such a corridor can be easily converted into polytopic constraints and seamlessly integrated with an optimization scheme. Furthermore, a computationally tractable and hierarchical scheme is designed to decide the optimal merging instant and generate the expected trajectory. In this scheme, the trade-off between optimality and complexity in calculating merging instants can be well balanced through feasibility estimation and optimum searching. Furthermore, the trajectory is optimized by model predictive control (MPC) technique. Finally, both numerical simulations and naturalistic validations are conducted to verify the effectiveness. The comparison results indicate the convincing superiority of our scheme in achieving safer driving in an uncertain environment. The code is now available athttps://github.com/Aeolson/code_otp.git

Pedestrian crossing intention prediction is critical for autonomous vehicles. Analyzing pedestrian motion characteristics provides a preemptive decision-making basis for the planning module. The prediction accuracy directly affects the timeliness of collision avoidance mechanisms in vehicles, especially in mixed traffic scenarios, which is crucial for the safety and reliability of autonomous driving systems. Existing methods struggle to jointly achieve accuracy and real-time performance due to motion-unaware representations and inflexible fusion, which introduce redundancy and latency on embedded platforms. To tackle this problem, we propose PCIP, which leverages adaptive multimodal fusion to achieve efficient and robust crossing intention prediction. PCIP proposes three key contributions: 1) The dynamic selection of the quasi-polar coordinate system accurately characterizes the relative motion trends between pedestrians and vehicles through adaptive multi-pole selection. 2) The Koopman-based CI-TDE module linearizes the spatiotemporal dynamics of skeletal actions, significantly enhancing the expressive capability of temporal features that describe pedestrian motions. 3) A multi-scale dynamic weighting strategy based on distance and angle effectively suppresses noisy modalities and enhances pivotal features. Experiments show that PCIP achieves intention classification accuracies of 89% and 93% on the JAAD and PIE datasets, respectively, and achieves a real-time inference speed of 0.5ms per frame on an embedded platform (NVIDIA Jetson AGX). This study provides a lightweight and robust solution for pedestrian intention prediction in complex traffic scenarios.

Vehicle trajectory prediction is important for automated vehicles to understand driving scenarios. This paper proposes an encoder-decoder network-based parameterized transfer learning (EDN-PTL) model to predict vehicle trajectory. To improve trajectory prediction accuracy, the motion interaction between the target vehicle and the surrounding vehicles is considered, and a multidimensional spatiotemporal input expansion (MSIA) strategy is proposed to extend the feature dimensions. Additionally, global and local scale features, as well as long and short horizon features, are extracted and used for interactive vehicle trajectory prediction by a CNN and LSTM-based encoder-decoder network (CNN-LSTM-EDN). Moreover, the features extracted by CNN-LSTM-EDN are integrated using a stacked convolutional social pooling network (SCSPN). To enhance the environmental adaptability of the trajectory prediction model, a PTL strategy is proposed to enable transfer learning capabilities of EDN-PTL. Based on the PTL strategy, trajectory prediction accuracy is maintained even when applied to untrained environments. The proposed EDN-PTL model is validated on three types of publicly available naturalistic datasets and compared with several baselines and state-of-the-art (SOTA) methods. The validation results demonstrate that the proposed EDN-PTL achieves better prediction accuracy, robustness, and environmental adaptability compared to the baselines and SOTA methods.

Drivers' perception of risk determines their acceptance, trust, and use of the Automated Driving Systems (ADSs). However, perceived risk is subjective and difficult to evaluate using existing methods. To address this issue, a driver's subjective perceived risk (DSPR) model is proposed, regarding perceived risk as a dynamically triggered mechanism with anisotropy and attenuation. 20 participants are recruited for a driver-in-the-loop experiment to report their real-time subjective risk ratings (SRRs) when experiencing various automatic driving scenarios. A convolutional neural network and bidirectional long short-term memory network with temporal pattern attention (CNN-Bi-LSTM-TPA) is embedded into a semi-supervised learning strategy to predict SRRs, aiming to reduce data noise caused by subjective randomness of participants. The results illustrate that DSPR achieves the highest prediction accuracy of 87.91% in predicting SRRs, compared to three state-of-the-art risk models. The semi-supervised strategy improves accuracy by 20.12%. Besides, CNN-Bi-LSTM-TPA network presents the highest accuracy among four different LSTM structures. This study offers an effective method for assessing driver's perceived risk, providing support for the safety enhancement of ADS and driver's trust improvement.

We proposed an unsupervised pipeline AW-FRVP for denoising of point clouds in adverse weather to alleviate the noise points’ impact on perception. Specifically speaking, AW-FRVP consists of two stages: firstly, AWDenoiseNet was adopted to gain efficient pseudo labels by converting the point cloud into frequency domain with depth and intensity features. Secondly, the semantic segmentation network FRVPNet was trained based on the generated pseudo labels and tested for binary classification of noise and environmental points. In the first stage, the KNN-FLATTEN and Mamba-RFE feature extractor were equipped to alleviate the many-to-one issue while converting the point cloud to a range perspective and strengthening the key feature extraction in the frequency domain. In the second stage, point features corresponding to different perspectives were transformed and fused, and two modules named FPFusion-with-CSMA, as well as Triple-Points-Fusion-head, were adopted to gain point-wise features with rich information for subsequent binary classification. Amount of experiments show that the AWDenoiseNet can gain real-time and efficient point cloud denoising with precision of 93.47% on WADS and recall of 94.70% on Weather-KITTI (Rain Scenes). Without annotation, the AWDenoiseNet can gain the SOTA performance among the unsupervised denoising methods. What’s more, the training process of FRVPNet based on pseudo labels generated by AWDenoiseNet can gain sound denoising performance. Both AWDenoiseNet and FRVPNet can achieve real-time denoising with frequencies of about 165 FPS (Frame Per Second) and 19 FPS. In summary, our two-stage unsupervised denoising pipeline can achieve real-time SOTA label-free denoising with a slight gap to those fully-supervised methods (3.65% in precision on WADS, and 5.71% in recall on rain scenes of Weather-KITTI) and our method can significantly reduce the interference of noise in point clouds on perception algorithms for internet of traffic participants.

Accurate prediction of vehicle motions is imperative for enabling cooperative perception and planning of autonomous vehicles, however effective modeling of complex spatio-temporal interactions and long-term dependencies between vehicles remains a formidable challenge. To tackle these issues, we propose a novel motion query-based social transformer network (MOST) for vehicle trajectory and intention prediction through a multi-task approach, which is composed of temporal transformer encoder module, social interaction module and motion queries-based multi-task feature decoder in a hierarchical manner. The temporal transformer is responsible for capturing long-range temporal correlations of individual motion states through a self-attention mechanism with residual connection, while the spatial interaction dependencies between vehicles are acquired through the social interaction module by constructing social tensors. Furthermore, Considering the uncertainty and diversity of future vehicle behaviors, a motion query-based feature decoder is proposed, which is equipped with learnable parameters to assimilate prior knowledge and generate multiple possible future trajectories and intentions. To assess our model’s effectiveness, we carried out comprehensive experiments on the open-source NGSIM and HighD dataset. The results demonstrated that our approach reaches unparalleled performance, with an average prediction accuracy improvement of about 50% on the NGSIM dataset and 20% on the HighD dataset compared to the state-of-the-art method.

Vehicle-to-Vehicle (V2V) cooperative perception markedly broadens scene comprehension by aggregating observations from multiple connected vehicles. However, existing methods often suffer from performance degradation when confronted with domain gaps between training and deployment environments. To bridge this domain gap, we introduce V2V-APG, an Adversarial Progressive Generalization framework for cross-domain V2V cooperative perception. V2V-APG first leverages a Spatial-Interaction Perception Attention network to simultaneously model intra-vehicle spatial dependencies and inter-vehicle relationships, yielding rich, discriminative features. Additionally, we propose a Domain-Aware Dynamic Gradient Reversal Layer to dynamically align feature distributions based on real-time domain discrepancies. Finally, a Confidence-Guided Progressive Pseudo-label Refinement strategy iteratively refines target-domain pseudo-labels through adaptive thresholding and curriculum learning, boosting supervision quality without ground-truth annotations. Extensive experiments on the OPV2V and V2V4Real datasets demonstrate that V2V-APG outperforms state-of-the-art methods in cross-domain tasks, achieving superior generalization and robustness.

Pedestrian trajectory prediction plays a critical role in ensuring the safe operation of autonomous vehicles. Predicting from an egocentric view can eliminate the cumulative computational errors associated with scene perspective transformations. However, compared to predictions from a bird’s-eye view, a key challenge in the egocentric setting is that both the ego vehicle’s motion and the pedestrian’s motion simultaneously influence the target’s movement. To address this, we propose an agent-wise motion fusion network (AANet), which efficiently predicts the multimodal trajectories by learning the agent-wise motion step by step, and history trajectories feature in a two-stream structure. Specifically, we utilize the trajectory of the pedestrian, the ego vehicle and pedestrian motion to predict the multimodal trajectory of the pedestrian. One stream of the AANet studies the contextual information by the step-wise attention of the agent-wise motion to enhance the scenario understanding, while the other stream studies the temporal relationship of the trajectory. In addition, a query-based multistage decoder is designed and the prediction of the crossing intention of the pedestrian serves as an auxiliary task, which helps to understand the high-level motivation of the future motion of the pedestrian. Finally, the prediction results on the joint attention for autonomous driving (JAAD) and pedestrian intention estimation (PIE) datasets improve approximately 13% and 12%, respectively, demonstrating the effectiveness and our model achieves state-of-the-art performance.

Accurate prediction of the future trajectory of traffic agents is imperative to the effective motion planning of autonomous vehicles and mobile robots. Despite enormous progress that has been made toward trajectory prediction, dynamic and crowded traffic scenarios pose major challenges to the understanding and forecasting of traffic agents’ motion behavior. In this paper, we propose a novel trajectory prediction method from the perspective of temporal modeling and social interaction. Specifically, we first put forward a recurrent modeling approach to learn temporal features in favor of capturing long-range and short-range temporal dependencies of individual agents. Then, we construct a social feature learning module to capture the sparse and directional interactions among agents while suppressing the spurious connections. Finally, to reduce the accumulated error during prediction, a coordinated bidirectional decoding module is developed where temporal and social features can be properly integrated into the forward and backward prediction processes. Extensive experiments are performed on four real-world trajectory prediction benchmarks, and the results demonstrate the superiority of our method compared with other competing approaches. Detailed ablation studies are also performed to evaluate the effectiveness of each model component. Note to Practitioners—Motion planning is one of the crucial components of autonomous systems, such as intelligent vehicles and mobile robots. For example, the safety and efficiency of motion planning can be drastically improved if the future trajectories of surrounding agents, e.g., pedestrians, bicyclists, cars, etc., can be accurately forecasted. Motivated by the above requirements, this article develops an advanced deep learning model that can learn temporal and social features from trajectory data and perform accuracy prediction. This work aims to enhance the bidirectional recurrent network for dealing with trajectory data with evident temporal characteristics. In addition, this work introduces a novel adaptive social interaction modeling approach that overcomes the inherent defect of the fixed threshold method. This work also addresses the error accumulation problem in the prediction. The proposed model is evaluated on several datasets, and the results demonstrate its effectiveness. Our approach has broad application prospects in autonomous driving and mobile robots.

Trajectory prediction is crucial in understanding human behavior around intelligent agents, such as self-driving vehicles or social robots. Nevertheless, conventional approaches often fall short in effectively modeling the intricate trajectory-scene interaction. As they typically rely on simple feature concatenation methods which can introduce scene information irrelevant to motion trajectories, this leads to insufficient understanding of the scene context and a decrease in prediction performance. To tackle the issue, we propose an Adaptive Progressive Transformer-based Trajectory Predictor, APT-TP, which precisely forecasts future trajectories under the fine-grained trajectory-scene interaction constraint. The scene semantic maps and trajectory heatmaps are initially fused at a coarse-grained semantic level through a cross-attention mechanism. Afterward, inverse reinforcement learning is introduced to learn the trajectory-scene interaction from the fused results and generate possible future path plans through the Gumbel-Softmax sampling strategy. Finally, the generated path plans regarding the trajectory-scene interaction constraint are fused with the refined motion features at a fine-grained policy level through a novel APFormer. An adaptive motion token extractor is used to mitigate the redundancy in the refined motion features. APFormer fuses motion information and path plans progressively to generate future trajectories. APT-TP achieves promising performance on two benchmark datasets, Stanford Drone Dataset (SDD) and ETH/UCY, revealing its superiority. Moreover, qualitative evaluations demonstrate its effectiveness in exploring the trajectory-scene interaction, which is beneficial for ameliorated trajectory prediction performance.

This study presents a novel Multi-Agent Reinforcement Learning (MURL) architecture for autonomous vehicle (AV) navigation in complex urban traffic environments. By integrating a Social Value Orientation (SVO) model into a model-free SARSA reinforcement learning framework, our approach effectively balances individual agents’ social preferences with safety and performance objectives. A logistic regression-based risk assessment module evaluates collision probabilities in real time by analyzing spatiotemporal dynamics such as distances and velocities. Additionally, a dynamic optimizer adapts the learning rate and exploration strategies of the SARSA algorithm to provide efficient convergence to optimal policies. Extensive simulation experiments demonstrate that the proposed method significantly enhances safety and efficiency, achieving a 55.6% reduction in collision risk and increasing average rewards per episode by 2.1 compared to traditional SARSA without SVO. Furthermore, the optimized policy reduces average episode length, indicating the framework’s effectiveness in providing robust decision-making and adaptability across various traffic scenarios.

Pedestrian trajectory prediction is critical for understanding human behavior. The prevailing approaches employ neural networks to predict trajectories from large amounts of trajectory data. However, pedestrian trajectory data exhibits a long-tail distribution, which presents challenges in accurately predicting the future trajectories of tail samples. Previous research utilized contrastive learning and loss reweighting to tackle the long-tail distribution challenge in trajectory prediction. Although this approach enhanced the tail samples’ performance, it reduced the head samples’ performance. In order to address this limitation, we propose a trajectory prediction framework based on dynamic subclass-balancing contrastive learning in this work. Firstly, we obtain general motion patterns by clustering future trajectory data. We use the adaptive motion pattern refinement block to refine the general motion patterns, providing accurate guidance for the model and thus facilitating the recognition of tail motion patterns. Subsequently, we propose dynamic subclass-balancing contrastive learning to address the long-tail distribution issue of trajectory data on the encoder, which includes subclass-balancing clustering and dynamic dual-level contrastive learning. Subclass-balancing clustering is employed on the head trajectory data to achieve subclass balance across the dataset. Afterward, we perform dynamic dual-level contrastive learning for motion features to achieve instance balance and optimize the feature space. Finally, we use enhanced motion features to adjust the predicted trajectories through the trajectory proposal refinement block, achieving progressive refinement. This addresses the long-tail distribution issue of trajectory data on the decoder and improves the model’s generalization capability. Experimental results demonstrate that our method outperforms state-of-the-art long-tail trajectory prediction methods in addressing the long-tail distribution issue, improving the performance on both head and tail samples. The code will be released athttps://github.com/YanCCZU/DSBCL-PRM.Note to Practitioners—This work aims to tackle the long-tail distribution issue of pedestrian trajectory prediction while improving the model’s generalization capability. Existing methods mitigate the impact of the long-tail distribution issue on the encoder using contrastive learning. However, their overemphasis on the tail samples through loss reweighting has reduced the head samples’ performance. This work proposes a dynamic subclass-balancing contrastive learning module, which classifies head samples into several subclasses, each with a similar sample number in the tail classes. It performs dynamic dual-level contrastive learning based on class and subclass labels to achieve subclass and instance balance, improving the performance in head and tail samples. We utilize general motion patterns from the training set to guide the prediction of future trajectories. Moreover, we propose a progressive refinement strategy consisting of two refinement blocks to mitigate the impact of the long-tail distribution issue on the decoder and improve the model’s generalization performance. First, we adaptively refine motion patterns based on the difference between observed trajectories and historical motion patterns to provide accurate guidance. Then, we adjust the original predicted trajectories using the enhanced motion features, mitigating the impact of the long-tail distribution issue on the decoder while improving the model’s generalization and adaptability in unknown scenes. Our method’s simplified and effective model design ensures excellent real-time performance. Consequently, it is well-suited for deployment of edge devices in areas such as autonomous driving, intelligent surveillance, and social robotics. The proposed method enables accurate prediction of infrequent future trajectories in various scenarios, thus supporting safer decision-making.

Autonomous driving has become an inevitable trend in automotive development. Reinforcement Learning (RL) is extensively used in autonomous vehicle motion planning, demonstrating good generalization but facing challenges of long training times and lack of consideration of uncertainty. To address these challenges, an Adaptive Heuristic Reinforcement Learning (AHRL) approach is proposed. First, this study improves upon the Dueling Double Deep Q Network (D3QN) algorithm by proposing the Adaptive Heuristic Dueling Double Deep Q Network (Adapt-HD3QN) algorithm. Specifically, heuristic functions from search-based planning algorithms are incorporated into the RL reward terms to guide heuristic learning and enhance learning efficiency. Additionally, considering the uncertainties in real-world driving environments, such as the movement of other traffic participants and building occlusions, a Mixed Artificial Potential Field (Mix-APF) is implemented to address static and dynamic obstacle avoidance. Furthermore, potential collisions between the autonomous vehicle and other vehicles in occluded areas are modeled as a zero-sum game, with a Dynamic Bayesian Network (DBN) used for prior modeling of potential vehicles, aiding in constructing the potential vehicle’s forward hidden set. Finally, a signal-free intersection scenario, a typical crash-prone road type, is constructed on CARLA, incorporating static obstacles and other traffic participants. Experimental results demonstrate that the proposed Adapt-HD3QN algorithm exhibits superior safety, training efficiency, and traffic efficiency in scenarios with road environment perception uncertainties caused by occlusion.

Reinforcement learning (RL) offers a powerful paradigm for adaptive motion planning in complex driving environments. However, applying RL to autonomous driving remains challenging due to uncertainty from partial observability and the stochastic, multimodal behaviors of traffic participants. This paper presents a novel risk-aware RL framework for non-conservative motion planning under uncertainty. By integrating Partially Observable Markov Decision Processes (POMDP) with a deep RL-based policy optimization scheme, the proposed approach explicitly models aleatoric uncertainty via a Gaussian Mixture Bayesian Belief Updater and a time-varying risk field. Additionally, an Adaptive Context-aware Attention (ACA) module is employed to prioritize critical targets for enhanced interaction modeling dynamically. Extensive experiments on the CARLA simulator show that the framework generalizes well across diverse traffic conditions, improving average reward by 65.74% and 64.02% in low-speed dense and high-speed sparse scenarios. It remains robust in challenging situations such as overtaking and sudden lane changes in the PeMS dataset. Furthermore, distributed deployment tests confirm a real-time performance of 10 Hz on a hardware-in-the-loop platform, demonstrating the feasibility of practical deployment.

Accurately predicting the actions of vulnerable road users (VRUs) is crucial for improving traffic flow and enhancing VRU safety. The unpredictable nature of VRU trajectories poses a significant challenge. To address this, we introduce the Probabilistic Multimodal Trajectory Prediction Network (PMTPN), which effectively forecasts multimodal trajectories and their corresponding probabilities by utilizing a multitask learning framework that integrates trajectory and probability predictions. The network processes diverse input modalities, including bounding boxes, pedestrian pose, and ego-vehicle motion information. We enhance prediction performance by employing specialized encoders to extract distinct features from these inputs and a fusion module to integrate the data efficiently. To manage the variability in pedestrian actions, our model incorporates learnable motion queries that serve as reference points for predicting various potential outcomes. These queries are iteratively refined through attention operations with historical context in a multi-layer decoder. Additionally, a multi-gate mixture-of-experts (MMoE) module within the decoder helps mitigate the challenges of multitask learning. Our method significantly enhances trajectory prediction accuracy and provides probabilities for each predicted trajectory, demonstrating state-of-the-art results on the JAAD and PIE datasets.

With the development of intelligent connected vehicle technology, human-machine shared control has gained popularity in vehicle following due to its effectiveness in driver assistance. However, traditional vehicle following systems struggle to maintain stability when driver reaction time fluctuates, as these variations require different levels of system intervention. To address this issue, the proposed human-machine shared vehicle following assistance system (HM-VFAS) integrates driver outputs under various states with the assistance system. The system employs an intelligent driver model that accounts for reaction time delays, simulating time-varying driver outputs. Acontrol authority allocation strategy is designed to dynamically adjust the level of intervention based on real-time driver state assessment. To handle instability from driver authority switching, the proposed solution includes a two-layer adaptive finite time sliding mode controller (A-FTSMC). The first layer is an integral sliding mode adaptive controller that ensures robustness by compensating for uncertainties in the driver output. The second layer is a fast non-singular terminal sliding mode controller designed to accelerate convergence for rapid stabilization. Based on the driver-in-the-loop experimental results using the intelligent cockpit system, the performance of the HM-VFAS was evaluated. Results show that the proposed control strategy maintains a safe distance under time-varying driver states, with the actual acceleration error relative to the target acceleration maintained within$\pm 0.6\!\ \text {m/s}^{2}$and the maximum acceleration error reduced by$1.3\!\ \text {m/s}^{2}$. Compared to traditional controllers, the A-FTSMC controller offers faster convergence and less vibration, reducing the stabilization time by 26.8%.

The control problem for connected vehicular platoons requires balancing control optimality, saving computational and communication resources, and ensuring security. In this paper, a fuzzy prescribed performance adaptive optimal control strategy is developed for platoon system, and a distributed event-triggered (ET) mechanism is introduced. The main contributions include: 1) A prescribed performance adaptive dynamic programming (ADP) control architecture under distributed event-triggering is developed. The designed control method not only ensures the safety distance requirements of the platoon, but also significantly reduces the computing and communication costs, while guaranteeing the control optimality under the above objectives; 2) The stability proof of the platoon system considering the above complex control objectives is completed. Stable convergence of fuzzy logic system (FLS) is ensured by designing an experience replay-based critic update rule; 3) Considering the practicability in real working conditions, the cost function of the ADP controller robust to actuator saturation and disturbance is designed. Compared with existing methods, our approach achieves optimal control, robustness, and enhanced communication efficiency. Finally, the effectiveness and applicability of the controller are verified by simulations.

Lane detection is one of the most critical tasks in autonomous driving. In the past few years, due to the development of deep neural networks, lane detection approaches using onboard sensors like cameras and LiDAR have been proven to be effective ways to improve performance. In contrast to the camera-based scheme, LiDAR-based lane detection exhibits remarkable robustness to varying lighting conditions. This paper proposes a novel cyclic-shifting group convolution (CSGConv) module. Compared with classical group convolution, the proposed CSGConv module can efficiently promote information exchange among different group feature channels and reduce the high computational burden in current LiDAR-based lane detection networks. Then, a cross stage partial CSGConv (CSPCSG) block is designed to enlarge the receptive field and improve detection performance, especially when the lanes are curved or occluded. Subsequently, a LiDAR-based lane detection network, CSGNet, is put forward by integrating the CSPCSG block. Experiments and ablation tests on the dataset illustrate that our strategy achieves the highest 83.9% overall F1-score. Compared with the current state-of-the-art LiDAR-based lane detection method LLDN-RW, our results exhibit 2.5 times faster and 39% reduction of FLOPs, which indicates less resources are required in the online process.

Human-machine cooperation can simultaneously leverage the strengths of both human drivers and machines, making it a promising solution for improving driving safety, comfort, and experience. This paper designs a heterogeneous synchronous reinforcement learning (HSRL)-based human-machine shared steering decision-making (HMSSDM) strategy for intelligent vehicles. First, the vehicle dynamics, which incorporate steering characteristics, are built to quantify human driver’s steering behavior. Additionally, the scenario-oriented driving constraints (SODCs) are established to demonstrate driving constraints from traffic participants, roadside obstacles, and traffic signs. Second, to enhance the rationality and reliability of steering behaviors, the human driver’s steering behavior is evaluated using a fuzzy logic strategy, and HSRL is proposed to simultaneously determine steering actions and allocate driving authority between the human driver and machine. The main advantage of HSRL is its ability to perform both continuous domain learning (CDL) and discrete domain learning (DDL) simultaneously. Finally, the proposed method is validated using a human and hardware-in-the-loop (HHiL) experimental platform. The comparison results demonstrate that the proposed method outperforms the comparison methods in terms of driving safety, comfort and experience.

To enhance the trajectory tracking capability of steer-by-wire (SBW) vehicles while reducing driver’s workload, a human-machine shared control (HMSC) strategy using improved reinforcement learning-based model predictive control (RL-MPC) is proposed. In this paper, two main contributions have been made: 1) for driver loop, a variable steering ratio (VSR) strategy applied to SBW system is designed based on an improved fuzzy controller, whose parameters are optimized through simulated annealing (SA) algorithm; 2) for intelligent control system loop, an improved RL-MPC method is proposed to realize the high precision steering tracking control for autonomous vehicles (AVs), in which MPC and deep deterministic policy gradient (DDPG) are deeply integrated to combine their short-term optimization ability and long-term value estimation capability. Moreover, to shorten the time of overall training and ensure that the optimal control strategy can be explored, the DDPG agent is pretrained before the parallel training of RL-MPC. CarSim-MATLAB/Simulink co-simulation results show that in the whole tracking process, the lateral position error and yaw angle error of the vehicle are significantly reduced, indicating that the tracking accuracy is greatly improved. Meanwhile, the steering wheel angle and speed are also reduced, which means that the driver will spend less energy during the steering process.

The path tracking control problem for autonomous vehicle with uncertain dynamics requires simultaneous consideration of control optimality and safety-based performance constraints. In this paper, an adaptive optimal control method with prescribed performance is proposed to solve this problem, which contains two contributions: 1) by introducing a prescribed performance function (PPF) into adaptive dynamic programming (ADP), the controller can constrain the tracking error of the system within a specified performance boundary while optimizing the control cost; 2) the critic-only ADP is used for the controller design, which simplifies the commonly used actor-critic ADP scheme, and the convergence of the estimation error is guaranteed under FE conditions. On this basis, the neural network identification technique is introduced to deal with the unknown dynamic parameters of the vehicle system. The control scheme is able to strictly guarantee user-defined vehicle performance specifications with approximately optimal control performance. The stability of the closed-loop system is rigorously demonstrated by the Lyapunov method. In addition, the controller also embeds a radial basis function neural network (RBFNN) compensator to approximate the nonlinear external disturbances of the autonomous vehicle. Finally, the efficiency of the controller to achieve autonomous vehicle path tracking is verified by CarSim-Simulink simulation.

In this paper, the mixed traffic scenario with human-driven vehicles (HDVs) and connected and autonomous vehicles (CAVs) on freeway is considered. In this partly known nonlinear system, an optimal control algorithm using adaptive dynamic programming (ADP) is proposed to deal with the communication latence and drivers’ reaction time, which can stabilize the system under the influence of dead zone and saturation with minimal cost. There are three contributions in this paper. Firstly, in the used ADP algorithm, a critic neural network (NN) is designed to estimate the optimal value of the cost function, which is updated using online data instead of pre-gathered data. This means that the proposed controller can adapt to different parameters of different systems. Secondly, the reaction time of human driver and the time latence of the V2V communication are considered as the state and input delay of the nonlinear system, by adding the terms of delayed states to the optimal value function, the influence of the time delay can be minimized in the process of the critic NN updating. Thirdly, the saturation and dead zone of actuator are considered, by designing a new utility function of control value, the control value is limited from being out of the expected range. Under this condition, the stabilization of the overall system and the effectiveness of the proposed algorithm is proved and validated by means of simulation results.

Accurate anticipation of pedestrian crossing intentions is critical for preventing pedestrian-vehicle conflicts and ensuring road safety. This issue is a significant focus in intelligent transportation systems and autonomous driving. However, current approaches often face challenges, such as high computational costs due to complex scene understanding and inadequate consideration of spatiotemporal dependencies in pedestrian actions. To handle these challenges, we propose RAIDN (real-time action-interaction dual-branch network), which comprises the pedestrian action encoding and traffic-object interaction modules to anticipate pedestrian crossing intentions in real time. The pedestrian action encoding module employs a multi-scale graph transformer, efficiently extracting the intrinsic topology of long- and short-term action variations. This module effectively addresses issues of information redundancy in multi-channel graph convolution networks and local limitations in multi-scale temporal convolutions. Subsequently, the traffic-object interaction module introduces an interaction relation graph convolution network to excavate relevant traffic-object interactions, thereby shortening the prolonged scene semantic inference. Finally, global average pooling and attention layers fuse the action and interaction cues for real-time intention anticipation. The effectiveness of RAIDN has been validated on public datasets JAAD and PIE, achieving competitive metrics with Accuracy, ROC-AUC, F1-Score, Precision, and Recall rates of 0.89/0.92, 0.80/0.89, 0.66/0.85, 0.65/0.82, and 0.72/0.89 respectively. Notably, RAIDN demonstrates a remarkable inference time of just 0.28ms, outperforming other state-of-the-art methods and establishing its suitability for real-time applications in intelligent transportation and autonomous driving. Code is available athttps://github.com/wrysmile99/RAIG.

Autonomous vehicles (AVs) bring advantages such as comfort, safety, and eco-friendliness compared to conventional ones. The focus on comfort has become increasingly important as it plays a key role in the widespread acceptance of AVs. Passenger demand for alleviating motion sickness (MS) during travel has also driven extensive research in this area. This manuscript conducts a comprehensive and comparative review of published articles to provide up-to-date research outcomes on MS mitigation. It also examines the optimization methods employed over the past two decades to enhance motion comfort in AVs. The methods for detecting and resolving MS are elaborated upon, and the research gap and challenges are addressed. Furthermore, the manuscript presents the current efficacy of the research and proposes an anti-motion sickness control framework that utilizes a cloud control platform, which might offer potential future research directions in the field of autonomous driving. This study serves as a valuable reference for future efforts and opportunities to improve the motion comfort of AVs.

Passenger comfort is a crucial aspect that influences humans’ acceptance of automated vehicles. The passenger comfort score (PCS) is closely related to the passengers’ psychological states, however, comfort quantification methods based on the passengers’ psychophysics mechanism are rare. This research pioneers a passenger comfort quantification model (PCQM) specifically designed for automated vehicles, demonstrating the model’s ability to accurately quantify subjective PCS under urban LCS. Three significant contributions form the basis of this study: 1) A dataset dedicated to comfort quantification is collected. A novel PCQM based on ensemble learning of psychophysics mechanism based sub-model and encoder-transformer based sub-model is proposed. The psychophysics mechanism model is derived from Stevens’ power law. 2) As a subjective indicator, the self-reported score (SRS), which is the indicator of PCS contains considerable noise. The PCQM addresses the issue of substantial noise prevalent in the subjective SRS by incorporating a semi-supervised learning strategy, which enhances data consistency and suppresses noise. 3) The efficacy of the proposed PCQM is corroborated via deployment on an automated vehicle, where the model’s real-time predictions strongly align with SRS from onboard passengers.

Vehicle control is one of the most critical challenges in autonomous vehicles (AVs) and connected and automated vehicles (CAVs), and it is paramount in vehicle safety, passenger comfort, transportation efficiency, and energy saving. This survey attempts to provide a comprehensive and thorough overview of the current state of vehicle control technology, focusing on the evolution from vehicle state estimation and trajectory tracking control in AVs at the microscopic level to collaborative control in CAVs at the macroscopic level. First, this review starts with vehicle key state estimation, specifically vehicle sideslip angle, which is the most pivotal state for vehicle trajectory control, to discuss representative approaches. Then, we present symbolic vehicle trajectory tracking control approaches for AVs. On top of that, we further review the collaborative control frameworks for CAVs and corresponding applications. Finally, this survey concludes with a discussion of future research directions and the challenges. This survey aims to provide a contextualized and in-depth look at the state of the art in vehicle control for AVs and CAVs, identifying critical areas of focus and pointing out the potential areas for further exploration.

In this paper, Vulnerable Road User (VRU) trajectory prediction for autonomous driving based on the Intention-Attention-Gate Recurrent Unit (IA-GRU), Improved Social Force Model (ISFM) and Adaptive Boosting (AdaBoost) is systematically investigated. Firstly, a novel IA-GRU is proposed for VRU (pedestrian, cyclist, and electric cyclist) trajectory prediction. VRU intention (waiting/crossing), VRU heterogeneity (age and gender), VRU-VRU interactions and VRU-dynamic vehicle interactions are taken into account. Attention is used to obtain the influence weights of the above factors used for VRU trajectory prediction. Secondly, a micro-dynamic ISFM is developed for VRU trajectory prediction. The impact of zebra crossing, collision avoidance with vehicles and VRUs, and VRU heterogeneity are considered. Moreover, traffic data collected by an unmanned aerial vehicle (UAV) is obtained and analyzed, and the parameters of the ISFM are calibrated by the Maximum Likelihood Estimation (MLE). Finally, a data-driven integrated approach based on the IA-GRU and ISFM is proposed, and AdaBoost is used to prevent the model from overfitting and improve the prediction accuracy. The results indicate that the integrated model outperforms the existing methods, and the prediction accuracy is improved by more than 11% based on the collected traffic data, which can give us great confidence to use the integrated model in the autonomous driving domain to improve the safety of VRUs.

In this paper, a composite longitudinal controller for velocity regulation of unmanned ground vehicles (UGVs) is proposed. First, both the empirical model and mixed logical dynamic model are developed in terms of the experimental data processing method. Second, a composite longitudinal control strategy incorporating steady-state controller and active disturbance rejection state feedback controller is designed for velocity regulation control during acceleration and deceleration. Finally, the stability analysis of velocity error is given with the global sector condition, and the comparative experimental tests show the effectiveness and robustness of the proposed composite longitudinal regulation control strategy.

This article presents an interacting multiple model (IMM) for short-term prediction and long-term trajectory prediction of an intelligent vehicle. This model is based on vehicle's physics model and maneuver recognition model. The long-term trajectory prediction is challenging due to the dynamical nature of the system and large uncertainties. The vehicle physics model is composed of kinematics and dynamics models, which could guarantee the accuracy of short-term prediction. The maneuver recognition model is realized by means of hidden Markov model, which could guarantee the accuracy of long-term prediction, and an IMM is adopted to guarantee the accuracy of both short-term prediction and long-term prediction. The experiment results of a real vehicle are presented to show the effectiveness of the prediction method.

Trust is a crucial aspect for autonomous vehicles (AVs) and automated driving systems (ADSs) that maintains humans’ acceptance of such functions. However, trust dynamic models describing the human-vehicle relationship rarely exist. To this end, this paper develops a quantitative trust dynamic model of the driver to an adaptive cruise control (ACC) system and applies the proposed model to a trust-based ACC. Driver’s trust level is modeled as an objective evaluation index for assessing the individual perceived trustworthiness on automation. Three contributions have been made in this work: 1) a novel quantitative dynamic model describing the driver’s trust on ACC is proposed for the first time considering the effects of the driver’s taking-over and handing-over operations as well as the steady-state driving scenarios; 2) an improved control barrier function approach with guaranteed system stability and prescribed asymptotic performance is proposed to design the trusted-based ACC, satisfying the state, input, and safety constraints; and 3) a new prescribed performance function that does not require accurate value of the initial condition is proposed to restrict the tracking errors within the predefined asymptotic boundaries, given that the initial value of the trust is hard to obtain. High-fidelity CarSim-based simulations demonstrate the rationality of the proposed trust model and the validity of the proposed control approach.

This paper concerns the robust gain-scheduling control issue for autonomous path following systems with stochastic network-induced delay. Firstly, to effectively approximate the highly nonlinear tire dynamics, the linear fractional transformation formulations are employed to describe the tire cornering stiffness with a norm-bounded uncertainty. Secondly, by taking the data dropout and delay encountered in signal computation and transmission into account, a more generalized lumped delay form is proposed to unify the time-varying data dropout and network-induced delay. Moreover, a Markovian process is presented to describe the lumped delay as a stochastic distribution. Thirdly, to address the issue of varying vehicle velocity, a linear parameter varying model is established to capture vehicle lateral behaviors. Based on the stochastic stability theory, a new robust gain-scheduling path following control method is proposed for the autonomous vehicles. Finally, the experimental study is presented to bridge the gap between the theoretical and practical investigations on path following control of autonomous vehicles, and results validate the superior performance of the proposed method compared with existing works.

This study addresses the path-tracking control issue of autonomous vehicles (AVs) when the GPS measurement is temporarily unavailable. In such a case, the vehicle states, location or the curvature of the reference path might be unobtainable, while the camera can be potentially used to detect the path-tracking states. To this end, this paper proposes a fuzzy-observer-based composite nonlinear feedback (CNF) controller with a Takagi-Sugeno (T-S) vehicle lateral dynamic model to guarantee the normal path-tracking maneuver and improve the transient performance. A parallel distributed compensation (PDC)-based CNF control method is developed to realize the control objective with the T-S vehicle model considering the transient performance and actuator saturation. The closed-loop stability and H∞index performance integrating the tracking and estimation errors have been proved with a Lyapunov approach. The observer-based controller design has been implemented based on the formulation of the linear matrix inequalities (LMIs). High-fidelity simulations using CarSim-Matlab/Simulink have demonstrated the validity of the proposed approach in terms of enhancing the tracking performance under the input saturation and disturbances in GPS-denied environments.

This paper proposes a novel robust path-following strategy for autonomous road vehicles based on type-2 fuzzy PID neural network (PIDT2FNN) method coupled to an Extended Kalman Filter-based Fuzzy Neural Network (EKFNN) observer. Uncertain Gaussian membership functions (MFs) are employed to self-adjust the universe of discourse for MFs using the adaptation mechanism derived from Lyapunov stability theory and Barbalat's lemma. External disturbances are significant in autonomous vehicles by changing the driving condition. Furthermore, parametric uncertainties related to the physical limits of tires and the change of the vehicle mass may significantly affect the desired performance of autonomous vehicles. The robustness of the proposed controller against the parametric uncertainties and external disturbances is compared with one active disturbance rejection control (ADRC) algorithm, and a linear-quadratic tracking (LQT) method. The obtained results in terms of the maximum error and root mean square error (RMSE), demonstrate the effectiveness of the proposed control algorithm to reach the minimized path-tracking error.

This article investigates the integrated lane-keeping and roll control for autonomous ground vehicles (AGVs) considering the transient performance and system disturbances. The robust integral of the sign of error (RISE) control strategy is proposed to achieve the lane-keeping control purpose with rollover prevention, by guaranteeing the asymptotic stability of the closed-loop system, attenuating systematic disturbances, and maintaining the controlled states within the prescribed performance boundaries. Three contributions have been made in this article: 1) a new prescribed performance function (PPF) that does not require accurate initial errors is proposed to guarantee the tracking errors restricted within the predefined asymptotic boundaries; 2) a modified neural network (NN) estimator which requires fewer adaptively updated parameters is proposed to approximate the unknown vertical dynamics; and 3) the improved RISE control based on PPF is proposed to achieve the integrated control objective, which analytically guarantees both the controller continuity and closed-loop system asymptotic stability by integrating the signum error function. The overall system stability is proved with the Lyapunov function. The controller effectiveness and robustness are finally verified by comparative simulations using two representative driving maneuvers, based on the high-fidelity CarSim-Simulink simulation.

This paper investigates the lane keeping control of autonomous ground vehicles (AGVs) considering the rollover prevention and input saturation. An enhanced state observer-based sliding mode control (SMC) strategy is proposed to achieve the control purpose and maintain the lane keeping errors as well as the roll angle within the prescribed performance boundaries. Three contributions are made in this paper. First, a prescribed performance function (PPF) is proposed in the controller design, aiming to implement the error transformation so as to constrain the controlled variables within the prescribed performance boundaries. Second, a modified sliding surface is developed incorporating two nonlinear functions, whose specialities and benefits are taken advantage of: one is a barrier function to restrict the load transfer ratio (LTR) in a safe boundary to guarantee the roll stability; another is a monotonely decreasing function to adaptively change the damping ratio of the closed-loop system to improve the transient performance, including reducing the transient overshoots and steady-state errors. Third, a modified multivariable adaptive SMC controller is proposed to achieve the integrated lane-keeping and roll control in the presence of the input saturation and bound-unknown disturbances. The stability of the closed-loop system is rigorously proved via the Lyapunov function. Finally, the effectiveness of the proposed control strategy is verified with a high-fidelity and full-car model via the CarSim platform.

This paper investigates the lane keeping control for four-wheel independently actuated autonomous vehicles. To guarantee the vehicle safety when the active-steering motor entirely fails, the steering manoeuvre is accomplished by the differential drive assisted steering (DDAS), which is generated by the differential moment between the front wheels. A novel adaptive multivariable super-twisting control strategy is proposed to realize the control objective in finite time, considering the multiple unknown and mismatched disturbances of the steering system with the chattering effect removed. In the sliding surface, a nonlinear function is designed to adaptively change the damping ratio of the closed-loop system so as to improve the transient performance of the lane keeping control in the faulty condition. The finite-time convergence of the closed-loop system is proved by Lyapunov function technique. Results of CarSim-Simulink simulations with a high-fidelity and full-car model have verified the effectiveness and robustness of the proposed controller in the lane keeping control with DDAS and guaranteeing high performance.

Differential drive assistance steering (DDAS) is an emerging assisted steering mechanism in in-wheel-motor driven (IWMD) electric vehicles, yielded by the differential moment of the front tires in the steering system. DDAS can steer the front wheels when there is no steering power from the steering motor, and thus can be used as a redundant steering mechanism. To realize the yaw control when the active front steering entirely breaks down and guarantee the transient control performance therein, this paper proposes an integral sliding mode control (ISMC) approach for IWMD electric vehicles steered by DDAS. Two contributions are made in this paper: 1) An improved disturbance observer based ISMC strategy is designed to cope with the unknown mismatched disturbances, and the composite nonlinear feedback technique is employed to design the nominal part of the controller to restrain overshoots and remove steady-state errors considering the tire force saturations; 2) An adaptive super-twisting control approach is proposed to deal with the disturbances with unknown boundaries using a continuous controller while eliminating the chattering effect. The system stability and robustness are proved via Lyapunov approach. CarSim-Simulink simulation has verified the effectiveness of the proposed control approach in the case of the steering fault.

This paper investigates the path-following control problem for four-wheel independently actuated autonomous ground vehicles through integrated control of active front-wheel steering and direct yaw-moment control. A modified composite nonlinear feedback strategy is proposed to improve the transient performance and eliminate the steady-state errors in path-following control considering the tire force saturations, in the presence of the time-varying road curvature for the desired path. Path following is achieved through vehicle lateral and yaw control, i.e., the lateral velocity and yaw rate are simultaneously controlled to track their respective desired values, where the desired yaw rate is generated according to the path-following demand. CarSim-Simulink joint simulation results indicate that the proposed controller can effectively improve the transient response performance, inhibit the overshoots, and eliminate the steady-state errors in path following within the tire force saturation limits.

This paper presents a robust H∞path following control strategy for autonomous ground vehicles with delays and data dropouts. The state measurements and signal transmissions usually suffer from inevitable delays and data packet dropouts, which may degrade the control performance or even deteriorate the system stability. A robust H∞state-feedback controller is proposed to achieve the path following and vehicle lateral control simultaneously. A generalized delay representation is formulated to include the delays and data dropouts in the measurement and transmission. The uncertainties of the tire cornering stiffnesses and the external disturbances are also considered to enhance the robustness of the proposed controller. Two simulation cases are presented with a high-fidelity and full-car model based on the CarSim-Simulink joint platform, and the results verify the effectiveness and robustness of the proposed control approach.

The path-following problem for autonomous vehicles is investigated in this paper. The desired vehicle heading is commonly chosen as the tangent direction on the desired path. This paper points out that the traditional definition of the desired heading may deteriorate the path-following performance, particularly when the vehicle is tracking a path with large curvature. That is because the sideslip angle control and the yaw rate control are conflicting in the presence of sliding effects, and the sideslip angle does not equal to zero when the vehicle is tracking a curve path. This paper further provides an amendment to the definition of the desired heading, which realizes a more accurate path-following maneuver. In the controller design phase, backstepping is used to generate the required yaw rate, and an LQR controller is adopted to obtain the optimal active front steering input. The CarSim-Simulink joint simulation verifies the reasonability of the amendment to the desired heading.