Vehicle-to-Everything (V2X) collaborative perception bolsters the performance of autonomous vehicles by overcoming occlusion challenges and expanding their sensing range. However, current methodologies frequently overlook the intrinsic uncertainties associated with localization inaccuracies, asynchronous measurements, and diverse agent models. Such uncertainties may weaken fusion reliability, leading to performance worse than single-vehicle perception. To address this pivotal challenge, we introduce BELT-Fusion, a cohesive probabilistic framework tailored for reliable V2X late fusion. Our framework offers two notable advantages. The first is explicit agent-level uncertainty modeling, where classification uncertainty is captured via evidential deep learning and regression uncertainty via Bayesian neural networks. This capability allows task-specific reliability assessments to be effortlessly incorporated into existing object detectors. Second, our framework introduces an uncertainty-aware adaptive fusion representation. This representation dynamically guides object selection and weight allocation based on measurable fusion-level uncertainty, ensuring reliable fusion results without retraining and enabling plug-and-play functionality. To validate BELT-Fusion’s efficacy, we conducted evaluations focusing on 3D object detection in both real-world and simulated scenarios using the OPV2V and DAIR-V2X datasets. BELT-Fusion improved [email protected] by 7.16% in noisy settings and 3.84% under ideal conditions over uncertainty-agnostic baselines, demonstrating its robustness under challenging and noisy conditions. Our code will be available athttps://github.com/ZhiguoZhao/BELT-Fusion

This paper presents a novel, high-performance multi-vehicle detection and tracking (MVDT) framework to extract vehicle trajectories from roadside 3D point clouds. First, we developed a vehicle detector called the deep evidential occupancy grid model. This model uses PointNet to extract features from vertically organized point cloud pillars, generating a bird’s-eye view (BEV) grid feature map, followed by a 2D convolutional neural network to capture both local and global spatial features. The detection head uses a Dirichlet distribution to gather “evidence” from gathered features that indicate grid occupancy or vacancy, quantifying the model’s “confidence” through a probabilistic representation of uncertainty. An adaptive post-processing method is applied to output detection results with associated uncertainty. Second, we integrated uncertainty into the tracker, developing an uncertainty-aware multi-vehicle tracking (UMVT) model. This model manages the trajectory initialization and termination strategies, enhancing tracking robustness in complex scenarios. Comprehensive experiments on two real-world datasets, DAIR-V2X and V2X-Real, demonstrate that the proposed MVDT framework outperforms state-of-the-art methods in roadside perception, achieving improvements in both detection and tracking tasks. The effectiveness has also been validated using point cloud data from the Shanghai-Nanjing Expressway, showcasing its excellent practical applicability.

Black-box decision-making algorithms based on deep learning and reinforcement learning have demonstrated significant advancements in autonomous vehicles (AVs), but still encounter challenges due to unpredictable failures in long-tail scenarios. The efficient evaluation of the safety of these algorithms is essential for the further development of AVs. This paper introduces a novel testing and evaluation method that leverages subset simulation (SuS) with adaptive Markov chain Monte Carlo (ApMCMC). The method is designed to facilitate the occurrence of rare events and expedite the probability assessment process, thereby enabling rapid evaluation of the failure probability of deep reinforcement learning (DRL)–based end-to-end decision-making algorithms. The car-following scenario is used as a case study to demonstrate the efficacy of our proposed method and four state-of-the-art DRL-based algorithms are trained as evaluation objects. Experimental results indicate that our method maintains evaluation errors within 5%. Compared with the naive Monte Carlo method and other existing evaluation techniques, our approach significantly enhances evaluation accuracy, reduces variance, and improves confidence levels, while accelerating evaluation efficiency by 5-200 times. Additionally, our method can easily handle end-to-end black-box decision-making systems and exhibits the adaptability to be seamlessly extended and applied to other scenarios.

Full-sample vehicle trajectories collected by roadside sensors are pivotal for advancements in active traffic management and vehicle-highway automation applications. However, sensor inefficiencies and environmental factors often lead to incomplete and noisy data, impeding practical applications. To overcome this hurdle, we propose a data-driven and kinematics-aware method for reconstructing full-sample vehicle trajectories. It employs an encoder-decoder architecture, where the data-driven encoder utilizes graph-based map modeling to capture road network characteristics and models multi-vehicle interactions based on real-world driving attention patterns. The model-driven decoder integrates conditional variational autoencoder with a vehicle kinematic model, to ensure that the generated trajectories adhere closely to vehicle kinematic principles. Experiments on large-scale trajectory datasets demonstrate that our approach exhibits good performance in trajectory reconstruction. Furthermore, case studies on the DAIR-V2X dataset indicate that our approach could improve raw trajectory data quality and downstream prediction accuracy.

Cooperative perception, using vehicle-to-everything (V2X) technologies for perceptual data sharing between autonomous vehicles (AVs) and intelligent infrastructure, is considered a solution to many single-agent perception challenges. Early fusion, a data fusion scheme for the cooperative perception of AVs, provides a universally available data-sharing approach but has been criticized for its huge bandwidth consumption. This paper proposes a safety field (SF)-based vehicle-infrastructure cooperative perception approach by quantifying the driving risk in complex traffic scenarios. Leveraging the SF theory and point cloud downsampling, we design a delay-aware early fusion framework with adaptive communication volume control. We propose a latency-compensation error (LCE) for performance evaluation considering data transmission delay. The proposed framework is tested and verified in simulated city environments and simulated and real-world datasets. The experimental results show that the proposed approach increases the average precision (AP) and reduces the LCE compared with base models within a limited communication budget.

Parking scenarios present huge challenges for the prediction of intentions and situational awareness of autonomous vehicles. Unlike urban roads, parking scenarios feature greater freedom of movement, denser interactions, and more diverse intents. To address this issue, we propose the Parking Memory Interaction Transformer (PMI-Transformer) for predicting vehicle intent and trajectory based on cooperative vehicle-infrastructure systems (CVIS). Our model utilizes long-term semantic observation data and short-term historical trajectories to predict a target vehicle’s future intentions and trajectories from roadside perception units in CVIS. Based on global spatiotemporal features, a memory encoding strategy extracts essential information from the vehicle’s long-term history. At the same time, a hypergraph-based multiscale interaction module captures dense, diverse interactions in multi-vehicle environments. We incorporate parking occupancy data and calculate affinities between the vehicle and nearby spots to better characterize vehicle intent within the parking context. Experimental results show that our model outperforms state-of-the-art methods in intent and trajectory prediction, with visualizations confirming that PMI-Transformer effectively captures key historical information and social interactions in parking scenarios.

This article introduces a sustainable human-guided reinforcement learning (RL) framework to address the challenge of learning performance degradation when the human guidance is suspended. First, a compensation reward based on the historical similarity between the RL agent and human guidance history is designed to ensure the continued influence of human guidance. To avoid cumulative errors in value function approximation caused by fitting the new reward, including the compensation reward, a novel RL paradigm is proposed, which bypasses value function fitting and directly optimizes the policy using historical similarity. This paradigm develops a new historical similarity-based learning objective for RL to leverage human guidance more efficiently and achieve alignment with human behavior. Furthermore, the proposed paradigm enables the fine-tuning of the RL agent to address the long-tail problem. Experimental results demonstrate the advantages of the proposed method in terms of sustainable guidance and optimal performance in the autonomous driving, achieving a 15% increase in optimal performance compared with existing state-of-the-art (SOTA) methods.

Considering the inefficient resource allocation (RA) and high quality of service (QoS) requirement in vehicular communications, this paper proposes two dynamic RA algorithms, spectral clustering based greedy (SCGR) algorithm and multi-agent deep reinforcement learning (DRL) algorithm, to maximize both the sum capacity of the vehicle-to-infrastructure (V2I) uplinks and the total energy efficiency (EE) of the vehicle-to-vehicle (V2V) links by assigning the proper power and resource block (RB) to each V2V link in spatial modulation (SM) multiple-input multiple-output (SM-MIMO) vehicular system. For the SCGR algorithm, the spectral clustering (SC) scheme is first utilized to group the V2V links for the suitable RBs. Then, the optimal power is distributed to each V2V link by the greedy (GR) algorithm. For the DRL algorithm, a decentralized model-free network, improved multi-agent deep Q fully connected neural network (IDQFN), is developed to simultaneously find the best power allocation (PA) and RB assignment (RBA). Moreover, the SM technology is exploited to convey the information through the V2I and V2V links and improve the system capacity. Numerical results reveal that the proposed SCGR and IDQFN RA schemes outperform the typical RA algorithms, and the IDQFN scheme achieves better EE than the SCGR scheme, while the SCGR algorithm obtains the optimal average bit error rate (ABER) performance.

Sensors, such as cameras, millimeter-wave radar, and LiDAR, are widely deployed in cooperative vehicle-infrastructure systems. The demand for calibration of initial installation, damage replacements, and unstable installation has risen dramatically. Traditional methods require on-site operation and road closure; thus, repeated calibration can severely affect traffic conditions and expose operational personnel to potential safety threats. As more and more autonomous vehicles (AVs) flood the roads, this paper proposes an automatic calibration framework of roadside sensors by leveraging the high-precision positioning and perception data of AVs. First, we design a graph-based target-matching algorithm using an AV’s surrounding traffic perception data to identify the AV of interest from a dataset of multiple target trajectories recorded by roadside sensors. A line search algorithm is then designed to adjust the clock delay between sensors and establish the temporal correspondence, where a Gaussian process is applied to estimate the vehicle state in continuous time. Finally, we develop a least squares optimization model to complete the final calibration with the AV positioning data. The influence of measurement noise and missed detections on the proposed calibration framework are analyzed in simulated scenarios based on a Next Generation SIMulation (NGSIM) dataset, and the practicability is validated based on real-world data collected at Donghai Bridge, Hangzhou Bay Bridge, and DAIR-V2X dataset. It is shown that the proposed target matching algorithm can identify an AV trajectory from roadside sensor data with 20%-90% higher accuracy than baseline models, and the framework can accurately estimate the spatial and temporal parameters even with poor data quality. The mean least squares error of the trajectory alignment reaches centimeter-level accuracy.

Cameras, millimeter-wave radars, and lidars are widely deployed on smart roads to obtain personalized vehicle trajectories for advanced traffic control and risk avoidance. However, these asynchronous roadside sensors need to be spatiotemporally calibrated accurately before they are put into service. Traditional manual manipulation methods are inefficient and will affect traffic operation and safety. A rapid and convenient method has become essential under the trend that large amounts of roadside sensors need to be tested and calibrated frequently. As more and more connected and automated vehicles (CAVs) flood the smart roads, this paper proposes a novel spatiotemporal calibration framework using the positioning and perception data of CAVs. First, a trajectory matching algorithm is designed using motion feature and point feature histogram sequences as the descriptors, which can determine the approximate spatiotemporal correspondence for the CAV from the roadside trajectory dataset. An optimization method is then formulated to tune transformation parameters through the Gaussian Process trajectory representation and Gauss-Newton algorithms, considering the sampling frequency deviation and measurement noise. Based on numerical analysis via the NGSIM and HighD datasets, it is shown that the proposed calibration method can significantly reduce transformation errors and perform robustly in different scenarios. The feasibility and practicability of the calibration method are further validated through real-world experiments at Tongji University and on the Donghai Bridge in Shanghai, China. This study provides an economical and practical way for spatiotemporal calibration of roadside sensors in an era of CAVs.

Potential fields have been integrated with local path-planning algorithms for autonomous vehicles (AVs) to tackle challenging scenarios with dense and dynamic obstacles. Most existing potential fields are isotropic without considering the traffic agent’s geometric shape and could cause failures due to local minima. We propose a three-dimensional potential field (TriPField) model to overcome this drawback by integrating an ellipsoid potential field with a Gaussian velocity field (GVF). Specifically, we model the surrounding vehicles as ellipsoids in corresponding ellipsoidal coordinates, where the formulated Laplace equation is solved with boundary conditions. Meanwhile, we develop a nonparametric GVF to capture the multi-vehicle interactions and then plan the AV’s velocity profiles, reducing the path search space and improving computing efficiency. Finally, a local path-planning framework with our TriPField is developed by integrating model predictive control to consider the constraints of vehicle kinematics. Our proposed approach is verified in three typical scenarios, i.e., active lane change, on-ramp merging, and car following. Experimental results show that our TriPField-based planner obtains a shorter, smoother local path with a slight jerk during control, especially in the scenarios with dense traffic flow, compared with traditional potential field-based planners. Our proposed TriPField-based planner can perform emergent obstacle avoidance for AVs with a high success rate even when the surrounding vehicles behave abnormally.

Roadside sensors, such as camera and millimeter-wave (MMW) radar, provide traffic information beyond the visual range of intelligent vehicles in cooperative vehicle-infrastructure systems. Unlike onboard equipment, roadside sensors are affiliated with different systems and lack synchronization in both space and time. In this paper, we propose a novel spatio-temporal synchronization method of asynchronous roadside MMW radar–camera for sensor fusion, which utilizes features of the scenario to extract lane line corner points to pre-calibrate the camera. Based on the consistent time flow rate of the separate sensors, multiple virtual detection lines are set up to match the time headway of successive vehicles and conduct objective matching to track data. Finally, a synchronization optimization model is formulated and a constrained nonlinear minimization solver is applied to tune the parameters. Measure data from Donghai Bridge in Shanghai is applied to verify the feasibility and effectiveness of the method. The results determine that there are 33 frames (33*40 ms) of temporal deviation between the camera and the radar in this case. After the synchronization, the average spatial deviation is reduced from 2.47 m to 0.42 m in the X-direction and 64.06 m to 2.34 m in the Y-direction, respectively. This study provides an economical and effective way to solve the problem of spatio-temporal synchronization of roadside sensors.

This paper develops a new approach for tram prioritization integrating an offline traffic signal timing planner with an online tram progression controller. The offline planner optimizes tram progressions by resynchronizing traffic signals to minimize tram running times, taking into account the effect of the resynchronization on other vehicles. The online controller aims at enhancing tram reliability by adopting three control strategies-green extension, vehicle holding, and speed guidance-to instruct the trams to travel within appropriate progressions. The real-world case studies are presented to demonstrate that, comparing with the state-of-the-practice approach, the proposed approach has the potential to improve the service quality by shortening tram running time and passenger waiting time, and to mitigate the negative impact on other vehicles by avoiding triggering unnecessary green extensions.